All reviews
Home / Reviews / Air AI
AI voice-agent software · sold as a licensed “agency-in-a-box” business opportunity

Air AI Technologies, Inc.

A five-figure license to resell an AI phone agent that a federal complaint says could not reliably place an outbound call - sold with no disclosure document, no earnings claim statement, and a buy-back the sellers admitted they had no cash to honor.

Reviewed July 29, 2026 Founded Business start date recorded as 9 February 2023 on the BBB file; the FTC complaint pleads the conduct ran from at least February 2023 until the suit was filed in August 2025 Confidence: Medium-High
FGRADE
2.3/10
Weighted composite

DEFUNCT - OWNERS PERMANENTLY BANNED FROM BUSINESS OPPORTUNITIES

Entry ran from $9,800 to $100,000, often financed at up to 24.99% APR by the seller itself; the FTC alleged roughly $19 million of consumer losses, and $50,000 is what is actually payable for redress.

The question you came with

Can you actually make money with Air AI?

NO No - not on the numbers this company publishes

No. Not as a business, and no longer as a fact either: the three owners are permanently barred by a stipulated federal order announced 24 March 2026 from selling or marketing any business opportunity, and the offer itself is gone. It is written up here because licenses shaped exactly like this one are still being sold by other people, and the shape is worth recognizing before the deposit rather than after.

Entry ran from $9,800 to $100,000, frequently financed at up to 24.99% APR through the sellers own lending platform, so an optimistic decision turned into a hard debt that outlived the product. The FTC complaint pleads that buyers were given no disclosure document and no earnings claims statement at any point. That means there was never a denominator - no cohort, no percentage, nothing to hold a sales call against. What buyers received instead was a list of numerators.

The rest is arithmetic. Roughly $19 million of alleged consumer losses. Fifty thousand dollars actually payable for redress, on a judgment largely suspended for inability to pay, which is about a quarter of a cent on the dollar. The buy-back promise carried a 120-day floor and no ceiling, so indefinite delay was contractually survivable. And the input cost ran $0.11 to $0.32 a minute against $0.05 to $0.09 on the open market with no entry fee at all.

What is true in its favor is not nothing. There was no downline, no matrix and no commission for enrolling another licensee. No token, no staking, no pooled fund and no promised return on capital. Buyers were told to go out and work an agency, which is a different failure from paying for a position in a chain. And the whole file is now public: the complaint, the docket and the settlement can all be read.

What it costs to be in
$9,800–$100,000

for the licensing business opportunity, or $15,000–$30,000 for the “Access Card”; frequently financed at up to 24.99% APR through the sellers’ own lending platform

What would have to change
  • A disclosure document handed over seven days before payment, and an earnings claims statement carrying the number and percentage of all purchasers who achieved the advertised result. The Business Opportunity Rule requires both, and the complaint pleads neither was ever provided.
  • A fixed published price. The complaint alleges the price was adjusted to each consumer's ability to pay. A number that moves with the buyer's bank balance is not a price, and no purchaser could compare what they were quoted against what anybody else paid.
  • A buy-back clause with a ceiling on it. The agreement promised a full buy-back for any reason but no sooner than 120 days after signing, and set no maximum time to pay, which is what made indefinite delay survivable.
  • Software that performs the function it is licensed for. The complaint records an agent that was often unavailable and could not reliably place outbound calls, schedule or transcribe accurately, while the financing tool worked.

That call is computed, not chosen - the rule reads three of the nine published dimension scores and is printed on the methodology page. It describes this company's plan and the figures it publishes about the people already in it. It is not a prediction about you, and nothing on this site is advice.

~$19M
Alleged consumer losses
pleaded in the FTC complaint filed 25 August 2025
$50,000
Actually payable for consumer redress
on an $18 million judgment largely suspended for inability to pay
None
Disclosure documents or earnings claim statements furnished
the complaint pleads that neither was ever provided
21 of 22
BBB complaints left unanswered in three years
F rating, not accredited, two alerts on file

Legal status

UNDER ENFORCEMENT - and, on the participant side, closed. On 25 August 2025 the Federal Trade Commission filed a complaint against Air AI Technologies, Inc., five affiliated Arizona entities and three individual owners in the US District Court for the District of Arizona, case 2:25-cv-03068-SMB before Judge Susan M. Brnovich, alongside a motion for a temporary restraining order. A complaint is an allegation and nothing in it has been proven at trial. Four counts were pleaded: false and unsubstantiated earnings claims and misrepresented refund guarantees under Section 5(a) of the FTC Act, the Telemarketing Sales Rule (16 C.F.R. Part 310) as to the telephone-sold Access Card, and the Business Opportunity Rule (16 C.F.R. Part 437) as to the licensing offer. A stipulated preliminary injunction was entered on 10 September 2025 and is confirmed in force. The defendants answered on 31 December 2025 with a jury demand; the parties jointly moved to stay for settlement in early February 2026 and the stay was granted. On 24 March 2026 the FTC announced a stipulated order for permanent injunction, monetary judgment and other relief, filed on an unopposed motion by a 2-0 Commission vote: the corporate defendants and owners Caleb Maddix, Ryan O’Donnell and Thomas Lancer are permanently banned from selling or marketing any business opportunity, under an $18 million monetary judgment largely suspended on an inability-to-pay basis with $50,000 payable for consumer redress. The defendants agreed to that order. As with settlements of this kind, it resolves without an admission of liability, and there has been no judicial finding of fraud after trial. Formal entry of the stipulated final order by the court could not be confirmed as of 29 July 2026 and is recorded in the unverified list below.

Confidence: Medium-High

Primary sources fetched directly where possible. Everything we could not verify is listed at the bottom of this page by name.

What this actually is

Follow the money

An Arizona-based seller of an AI voice agent that ran two distinct offers, and conflating them is the most common error in third-party write-ups. The “Access Card” at $15,000 to $30,000 bundled software access with coaching and was sold over the telephone, which is why the Telemarketing Sales Rule count attaches to it specifically. The “licensing business opportunity” at $9,800 to $100,000 was a tiered agency-in-a-box: the right to use and resell the software to other businesses, a thousand pre-vetted qualified business owner leads, coaching and group consulting calls, access to a training platform, and a share of the revenue generated by call usage from businesses the licensee referred. Both carried per-minute charges on top of the entry fee, and telephony was billed separately again.

The Business Opportunity Rule is the spine of this file and it is the most useful thing a reader takes away, so it is worth setting out in plain English. Under 16 C.F.R. Part 437, an offer is a covered business opportunity when three things are true together: the seller solicits someone to enter a new business, that person has to make a required payment, and the seller represents that it will provide locations or accounts, supply outlets or customer leads, or buy back what the purchaser produces. When the Rule applies, the seller must hand the buyer a one-page disclosure document at least seven days before any signature or payment; and if the seller makes any earnings claim, it must provide a written earnings claim statement giving the timeframe, the characteristics of the purchasers who achieved that result, and - the part that decides everything - the number and percentage of all persons who bought the opportunity and achieved at least that level of earnings, with written substantiation available on request.

The FTC’s charging decision resolves the applicability question. The complaint pleads a new business (an agency reselling AI voice agents), a required payment ($9,800 to $100,000), and the third element satisfied twice over - a promise of a thousand pre-vetted leads, and a promise of a full buy-back. The general lesson the Commission is signposting is worth reading twice by anyone selling software with an income story attached: wrapping a software license in leads plus a buy-back plus earnings claims converts a SaaS sale into a regulated business opportunity, with all the disclosure obligations that follow. The complaint then pleads that defendants furnished neither the disclosure document nor any earnings claim statement. What buyers got instead of a denominator was a list of numerators.

The offer no longer exists in any form. As observed on 29 July 2026, air.ai resolves to “Air,” the rebranded identity of defense-analytics firm Govini - an entirely unrelated enterprise platform with federal security accreditations and a request-a-demo motion, no voice agent, no license and no pricing. Independent trackers list the original product as defunct. The stipulated order forecloses the offer permanently as to its three owners. A handful of search-driven “review 2026” pages still write about the product in the present tense and quote live-sounding prices; those are stale affiliate comparison pages, not evidence of an active offer. Anyone encountering an Air AI license for sale today should treat it as the resale of a dead asset or an unrelated party trading on the name.

The redress arithmetic

Approximately $19 million in alleged consumer losses, as pleaded in the FTC complaint, set against the $18 million monetary judgment in the stipulated order of 24 March 2026 and the amount actually payable under it. The judgment is largely suspended on an inability-to-pay basis; suspension of that kind is typically subject to reinstatement if the financial representations that supported it prove untrue.

94%
Suspended on an inability-to-pay basis ($17,950,000)Alleged loss above the judgment amount ($1,000,000)Actually payable for consumer redress ($50,000)
ProductPricePays
Licensing business opportunity (the agency-in-a-box)
Tiered with purported benefits increasing the more a consumer paid. Included the right to resell the software, “a thousand pre-vetted, qualified business owner leads,” coaching, group consulting calls, a training platform and a support channel. This is the product charged under the Business Opportunity Rule.
$9,800–$100,000
one-time, tiered
resale margin + usage share
Air AI Access Card
Software access plus coaching, sold by telephone - which is why the Telemarketing Sales Rule count attaches here. The complaint alleges the price was adjusted to each consumer’s ability to pay. Guarantee offered: double or triple the investment within six months or a full refund, plus an unconditional 72-hour refund.
$15,000–$30,000
one-time
Outbound calling
A BBB complainant who paid $50,000 reports the rate rising from 11¢ to 33¢ after purchase; a petition organiser independently describes actual costs three times the advertised claim. A 3x input-cost move destroys the unit economics of every client contract already signed.
$0.11/min advertised
per minute
Inbound calling
Roughly four to six times the per-minute rate of open-market platforms that charge no entry fee at all. Realistic all-in was reported at $0.19 to $0.32 and above.
$0.32/min
per minute
Telephony
Twilio charges sat on top of the per-minute platform rate rather than inside it. Reviewers consistently describe the quoted minute price as an understatement of the delivered cost.
billed separately
usage
In-house financing
The sellers originated loans through their own lending platform and brokered third-party loans. A seller that finances the purchase of its own income opportunity at near-credit-card rates converts a buyer’s optimism into a debt obligation that survives the product’s failure. One BBB complaint records financing remaining active after service was canceled.
up to 24.99% APR
on the balance
Contract minimum
No free trial. Every mainstream competitor in the category offers self-serve signup, month-to-month terms and trial credit.
12 months
term
Third-party services to make it work
Recorded in a BBB complaint. Integration and deployment support gaps are the dominant complaint theme, rather than the advertised count of 5,000-plus app integrations.
$10,000+ in one documented case
as needed
Background check

Who runs it, and what they ran before

CM
Caleb Matthew Maddix
Co-founder, public-facing chief executive, managing member; Arizona resident

Born June 2000. His own published biography is an influence-marketing résumé rather than an operating one: a children’s book about success co-authored with his father at 14, nine books in total through a family publishing imprint, stages shared with Grant Cardone, Gary Vaynerchuk, Kevin Harrington and Tony Robbins, early coverage in national and local press, and a self-asserted claim to have become a millionaire at 16 for which no supporting financials could be located. The one prior paid program identifiable by name is a YouTube-automation course sold at $997 upfront or two payments of $597 - the same genre as the Air AI license, two orders of magnitude cheaper. No prior collapsed venture could be found, and this is a first federal enforcement action rather than a repeat, which is worth stating fairly. What the record does not contain is any evidence of having built and shipped software at scale before attempting a deep-technology product.

RP
Ryan Paul O’Donnell
Co-founder and managing member; Virginia resident

Named as an individual defendant in the FTC complaint and covered by the permanent business-opportunity marketing ban in the stipulated order. Public information is thin and consists mainly of scraped-profile aggregators. No prior ventures, prior enforcement actions or operating track record could be confirmed either way, which is recorded here as an absence of information rather than as a finding.

TM
Thomas Matthew Lancer
Principal and co-founder; Idaho resident

Named as an individual defendant and likewise covered by the permanent ban. As with his co-defendant, the public record is limited to aggregator profiles; no prior business history could be verified. Both men consented to the stipulated order without admitting liability.

On
Ownership note
Reputation seeding around the lead founder

A Medium account under a handle combining the founder’s name with the company’s published a run of near-identical, self-laudatory profile pieces about him, one carrying a machine-mangled title in which “artificial intelligence” had been paraphrased into nonsense. That is spun content rather than editorial coverage, and it is the sort of material a prospective buyer typing “is this founder credible?” into a search box would have hit first. It proves nothing about the product. It says a good deal about the method, and it was visible before anyone wired money.

Registered address

Phoenix, Arizona, USA
Six corporate entities - Air AI Technologies, Inc. plus five Arizona LLCs - shared a single principal business address at 4742 N 24th Street, Suite 300, Phoenix, corroborated on the BBB profile. The complaint alleges more than $23,631,000 in commingled transfers between those corporate accounts. Customers reported being charged and contacted under a trading name, Scale 13, that did not match the brand they had bought. A six-entity cluster with multiple trading names and heavy intercompany movement is a structure that makes participant recovery difficult by construction, whatever the intent behind it. The company took no institutional venture capital: funding is listed at $0 and the business described as bootstrapped, which matters because a model built on large upfront payments and a full buy-back promise then has only one source of cash for refunds, which is new participant inflow. The complaint records the defendants’ own words on that point - “we have spent every last dollar that we have as owners. We are completely out of cash at this moment.” Self-reported figures put revenue at roughly $3.5 million of annualised recurring revenue as of June 2025 with 32 employees; the FTC pleads approximately $19 million of consumer losses over a few years, and the gap between those two numbers is the clearest available indication that the intake was entry fees rather than software subscriptions.

Compensation plan

What has to be true for you to get paid

To coverYou need
Recover a $25,000 license from agency margin ~50 client-months
at roughly $500 a month of net margin per reselling client, before any advertising cost
Recover the $50,000 license named in two first-person accounts ~100 client-months
against a pleaded finding that many licensees could not resell to anyone at all
Service financing at 24.99% APR on a $25,000 balance ~$6,250 a year in interest alone
payable whether or not the agent places a single call
Beat the open market on input cost not achievable at any volume
$0.11–$0.32/min plus telephony against $0.05–$0.09/min with no entry fee

Read this twice

This arithmetic is unusual on this site because the honest version of it cannot be completed. Every other breakeven table here is built from a published income disclosure; here the FTC pleads that no disclosure document and no earnings claim statement were ever furnished, so there is no denominator to divide by. What exists instead are individual timelines from the complaint record and the consumer file, and they all run the same direction: an AI agent not built after a month with $10,000 paid and no communication; a refund requested after six months of zero sales; eighteen months unable to use the software with no company response; one licensee abandoning the sales effort after months of troubleshooting persistent technical failures. The closest thing in the file to a population statistic is the FTC’s own framing - that many consumers who purchased the licensing business opportunity are not able to resell it to anyone, because the software did not work. The client-month figures above are therefore a model of what the offer would have required had the product functioned, not a description of what buyers achieved. Two things the model deliberately excludes because they cannot be quantified from the record: incremental advertising spend borne by licensees, which is nowhere in the file, and telephony-consent exposure from dialing supplied lead lists with a synthetic voice, which is a per-call statutory risk with no dollar estimate attached. The honest statement for a material share of buyers is that the realistic time to a first paying client was never.

Run your own numbers

Drag the sliders. Nothing here is stored or sent.

-
Cumulative net, after costs
Total referred license buyers -
Commission that month -
Total commissions earned -
Total you paid in -
Net -

HISTORICAL - the offer is no longer sold and the three owners are permanently barred from marketing any business opportunity under a stipulated order. This models what the reseller pitch claimed: an override on a referred license sale, against the entry license spread over twelve months. The realistic preset is zero referrals for a reason - the regulator’s complaint pleaded that most licensees could not resell to anyone at all. Your own subscription cost of $817/mo is included.

Your money

What it costs to replace this yourself

This is the exercise that decides the file, and it is unusually clean because the comparators are self-serve, publicly priced and available to anyone with a credit card. The question is not whether an AI voice agent is a real thing to sell - it is, the category is legitimate and growing. The question is what the license fee bought that the open market did not already sell at a fraction of the price, with no entry fee, no term commitment and free trial minutes.

What they sell youWhat you'd use insteadYour cost
License entry fee - $9,800 to $100,000Self-serve signup on an open-market voice platform$0
Access Card - $15,000 to $30,000The same capability with no card, no coaching upsell$0
Outbound at $0.11/min advertised, reported at $0.33Retell at $0.07/min with no platform fee~$0.07/min
Inbound at $0.32/minVapi at $0.05/min platform fee, roughly $0.15 all-in with telephony and models~$0.05–$0.15/min
Agency tier of the licenseVapi Agency plan at $500/mo, or Bland from $299/mo at pilot scale$299–$500/mo
12-month minimum, no free trialMonth-to-month, with 60 free minutes on one platform to test before paying$0
Financing at up to 24.99% APRNothing to finance - there is no entry fee to spread$0
“A thousand pre-vetted, qualified business owner leads”Your own list, with consent capture you control and can provead spend only
Coaching - delivered as group calls where one-to-one was promisedVendor documentation, public benchmarks and community forums$0
Measured latency above 1,000 ms with multi-second dead airRoughly 620 ms measured on one platform; 700–1,500 ms on another under real call conditionsincluded
Total as sold
$30,000–$80,000+ in year one
Total, built yourself
$200–$2,000 a month with no entry fee

Price-to-value

The gap is categorical, not marginal. The per-minute rate alone was roughly two to six times the open-market equivalent, before an entry fee that competitors do not charge at all, for a product measured slower and less reliable than both. Every one of those vendors offers self-serve signup and trial minutes, so any buyer could have run the comparison in an afternoon for nothing. The decisive test is simple: would a rational buyer have purchased this if no income offer were attached? On this pricing and this measured performance, no - not at any tier. Nobody pays $25,000 for access to a slower agent when a faster one is $0.07 a minute with a free trial. The license fee did not buy capability. It bought the income story.

Odds of profit

Three operators, five horizons

Probability of cumulative net profit

Hover any point for median, top decile and bottom quartile.

0% 25% 50% 75% 100%3 mo6 mo1 yr3 yr5 yr 3% 4% 1%
Access Card buyer - paid roughly $20,000 by telephone for software plus coaching, ran it as a tool in an existing businessEntry-tier licensee - $9,800 to $25,000 for the agency license, paid in cash, worked the supplied lead listHigh-tier financed licensee - $50,000 to $100,000, borrowed at up to 24.99% APR, hired staff against the promise

Access Card buyer

paid roughly $20,000 by telephone for software plus coaching, ran it as a tool in an existing business

HorizonP(profit)Median
3 mo 2% −$21,000
6 mo 2% −$23,000
1 yr 3% −$26,000
3 yr 3% −$26,000
5 yr 3% −$26,000

Entry-tier licensee

$9,800 to $25,000 for the agency license, paid in cash, worked the supplied lead list

HorizonP(profit)Median
3 mo 1% −$16,000
6 mo 2% −$18,000
1 yr 3% −$21,000
3 yr 4% −$21,000
5 yr 4% −$21,000

High-tier financed licensee

$50,000 to $100,000, borrowed at up to 24.99% APR, hired staff against the promise

HorizonP(profit)Median
3 mo 0% −$54,000
6 mo 0% −$59,000
1 yr 1% −$68,000
3 yr 1% −$88,000
5 yr 1% −$96,000

Methodology note. ANCHORED to the published record: the $9,800–$100,000 license band and the $15,000–$30,000 Access Card band pleaded in the FTC complaint; the $0.11 outbound and $0.32 inbound per-minute rates, and the 11¢-to-33¢ increase reported by a $50,000 buyer and corroborated by a petition organiser; financing at up to 24.99% APR through the sellers’ own platform; the twelve-month contract minimum with no free trial; the $10,000-plus of third-party services recorded in one BBB complaint; the $30,000–$80,000 realistic year-one total cited by two independent reviewers; and the FTC’s pleaded characterisation that many licensees could not resell the license to anyone. MODELED by us: every percentage, every median and both tails. There is no cohort data in existence for this offer, because - as the complaint pleads - no earnings claim statement was ever produced, so nobody, including the FTC, has published the number and percentage of purchasers who achieved any stated result. The share in cumulative profit is set near zero rather than at zero because the record does contain documented refunds, including one paid after nine months and a complaint to a state attorney general and one recorded by the BBB as approved - recovery was possible, if slow and adversarial, and the “top” column is largely a refund column rather than a trading-profit column. The flat medians after year one reflect the offer ceasing to exist: costs stop because the product stops, not because anything improved. Treat these tables as an illustration of the shape of the loss, not as measurement.

Go-to-market

Where you are actually allowed to promote this

Platform policy reads, not verifications. Check every one before you spend a dollar - enforcement changes faster than the written policy does.

Channel
Status
Notes
Marketing any business opportunity, by the three owners
PERMANENTLY BANNED BY STIPULATED ORDER
The order announced on 24 March 2026, which the defendants agreed to, permanently bars the corporate defendants and Caleb Maddix, Ryan O’Donnell and Thomas Lancer from selling or marketing any business opportunity. The ban is specific to business opportunities, telemarketing misrepresentation and unsubstantiated earnings claims; it does not bar them from operating a business generally.
The seven-day disclosure document
PLEADED AS NEVER FURNISHED
The Business Opportunity Rule requires a one-page disclosure document in the buyer’s hands at least seven days before any signature or payment. The complaint states that defendants do not provide consumers with disclosure documents as required by the Rule. That is the count that makes this file what it is.
Earnings claim statements
PLEADED AS NEVER FURNISHED
Any seller making an earnings claim must give a written statement setting out the timeframe, the characteristics of purchasers who achieved the result, and the number and percentage of all purchasers who achieved at least that level - with substantiation on request. The complaint states none were furnished, against claims including “$450,000 in revenue in their first thirty days” and “$79,620 in the past 30 days.”
Telephone sales of the Access Card
CHARGED UNDER THE TELEMARKETING SALES RULE
Count III pleads violations of 16 C.F.R. Part 310 as to the Access Card specifically, because it was sold by phone. The complaint describes a five-step funnel: social advertising, a qualifying call in which prospects were “uniformly assured” they were a “perfect” candidate, a mandatory pre-call video doubling down on earnings claims, a closing call using a “limited number of Licenses,” and the refund guarantee deployed to overcome the final objection.
Reseller advertising and income-claim policy
NONE LOCATED
No licensee-facing advertising policy, income-claim policy or compliance guide could be found anywhere in the public record. In most files on this site the policy is the constraint; here there appears not to have been one, and the earnings claims were propagated by the company’s own staff rather than by an arms-length affiliate layer there was any need to police.
Outbound AI voice dialing by licensees
REGULATED SINCE FEBRUARY 2024 - AND THE RISK SITS ON THE BUYER
On 8 February 2024 the FCC issued a unanimous declaratory ruling holding that the telephone-consumer statute’s restrictions on artificial or prerecorded voice calls cover AI-generated voices. Practical effect: prior express consent, and prior express written consent for marketing calls, plus caller identification and a working opt-out. Damages are statutory and per call. A licensee handed a thousand leads and told to dial them with a synthetic voice was pointed at this with no compliance tooling in the box.
Recording or transcribing calls into all-party consent states
LICENSEE’S EXPOSURE, UNADDRESSED
Several states require all-party consent to record, and cross-border calls generally take the stricter rule; several have moved toward affirmative AI-disclosure duties on the call itself. Nothing in the license description reviewed references consent capture, do-not-call scrubbing or state-aware recording disclosure. This is the single most under-discussed participant risk in the file.
Social media advertising by the company
DISPUTED CLIENT ATTRIBUTIONS
A petition organiser alleges Instagram advertising falsely attributed client campaigns to Apple and Tesla. Whether those specific advertisements ran is unverified and is recorded as an allegation by an interested party, not a finding - but it is consistent with the BBB’s July 2024 substantiation challenge, which the business did not answer.
Buying a license today
THERE IS NOTHING LEFT TO BUY
As observed on 29 July 2026, air.ai serves an unrelated defense-analytics platform. Independent trackers list the original product as defunct. Stale search-driven review pages still quote live-sounding pricing; treat any Air AI license offered for sale now as the resale of a dead asset or an unrelated party trading on the name.
The evidence

Red flags and green flags

Red flags

15
1A permanent federal ban on all three owners, by consent
The stipulated order announced 24 March 2026 bars Caleb Maddix, Ryan O’Donnell and Thomas Lancer from selling or marketing any business opportunity, permanently. It is a consent settlement with no admission of liability and no judicial finding of fraud after trial - and it is still the most severe structural relief available short of one.
2No disclosure document and no earnings claim statement, ever
The FTC complaint pleads both, flatly. Under 16 C.F.R. Part 437 the first must be handed over seven days before payment and the second must carry the number and percentage of all purchasers achieving the claimed result. Buyers received numerators with no denominator.
3The core product could not do its core job
Per the complaint, the conversational agent was “often unavailable,” malfunctioned when present, could not perform outbound calling, scheduling or accurate transcription, required extensive pre-scripting and produced fabricated responses. The analytics tool never materialised. The financing tool worked.
4The guarantee was the close, and it was not funded
Refund and buy-back promises were used to overcome the final objection, then “very rarely” honored per the complaint, with buyers “strung along for months.” The defendants’ own words, quoted in the pleading: “we have spent every last dollar that we have as owners. We are completely out of cash at this moment.”
5The buy-back clause had a floor and no ceiling
The agreement promised a full buy-back for any reason at all “but no sooner than 120 days after signing.” There is no stated maximum time to pay. That drafting choice makes indefinite delay contractually survivable, and indefinite delay is what buyers reported.
6Price was allegedly set by ability to pay
The complaint alleges defendants adjusted the price based on consumers’ ability to pay, extracting as much money from each individual consumer as possible. A price that moves with the buyer’s bank balance describes an extraction model rather than a product one.
7In-house lending at up to 24.99% APR
The sellers originated loans through their own platform and brokered third-party loans, so a buyer’s optimism became a hard debt obligation that outlived the product. One BBB complaint records financing remaining active after the service was canceled.
8The per-minute rate reportedly tripled after purchase
11¢ to 33¢, reported by a BBB complainant who paid $50,000 and corroborated independently by a petition organiser describing costs three times the advertised claim. For a licensee who had modeled client contracts at 11¢, a 3x input-cost move destroys the economics of every contract already signed.
9Six commingled entities and a second trading name
More than $23,631,000 alleged to have moved between corporate accounts across a six-entity Arizona cluster sharing one address, trading as both Air AI and Scale 13. Customers reported being charged under a name that did not match the brand they bought.
10BBB rating F, with 21 of 22 complaints unanswered
Not accredited, two alerts on file - the July 2024 advertising review and the August 2025 pending government action. A BBB rating is a private ratings body’s assessment rather than a regulator’s finding; the unanswered count is the part that carries information.
11The review distribution is not mixed, it is one-sided
The complaint cites a snapshot of 78 one-star reviews against 3 positive. An independent software directory shows 1.5 out of 5 across 113 reviews. Representative: “We paid them in full and we never received the product.”
12Price-to-value inverted against the open market
Two to six times the per-minute cost of self-serve competitors, plus a five- to six-figure entry fee those vendors do not charge at all, for measurably worse latency - above 1,000 milliseconds with reports of multi-second dead air, against roughly 620 milliseconds measured elsewhere.
13Scarcity and pressure engineered into the funnel
A mandatory pre-call video described in the complaint as doubling down on earnings claims, a uniform assurance that every prospect was a “perfect” candidate, and a closing call built on a “limited number of Licenses.” None of those are features of a product that sells on its merits.
14Telephony-consent liability pushed onto buyers with no support
Since the FCC’s February 2024 ruling, AI-generated voices are treated as artificial voices under the telephone-consumer statute, requiring prior express - and for marketing, prior express written - consent, with per-call statutory damages. Nothing in the license description references consent capture, do-not-call scrubbing or state recording rules.
15The founder’s track record was promotion, not product
Books, stages, a family publishing imprint and a $997 YouTube-automation course. No documented operating business that built and shipped software at scale before a first-time deep-technology build sold at five and six figures. That was all visible in public before anyone wired money.

Green flags

9
1There was no recruitment matrix
No levels, no downline depth, no genealogy, no rank advancement, and no commission for merely enrolling another licensee into a structure. Whatever else this was, it was not an endless chain, and a reader should not import that assumption from the category.
2No token, no staking, no passive-return offer
Nothing in the record describes an investment contract, a pooled fund, a yield on capital or a withdrawal gate. Buyers were told to go out and work an agency. That is why the securities dimension scores 8 in a file graded F - it is a genuinely separable fact and it is scored honestly.
3The license bought a real piece of software, not a place in a matrix
Non-functional software, on the FTC’s pleading - but a product with a specification, a platform and integrations, sold to be used. That is a different failure from paying for a position, and the distinction matters when a reader is deciding what lesson to carry into the next offer.
4No independent affiliate sales force earning recruitment commission
The complaint describes all sales activity as run by defendants’ own staff, with no external closers or affiliate telemarketers identified. That removes the classic recruiters-recruiting-recruiters dynamic and keeps responsibility for the claims with the company that made them.
5A federal regulator did act, and the file is complete and public
A prospective participant can now read the complaint, the docket and the settlement rather than guess at economics. That transparency arrived through a federal agency rather than the company - but it arrived, and it is more than most offers in this category ever produce.
6A court order has been in force since September 2025
The stipulated preliminary injunction entered 10 September 2025 is confirmed in force. Between filing and settlement there was continuous judicial supervision rather than a company left free to keep selling while a case moved slowly.
7Redress exists at all
$50,000 against roughly $19 million of alleged losses is about a quarter of a cent on the dollar and nobody should describe it as compensation. It is still more than the nothing that follows most collapses of this shape, and the suspension of the remaining judgment is typically reversible if the financial representations behind it prove untrue.
8The offer is no longer being sold
The domain now serves an unrelated defense-analytics platform, independent trackers list the product as defunct, and the stipulated order forecloses the offer permanently as to its owners. Nobody else is going to lose $50,000 to this specific license.
9No prior collapsed venture by the lead founder
This is a first federal enforcement action rather than a repeat pattern, and no earlier failed company could be located in his history. It is a fair point in mitigation on the serial-offender question, and it is recorded as such even though it says nothing about competence.
What would move this grade

We would like to be wrong about this

Upward

  • Verified entry of the stipulated final order followed by redress actually reaching buyers - $50,000 against roughly $19 million of alleged losses is about a quarter of a cent on the dollar, and meaningful distribution would move the operator-conduct and terms scores.
  • A published earnings claim statement with a real denominator: the number and percentage of all licensees who achieved the results claimed. It never existed, and it would be the single largest upward mover if it appeared.
  • Independent evidence of licensees who reached profitability, together with a functioning product at market price - sub-700 millisecond latency, reliable outbound calling, no entry fee, month-to-month terms and compliance tooling shipped in the box.

Downward

  • Entry of the stipulated final order confirmed with no redress distributed, or a finding that the defendants misrepresented their finances, which typically triggers reinstatement of the full suspended judgment.
  • A successor offer under new branding by any of the three principals, which would test the permanent business-opportunity ban directly; or state attorney general action or a certified class action adding a second enforcement layer.
  • Evidence that the launch demo was fabricated rather than a scripted best case, which would convert a substantiation failure into an outright fabrication; or evidence of licensees incurring telephone-consumer liability from dialing the supplied lead lists.
The better trade

Grade is F. A five-figure license to resell an AI phone agent that, per a federal complaint, could not reliably place a call - with no disclosure document, no earnings statement, and a buy-back the sellers said they had no cash to fund.

Start with what is fair, because it is real and it changes what a reader should take away. This was not an endless chain. There were no levels, no downline, no genealogy and no rank to climb; nobody was paid for enrolling another licensee into a structure. There was no token, no staking, no pooled fund and no promise of a passive return, which is why the securities dimension scores 8 here - buyers were sold work, and the work was reselling software to businesses that would actually use it. On paper that is compensation tied to sales to end users, which is the right side of the line the regulators draw. The underlying market is real too: AI voice agents are a legitimate and growing category served today by credible vendors at credible prices. The demand Air AI pointed at exists. Air AI simply did not serve it.

The Business Opportunity Rule is where the file turns, and it is the part worth learning. Under 16 C.F.R. Part 437, when a seller solicits someone into a new business, requires a payment, and then promises customer leads or a buy-back, the offer becomes a regulated business opportunity. The seller must hand over a one-page disclosure document seven days before any money changes hands, and if it makes any earnings claim at all it must provide a written statement giving the timeframe, the characteristics of the buyers who achieved that result, and the number and percentage of everyone who bought and reached that level. Air AI promised a thousand pre-vetted leads and a full buy-back, which satisfies the third element twice over, and it made earnings claims constantly - $450,000 in the first thirty days, $20k in three days, a million dollars a year. The complaint pleads that it furnished neither document. That is the whole lesson of this case for anyone selling software with an income story bolted on: leads plus a buy-back plus earnings claims turns a SaaS sale into a regulated offer, with everything that follows.

Then the commercial reality, which needed no regulator to see. The conversational agent was, per the complaint, often unavailable, unable to place outbound calls or transcribe accurately, and prone to fabricated responses; the only tool the FTC describes as fully operational is the one that lent buyers money at up to 24.99% APR. Independent testing put latency above a second with multi-second dead air, against roughly 620 milliseconds for an open-market alternative charging $0.07 a minute with no entry fee and free trial credit. A buyer could have run that comparison in an afternoon, for nothing, before wiring $25,000. The decisive question is the one to carry into the next offer of this shape: would you buy this product at this price if no income opportunity were attached to it? If the answer is no, the income story is doing all the work - and the income story is precisely the thing the seller is legally required to substantiate with a denominator, and precisely the thing that was never provided here.

1

Ask for the disclosure document before anything else

If an offer requires a payment, puts you into a new business, and promises leads or a buy-back, the Business Opportunity Rule applies and you are entitled to a one-page disclosure document seven days before you sign or pay. Ask for it in writing. A seller who cannot produce it has told you the most important thing about the offer in the cheapest possible way, and the request costs you nothing.

2

Demand the denominator, not the testimonials

Any earnings claim triggers a written earnings claim statement giving the number and percentage of all purchasers who reached that level, with substantiation on request. “One client did $79,620 last month” is a numerator. Ask how many people bought and how many reached that figure. If the answer is a screenshot, you have your answer.

3

Price the software as software, with the income story removed

Open-market voice platforms run $0.05 to $0.09 a minute with self-serve signup, month-to-month terms and free trial minutes, and small deployments cost $200 to $2,000 a month all-in. Build the agency on one of those. If a license costs five figures on top of a worse per-minute rate, the fee is buying the narrative, not the capability - and you can test the capability yourself before paying anyone.

4

Price the compliance you would be taking on

Since February 2024 an AI voice is an artificial voice under the telephone-consumer statute: prior express consent, written consent for marketing, caller identification, a working opt-out, and per-call statutory damages if you get it wrong. Add all-party recording consent in several states. Anyone handing you a lead list without consent records and without do-not-call scrubbing is handing you the liability, not the leads.

The only component the FTC describes as fully operational is the one that lent buyers the money.
Scorecard

Nine dimensions, weighted

Comp structure & KoscotDoes the plan pay for recruitment or for sales to real customers?
20%
3.0
On paper the architecture sits on the right side of the line, and that has to be said first. There were no levels, no downline depth, no genealogy, no rank advancement and no commission for merely enrolling another licensee. Pay was framed as margin on reselling software to real businesses, plus a share of the call usage those end businesses generated. That is compensation tied to sales to end users. Three rather than seven because the complaint pleads the paper did not describe the operation: many licensees could not resell the license to anyone, because the software did not work. When nobody downstream can sell, the only money entering the system is the next entry fee. Design earns credit; operation removes most of it.
Securities exposureAny passive return on capital? Howey, staking, tokens, withdrawal friction.
15%
8.0
Eight, inside a file graded F, and the reason is the point of scoring nine dimensions separately. Buyers were sold active work, not a passive return. Nothing in the record describes an investment contract, a token, staking, a pooled fund, a withdrawal gate or a yield on capital - the pitch was to run an agency and resell software you had to go out and place. No securities regulator has been involved at any stage and no Howey exposure to the participant could be located. That is a real, separable fact and it should not be dragged down by everything around it. Two points are withheld because the usage override was marketed with claims - “$1MM Every Single Month paid out” - pitched close enough to a return to blur the distinction.
Ownership & track recordWho runs it, what did they run before, and what happened to it.
15%
1.0
The stipulated order to which all three owners agreed on 24 March 2026 permanently bans them from selling or marketing any business opportunity. That is the most severe structural relief available short of a fraud judgment after trial, and it was consented to rather than litigated. Behind it: six commingled entities, multiple trading names, more than $23.6 million alleged to have moved between corporate accounts, and the defendants’ own statement that they were “completely out of cash.” In mitigation, and it is genuine, no prior collapsed venture by the lead founder could be found and this is a first enforcement action, not a repeat. It is not a point in favor of competence that the record before it was selling opportunity rather than building product.
Product reality & demandWould a rational buyer purchase this if no income offer existed?
12%
1.0
The complaint itemises the suite. The analytics tool never materialised. The platform worked but was limited. The conversational agent at the center of everything - the actual product - was “often unavailable,” malfunctioned when present, could not perform outbound calling, scheduling or accurate transcription, required extensive pre-scripting and produced fabricated responses. The only component the FTC describes as fully operational is the one that lent buyers money. Independent latency testing puts the agent above 1,000 milliseconds with reports of multi-second dead air, against roughly 620 milliseconds measured for one open-market competitor. Multi-second silence on a live sales call ends the call. Marketing promised phone conversations of ten to forty minutes indistinguishable from a human being.
Participant economicsReal cost in, realistic money out, and whether they publish the numbers.
10%
0.0
A floor case, and not by inference. The FTC pleads flatly that defendants do not provide consumers with disclosure documents as required by the Business Opportunity Rule and do not furnish earnings claims statements. So there is no denominator anywhere in this file - no cohort data, no number and percentage of purchasers achieving any stated result, nothing. What buyers received instead was a list of numerators: $450,000 in the first thirty days, $20k in three days, $79,620 in the past thirty days, a million dollars a year. The closest thing to a population statistic is the FTC’s own characterisation that many licensees could not resell to anyone at all. Zero is the correct score when the disclosure is not weak but absent.
Price-to-valueWhat the same capability costs on the open market.
8%
1.0
The comparison is categorical rather than marginal. Open-market voice-agent platforms charge roughly $0.05 to $0.09 a minute with no entry fee, self-serve signup, free trial minutes and month-to-month terms; typical all-in operating cost for a small deployment runs $200 to $2,000 a month. Air AI charged $0.11 a minute advertised for outbound and $0.32 for inbound, with telephony billed separately on top, a twelve-month minimum, no free trial - and a five- to six-figure entry fee that no competitor charges at all. A BBB complainant who paid $50,000 reports the per-minute rate rising from 11¢ to 33¢ after purchase, corroborated independently by a petition organiser describing costs three times the advertised claim.
Payout sustainabilityCan the company fund the plan out of margin, or only out of inflow?
8%
1.0
The buy-back was the close, and the complaint alleges it was not funded. The license agreement promised a full buy-back of the amount paid for any reason at all, “but no sooner than 120 days after signing” - a floor with no stated ceiling, which makes indefinite delay contractually survivable. Buyers report exactly that: a refund promised in January 2024 then delayed a further six months; a 120-day buy-back deferred with the owners citing insufficient funds; a nine-month wait resolved only after a complaint to a state attorney general. The FTC pleads that guarantees were “very rarely” honored. Against roughly $19 million of alleged losses, $50,000 is actually payable.
Marketing conductIncome claims, regulator run-ins, hype, deadline stacking.
7%
0.0
The company itself propagated the earnings claims, through social advertising, a mandatory pre-call sales video that the complaint describes as doubling down on earnings claims, and closing calls run by its own staff using scarcity - a “limited number of Licenses” - and a uniform assurance to every prospect that they were a “perfect” candidate. The BBB formally challenged two headline claims in July 2024, that the product could replace 100,000 sales and customer service representatives and hold ten-to-forty-minute human-sounding calls, and the business did not respond. A petition organiser alleges Instagram advertising falsely attributed client campaigns to Apple and Tesla; whether those specific advertisements ran is unverified. No income-claim or advertising policy of any kind could be located.
Operator terms & exitWho owns the customer, what you forfeit, how hard it is to leave.
5%
1.0
Price was not fixed: the complaint alleges it was adjusted to each consumer’s ability to pay, to extract as much from each individual as possible. A price that moves with the buyer’s bank balance is an extraction model rather than a product one. The buy-back clause sets a 120-day floor and no ceiling. The contract ran twelve months minimum with no free trial. Purchases were financed through the sellers’ own lending platform at rates up to 24.99% APR, and a BBB complaint records financing remaining active after service was canceled - the debt outlives the product. No reseller advertising or compliance policy was ever located, so licensees dialing supplied lead lists with a synthetic voice carried the telephony-consent exposure alone.
Weighted composite
2.30
F

Dimension profile

Further from center is better. Hover any point.

Comp structure& Koscot 3.0 Securitiesexposure 8.0 Ownership &track record 1.0 Product reality& demand 1.0 Participanteconomics 0.0 Price-to-value 1.0 Payoutsustainability 1.0 Marketingconduct 0.0 Operator terms& exit 1.0

Hard caps that bind here

Cap at F three individual owners are permanently banned by a stipulated federal order from selling or marketing any business opportunity, and the complaint pleads that the offer furnished neither the disclosure document nor the earnings claim statement that the Business Opportunity Rule requires. Those two facts are not gradations of a weak file - they are the outer boundary of one. A permanent ban consented to by the principals removes the operator from the category entirely, and an offer that provided no denominator at all cannot be underwritten by a buyer on any evidence, because there was no evidence to give them. No score on any other dimension can lift a file past that.
Cap at D− the product could not perform its core function. The complaint records that the conversational agent was often unavailable, could not reliably place outbound calls, schedule or transcribe accurately, and produced fabricated responses - while the financing tool worked. Even with the enforcement file set entirely to one side, a five-figure license to resell software that does not run is uneconomic at every tier, against open-market alternatives at $0.05 to $0.09 a minute with free trial credit and no entry fee. This cap is recorded separately because it would apply on the commercial evidence alone.

The lowest binding cap wins, regardless of the weighted arithmetic.

Sources consulted

What we read

Every source below links to the document itself. Tier 1 is a primary record - the company’s own plan, policy or disclosure, a court filing, a regulator’s decision or an SEC filing. Tier 2 is a self-regulatory or secondary regulator record, tier 3 reporting or academic work, tier 4 an open-market price comparison. Where a document can be moved or withdrawn, an archived copy is linked beside it. If a link is dead when you try it, that is a correction we want.

  1. FTC v. Air Ai Technologies, Inc. et al. - Complaint for Permanent Injunction, Monetary Judgments, and Other Relief, No. 2:25-cv-03068-SMB (D. Ariz., filed 25 August 2025) (PDF)
    RegulatorTier 1Federal Trade Commission · 2025-08-25archived copy

    FTC v. Air Ai Technologies, Inc. et al., Complaint filed 25 August 2025, D. Ariz. 2:25-cv-03068-SMB before Judge Susan M. Brnovich - the $9,800–$100,000 license band, the $15,000–$30,000 Access Card, pricing set by ability to pay, the “thousand pre-vetted, qualified business owner leads,” the 120-day buy-back clause quoted verbatim, financing at up to 24.99% APR, the four counts including the Business Opportunity Rule at 16 C.F.R. Part 437, the approximately $19 million of consumer losses, the $23,631,000 of intercompany transfers, and the pleading that no disclosure documents and no earnings claim statements were furnished - FTC complaint, software inventory - the analytics tool that never materialised, the platform described as functional but limited, the operational in-house lending tool, and the conversational agent described as often unavailable and unable to perform outbound calling, scheduling or accurate transcription while producing fabricated responses; also the group-only coaching against a one-to-one promise

  2. Federal Trade Commission v. Air Ai Technologies Incorporated et al., No. 2:25-cv-03068 (D. Ariz.) - docket, Judge Susan M. Brnovich
    Court recordTier 1Justia Dockets & Filings (U.S. District Court for the District of Arizona) · 2025-08-25archived copy
  3. "FTC Sues to Stop Air AI from Using Deceptive Claims about Business Growth, Earnings Potential, and Refund Guarantees to Bilk Millions from Small Businesses" - FTC press release, 25 August 2025
    RegulatorTier 1Federal Trade Commission · 2025-08-25archived copy

    FTC press release, 25 August 2025, announcing the suit and the motion for a temporary restraining order; FTC case page for Air.ai; docket records showing the stipulated preliminary injunction entered 10 September 2025, the answer with jury demand filed 31 December 2025, and the joint motion to stay for settlement granted in February 2026

    Not established by this document: The 31 December 2025 answer with jury demand is not individually addressable on either free docket mirror; it is reflected only in the docket text, not as a linkable filing.

  4. Plaintiff's Motion for a Temporary Restraining Order and Other Equitable Relief and Order to Show Cause re Preliminary Injunction (PDF)
    RegulatorTier 1Federal Trade Commission · 2025-08-25archived copy
  5. Air.ai - FTC case page (Legal Library: Cases and Proceedings), listing all filings and defendants
    RegulatorTier 1Federal Trade Commissionarchived copy
  6. Federal Trade Commission v. Air Ai Technologies Incorporated, 2:25-cv-03068 - CourtListener/RECAP docket
    Court recordTier 1Free Law Project (CourtListener)archived copy
  7. "Air AI and its Owners will be Banned from Marketing Business Opportunities to Settle FTC Charges the Company Misled Many Entrepreneurs and Small Businesses" - FTC press release, 24 March 2026 (2-0 Commission vote)
    RegulatorTier 1Federal Trade Commission · 2026-03-24archived copy

    FTC press release, 24 March 2026 - stipulated order for permanent injunction, monetary judgment and other relief, filed on unopposed motion by a 2-0 Commission vote: permanent ban on Caleb Maddix, Ryan O’Donnell and Thomas Lancer from selling or marketing any business opportunity, $18 million judgment largely suspended on inability to pay, $50,000 payable for consumer redress

  8. Stipulated Order for Permanent Injunction, Monetary Judgment, and Other Relief, No. 2:25-cv-03068-SMB - $18,000,000 judgment, $50,000 payable within 60 days, permanent business-opportunity ban on Caleb M. Maddix, Ryan P. O'Donnell and Thomas M. Lancer (PDF)
    RegulatorTier 1Federal Trade Commission · 2026-03-24archived copy
  9. Unopposed Motion to Enter Into Stipulated Order for Permanent Injunction, Monetary Judgment, and Other Relief (PDF)
    RegulatorTier 1Federal Trade Commission · 2026-03-24archived copy
  10. Air.ai case timeline entry - 24 March 2026
    RegulatorTier 1Federal Trade Commission · 2026-03-24archived copy
  11. Air Ai - BBB Business Profile, Phoenix AZ: F rating, not accredited, business start date 9 February 2023, 22 complaints with 21 unanswered, and the 11 July 2024 BBB advertising review of the "100,000 sales and customer service reps" and "70,000+ Businesses" claims
    Self-regulatoryTier 2Better Business Bureau (BBB Serving the Pacific Southwest) · 2024-07-31archived copy

    BBB business profile and complaint detail, Air Ai, Phoenix AZ - F rating, not accredited, business start date 9 February 2023, 22 complaints in three years with 21 unanswered, the July 2024 advertising review challenging the “100,000 sales and customer service reps” and “10-40 minute” claims with no business response as of 31 July 2024, the 11¢-to-33¢ per-minute increase reported by a $50,000 buyer on 14 June 2024, the $10,000-plus of third-party services, and financing remaining active after service cancellation

  12. Air Ai - BBB complaint detail (individual consumer complaints, including the 14 June 2024 per-minute price-increase complaint)
    Self-regulatoryTier 2Better Business Bureau (BBB Serving the Pacific Southwest)archived copy
  13. Retell AI pricing page - usage-based from $0.07/min with no platform fee
    Open-market comparisonTier 4Retell AI · 2026archived copy

    Independent latency benchmarking and pricing comparison, early 2026 - Air AI above 1,000 ms with multi-second dead air against roughly 620 ms measured for Retell and 700–1,500 ms measured for Bland; Retell at $0.07/min with no platform fee; Vapi at $0.05/min platform fee, roughly $0.15/min all-in, 60 free minutes, Agency plan $500/mo; Bland at $0.09–$0.15/min at pilot scale with subscriptions from $299

    Not established by this document: No published, methodologically identified latency benchmark measuring Air AI at >1,000 ms against Retell (~620 ms) and Bland (700-1,500 ms) could be located; the latency figures rest on the report's own testing, not a citable third-party study. Vapi's public pricing page shows Build and Scale tiers only - no $500/mo Agency plan is currently listed.

  14. Vapi pricing page - $0.05/min Vapi platform fee on the Build plan, 60+ minutes included
    Open-market comparisonTier 4Vapi · 2026archived copy
  15. Bland AI pricing page
    Open-market comparisonTier 4Bland AI · 2026archived copy
  16. Air AI reviews and product details - aggregate 1.5 / 5 across 113 reviews
    ReportingTier 3Tekpon · 2026archived copy

    Third-party software directory reviews - 1.5 out of 5 across 113 reviews, plus the FTC complaint’s snapshot of 78 one-star reviews against 3 positive; Trustpilot Canadian mirror for the nine-month refund paid only after a state attorney general complaint; representative verbatims on non-delivery and unresponsiveness

    Not established by this document: Trustpilot's main air.ai page returns HTTP 403 to automated fetches, so the live star distribution and the Canadian-mirror refund account were not read directly.

  17. air.ai customer reviews on Trustpilot
    ReportingTier 3Trustpilotarchived copy
  18. FCC Declaratory Ruling FCC 24-17, CG Docket No. 23-362, released 8 February 2024 - AI-generated voices are "artificial" voices under the TCPA (PDF)
    RegulatorTier 1Federal Communications Commission · 2024-02-08archived copy

    FCC Declaratory Ruling FCC 24-17, 8 February 2024 - AI-generated voices are “artificial” voices under the telephone-consumer statute, requiring prior express consent, prior express written consent for marketing calls, caller identification and a working opt-out, with per-call statutory damages; plus practitioner guidance on state all-party recording consent and emerging AI-disclosure duties

  19. "FCC Makes AI-Generated Voices in Robocalls Illegal" - FCC document page for the Declaratory Ruling
    RegulatorTier 1Federal Communications Commission · 2024-02-08archived copy
  20. air.ai as it stands in 2026 - the domain now serves "Air," the rebranded identity of defense-analytics firm Govini; no voice agent, license or pricing
    Company documentTier 1Air (formerly Govini) · 2026archived copy

    air.ai as observed 29 July 2026 - the domain serves “Air,” the rebranded identity of defense-analytics firm Govini, an unrelated enterprise platform with no voice agent, no license and no pricing; independent AI-agent trackers list the original Air AI product as defunct; company data aggregator listing $0 raised, roughly $3.5M ARR self-reported as of June 2025, $10.6M valuation and 32 employees

    Not established by this document: The AI-agent tracker listing the original Air AI product as defunct, and the company-data aggregator entry ($0 raised, ~$3.5M ARR, $10.6M valuation, 32 employees), are behind paywalled or login-gated aggregators and no free canonical URL was retrievable.

  21. Air pressroom - "Govini is now Air" rebrand announcements
    Company documentTier 1Air (formerly Govini) · 2026archived copy
  22. "Stop Air AI LLC (a robo call company) & Caleb Maddix from Defrauding Innocent People" - Change.org petition started 2 August 2024, 193 verified signatures, alleging a $50,000 license, an unhonoured 120-day guarantee and Instagram ads attributing campaigns to Apple and Tesla
    ReportingTier 3Change.org (petition by a self-identified Air AI licensee) · 2024-08-02archived copy

    Change.org petition started 2 August 2024 with 193 verified signatures, alleging a $50,000 license purchase, an unhonoured 120-day guarantee and Instagram advertising falsely attributing campaigns to Apple and Tesla; and a dedicated complaint site published by a self-described $100,000 licensee - both first-person accounts by interested parties, not adjudicated findings

    Not established by this document: The dedicated complaint site published by a self-described $100,000 licensee could not be located at a live URL in this pass.

  23. Petition update - "An Update on Air AI LLC"
    ReportingTier 3Change.orgarchived copy
Unable to verify

What we could not get

  • Whether the court has formally entered the stipulated final order as of 29 July 2026 - the FTC’s own 24 March 2026 announcement described it as filed on an unopposed motion and requiring judicial approval, and the accessible docket shows most recent activity in February 2026. What is confirmed and in force is the stipulated preliminary injunction entered 10 September 2025
  • The full text of the stipulated order - the filed PDF did not render substantive terms through the tools available, so the judgment amount, the suspension basis and the precise scope of the ban are taken from the FTC press release and contemporaneous trade coverage rather than from the order itself
  • Current corporate registration status of all six entities in Arizona and Delaware - active, dissolved or revoked could not be established in this pass
  • Exact current Trustpilot review count and star distribution - the main site returned an access error, so the figures used are the FTC complaint’s snapshot and secondary citations from a Canadian mirror and two review write-ups
  • The existence or content of any Air AI reseller advertising policy, income-claim policy or compliance guide - none could be located anywhere, which is recorded as an absence of evidence rather than proof that none existed
  • Prior business history of Ryan O’Donnell and Thomas Lancer, and any ventures by Caleb Maddix beyond the publishing imprint and the $997 YouTube-automation course; also what any of the three are doing commercially as of July 2026 and whether any successor voice-agent offer exists under new branding
  • Whether the original launch demo was fabricated as opposed to a scripted best case - the record supports the weaker claim only, that the controlled demo did not generalise to production, and the stronger allegation common in forum commentary could not be substantiated
  • Any state attorney general enforcement action or certified class action; chargeback rates or payment-processor action; typical licensee advertising spend; and any cohort-level data on time to first paying client - none of which exists publicly, because no earnings claim statement was ever produced

Not advice

This is independent analysis for decision-making, not legal, financial or investment advice. Allegations reported by third parties should be verified against primary court and regulatory records before any decision.

Who writes this

Researched by Claude. Reviewed by an editor.

Every report is researched and written by Claude, Anthropic’s AI assistant, from the company’s own plan documents, policies, terms and regulatory file - then reviewed before publication by Rob Fore, who checks the sources and the stage-label on every allegation.

  • Nine weighted dimensions, published with their weights
  • The editor checks the evidence and cannot change the number - the build rejects any page whose grade does not reconcile to its own arithmetic
  • Every affiliate position we hold is disclosed on the report it touches
  • No company has paid for a grade, and no report carries an affiliate link
Read the About page

Looking at something else?

Enter any company name or website. If a report exists it opens instantly; if not, we start one.

Try:
Common questions

Air AI - frequently asked

QWhat did the FTC allege about Air AI?
On 25 August 2025 the Federal Trade Commission filed a complaint in the District of Arizona against Air AI Technologies, Inc., five affiliated Arizona entities and three individual owners, case 2:25-cv-03068-SMB. Four counts were pleaded: false and unsubstantiated earnings claims and misrepresented refund guarantees under Section 5(a) of the FTC Act, the Telemarketing Sales Rule as to the telephone-sold Access Card, and the Business Opportunity Rule at 16 C.F.R. Part 437 as to the licensing offer. Stage-labeling matters here. A complaint is an allegation and nothing in it has been proven at trial. A stipulated preliminary injunction was entered on 10 September 2025 and is in force. On 24 March 2026 the FTC announced a stipulated order - one the defendants agreed to - permanently banning the owners from marketing any business opportunity, under an $18 million judgment largely suspended for inability to pay with $50,000 payable in redress. Settlements of this kind resolve without an admission of liability.
QWhat is the FTC Business Opportunity Rule and why did it apply?
The Rule, at 16 C.F.R. Part 437, covers an offer when three things are true together: the seller solicits someone to enter a new business, that person must make a required payment, and the seller represents that it will provide locations or accounts, supply outlets or customer leads, or buy back what the purchaser produces. When it applies, the seller must give the buyer a one-page disclosure document at least seven days before any signature or payment. And if the seller makes any earnings claim, it must provide a written earnings claim statement setting out the timeframe, the characteristics of the purchasers who achieved that result, and the number and percentage of all persons who bought the opportunity and achieved at least that level, with written substantiation available on request. Air AI promised a thousand pre-vetted leads and a full buy-back, satisfying the third element twice over, against a required payment of $9,800 to $100,000. The complaint pleads that neither document was ever furnished.
QHow much did an Air AI license cost?
The FTC complaint puts the licensing business opportunity at $9,800 to $100,000 in a tiered structure, with purported benefits increasing the more a consumer paid, and the separate Access Card at $15,000 to $30,000 sold over the telephone. The complaint alleges the price was adjusted to each consumer’s ability to pay. On top of the entry fee came per-minute charges - $0.11 advertised for outbound, $0.32 for inbound - with telephony billed separately again, on a twelve-month minimum with no free trial. One BBB complainant who paid $50,000 reports the per-minute rate rising from 11¢ to 33¢ after purchase, corroborated independently by a petition organiser. Purchases were frequently financed at up to 24.99% APR through the sellers’ own lending platform. Two independent reviewers put realistic year-one cost at $30,000 to $80,000 and above.
QIs Air AI still operating, and can you still buy it?
No. As observed on 29 July 2026 the air.ai domain serves “Air,” the rebranded identity of defense-analytics firm Govini - an entirely unrelated enterprise platform with federal security accreditations and a request-a-demo motion, with no voice agent, no license and no pricing anywhere on it. Independent AI-agent trackers list the original product as defunct, noting that the domain now belongs to an unrelated defense-technology company. The stipulated order announced in March 2026 forecloses the offer permanently as to its three owners, who are banned from selling or marketing any business opportunity. A handful of search-driven review pages still write about the product in the present tense and quote live-sounding prices; those are stale affiliate comparison pages rather than evidence of an active offer. Treat any Air AI license offered for sale today as the resale of a dead asset or an unrelated party trading on the name.
QWas Air AI a pyramid scheme?
No regulator alleged that, and the structure does not fit. There were no levels, no downline depth, no genealogy, no rank advancement and no commission for merely enrolling another licensee. Compensation was framed as margin on reselling software to real businesses plus a share of the call usage those end businesses generated - sales to end users, which is the right side of the line. There was also no token, no staking, no pooled fund and no promise of a passive return, so no securities question arises for the participant. That is why the securities dimension scores 8 in a file graded F. The failure was elsewhere: the complaint pleads that many licensees could not resell the license to anyone because the software did not work, which meant the only money reliably entering the system was the next buyer’s entry fee. A plan designed as sales-driven can still operate as inflow-driven.
Who wrote this report

Author, editor and publisher

C
Written by Claude AI
Reviewed by Rob Fore · Published by Listech Inc · July 29, 2026

This report was researched and written by Claude, Anthropic’s AI assistant, working from primary documents - Air AI’s own compensation plan, its policies and procedures, its terms of service, its income disclosure statement where one exists, and its regulatory and self-regulatory file. It was scored against nine weighted dimensions that are published in full, with their weights, on the methodology page.

Before publication it was reviewed by Rob Fore, who checks every source link, every figure against the document it came from, and every allegation against its stage-label - an investigation is not a finding, a warning letter is not an enforcement action, and a filed claim is not a verdict.

The editor does not set the grade. The published score is the weighted composite of the nine dimension scores, and the build refuses to emit a page where the two disagree by more than 0.06. A grade moves when the evidence moves it and not otherwise.

Rob Fore has marketed online since 1996, wrote Online MLM Marketing (2014), and is CEO of Listech Inc, the Nevada corporation that publishes this site. He holds affiliate positions in companies graded here - including LiveGood, which this site grades D, SendOutCards, which it grades C−, and the Home Business Academy, which it grades B−. Those positions are disclosed on the reports they touch, and changed nothing on this page.

About the author and our conflicts  ·  Contact the editor  ·  Corrections: corrections@opportunitygrade.com

Stay with it

Tell me if this grade changes

Air AI is graded F as of July 29, 2026. Grades move when the evidence moves - a new income disclosure, a regulatory action, a rewritten compensation plan. Leave your address and you will get one email if this one does.

One email when the grade moves, and nothing else. We will never use your address to promote an income opportunity of any kind, we do not sell, rent or share the list, and it is stored on our own infrastructure rather than with any company graded here. Unsubscribe removes everything.

Right of reply

Corrections

Every factual claim on this page is sourced, and the ones we could not stand up are named in the unable to verify list above. If something here is wrong, we want to know, and we would rather hear it from Air AI than from a reader.

Write to corrections@opportunitygrade.com. Point at the specific sentence and send the document that contradicts it - a plan document, a filing, an income disclosure, a policy page. We will check it against the primary source, correct the page if it is wrong, and say in the report that it was corrected and when. A grade moves if the evidence moves it.

This address reaches a person, not a form. We do not require a takedown demand, an NDA or a lawyer to accept a correction, and we do not remove a report because a company disputes its conclusion - only because the underlying facts turn out to be wrong.

Other published reports

Every report is written to stand alone. Graded on the same nine weighted dimensions and the same six legal tests. Twelve of 107, spread across the grade bands.

See all 107 published reports →