
Your agent spends the user’s money. A wrong guess is a charge.Founded by a software developer who runs an AI app.
Agents that find, compare and buy for people turn a bad recommendation into a purchase, a refund request and sometimes a bank dispute, and stores are now going to court over agents acting on their sites. We insure agents, in-store assistants and apps sold to merchants.
Tell us the situation.
One licensed human replies the same business day — not five agents, not an auto-responder. If the policy you have is already the right one, we will tell you.
We use this to quote and service your insurance, and we do not sell it or pass it to lead networks. Privacy policy.
The person placing your tech E&O runs an app with an AI coach of his own.

Kevin Kelsey
Founder | Principal Agent
Boring’s founder, Kevin Kelsey, owns and runs TrainRec, a fitness app with a built-in AI coach, and has placed insurance for AI apps. Insuring his own taught him how an underwriter reads a model. Personal guidance from an AI reads more like the work of a professional than like software, technology E&O forms commonly exclude bodily injury, and not every form will add it back. So whether your app answers money, health, hiring or farming questions, the gap between what it says and what the policy covers gets read by someone who has had to close it for his own product.
Send us what you have. We’ll tell you if it’s right.
Your current policy paperwork, a renewal notice, or nothing at all. A licensed agent reads it and replies the same business day.
What AI shopping agent businesses actually need.
Technology E&O
Claims that your agent bought the wrong item, the wrong quantity or at the wrong price, or acted outside what the user approved.
Without it — The loss is the user’s money, which is financial loss, and your terms will have limited it. Whether the limit holds is the claim.
Cyber & tech E&O
Breach response, ransomware and the notification costs that follow. Technology errors and omissions sits on the same policy and covers the software or service itself failing a client, which is the half most tech companies find out about too late.
Without it — A shopping agent holds sessions, saved cards or tokens for user accounts, and a breach of those is the larger liability.
Media liability
Covers claims arising from what you publish (defamation, copyright and trademark infringement, invasion of privacy and misuse of a person’s likeness) on an annual policy for creators, podcasters, publishers and authors.
Without it — Product comparisons, generated descriptions and review summaries are published statements about other businesses’ goods, and misleading ones draw advertising and defamation claims.
Directors & officers
Protects the personal assets of the people running the company when they are sued over a management decision. It applies to private companies and nonprofit boards, not just public ones, which is a common and expensive misunderstanding.
Without it — Payments partners, platforms and investors scrutinize who is accountable when an agent moves money.
General liability
Third-party bodily injury and property damage. It is the line nearly every contract names by default.
Without it — The certificate that retail partners and platforms ask for, to be read for technology and professional services exclusions.
Ecommerce seller coverage
If you also sell products yourself rather than only run the agent, the merchant’s coverage is on its own page.
Without it — A seller is in the chain of supply for what it ships, which an agent that only buys on a user’s behalf is not.
What the regulators actually wrote down.
Amazon.com Services sued Perplexity AI in the Northern District of California (No. 3:25-cv-09514) in November 2025 under the federal Computer Fraud and Abuse Act and California’s Comprehensive Computer Data Access and Fraud Act, to stop the use of Perplexity’s agentic web browser tool on Amazon.com; the district court granted a preliminary injunction.
Ninth Circuit, Amazon.com Services, LLC v. Perplexity AI, Inc., No. 26-1444 (Aug. 4, 2026), opinion, caption and background; filing month per press reports of the complaint · checked 2026-11-01
On August 4, 2026 the Ninth Circuit vacated that preliminary injunction and remanded, holding that Amazon was unlikely to succeed in showing Perplexity “accessed” Amazon computers for purposes of the Computer Fraud and Abuse Act. The opinion decided a request for a preliminary injunction, not the final outcome of the case.
Ninth Circuit, Amazon.com Services, LLC v. Perplexity AI, Inc., No. 26-1444 (Aug. 4, 2026), opinion by Judge M. Smith · checked 2026-11-01
Regulation E defines an unauthorized electronic fund transfer as one initiated by a person other than the consumer without actual authority and from which the consumer receives no benefit, and the definition excludes a transfer initiated by a person who was furnished the access device by the consumer unless the consumer has notified the financial institution that transfers by that person are no longer authorized.
12 CFR 1005.2(m), Regulation E (Consumer Financial Protection Bureau) · checked 2026-11-01
The Regulation E official interpretation says that if a consumer furnishes an access device and grants authority to make transfers to a person, such as a family member or co-worker, who exceeds the authority given, the consumer is fully liable for the transfers unless the consumer has notified the financial institution that transfers by that person are no longer authorized.
12 CFR Part 1005, Supplement I, comment 2(m)-2 · checked 2026-11-01
The FTC’s rule on consumer reviews and testimonials, 16 CFR Part 465, bans fake reviews, including reviews that misrepresent who wrote them such as AI-generated fake reviews, and bans businesses from creating or selling them; it took effect on October 21, 2024.
Federal Trade Commission, final rule banning fake reviews and testimonials (August 14, 2024) and Consumer Reviews and Testimonials Rule Q&A · checked 2026-11-01
The claims we actually see.
Ordered by how often we see them, not by how dramatic they are. Each one names the coverage that answers it — and the policy people wrongly assume already does.
Most common
The agent buys the wrong thing, or buys when the user did not intend it
Covered by Technology E&O
The user asks for a refund, then disputes the charge with the bank. Terms that say the user approved the order help, but how the approval was captured is what a court or a bank will look at.
Less common, severe
A store sues over the agent acting on its site
Covered by Technology E&O is the usual route, but the wording on claims from outside your customers matters
A platform’s claim is not a customer’s loss, and tech E&O may treat it differently. The Amazon case was brought under computer access laws, and policies differ on whether that is covered at all.
Common
Saved cards, tokens or account sessions are exposed
Covered by Cyber liability
Payment data brings card-brand rules and fines, and some cyber forms exclude or limit them. A user whose account is misused through your agent will also look at your contract.
Common
Product claims or review summaries mislead
Covered by Media liability
Summaries of other businesses’ reviews can misstate them, and the FTC’s fake review rule reaches reviews that misrepresent who wrote them, including AI-generated ones. Tech E&O may not include the publishing claim.
What our AI shopping agent clients ask us.
Why is an AI shopping agent different from other AI apps?
Most AI apps give an answer. A shopping agent acts, and the action costs money. A wrong recommendation is a complaint, but a wrong purchase is a charge, a return, a refund dispute and sometimes a chargeback with a bank. That moves the claim from advice to a financial loss with a payments party attached, and it is why the contract language about what the user approved matters more than for a chatbot.
What happened in Amazon’s lawsuit against Perplexity?
Amazon sued in the Northern District of California in November 2025, alleging that Perplexity’s agentic browser tool shopping on Amazon.com for users violated the federal Computer Fraud and Abuse Act and California’s computer access law. The district court granted a preliminary injunction. On August 4, 2026 the Ninth Circuit vacated it and sent the case back, holding that Amazon was unlikely to succeed in showing Perplexity accessed Amazon’s computers under the federal statute. It was a ruling on an injunction rather than the end of the case. For an agent builder the practical point is that a platform may sue, and that the legal question is still open.
Who pays if the agent buys something the user did not want?
It depends on what the user approved and what your terms say, and then on how the payment was made. Regulation E, which covers debit card and bank transfers, defines an unauthorized transfer as one initiated by someone without actual authority, and it excludes a transfer by a person to whom the consumer furnished the access device unless the consumer told the bank that person’s authority had ended. Whether an AI agent counts as such a person has not been decided in any source we verified, so we do not say. A lawyer should read how your flow captures approval.
Should the agent ask the user to confirm every purchase?
Underwriters like it and users sometimes do not. A confirmation step makes the user the decision maker, which helps the defense, and spending limits, an item list the user can see and an easy way to cancel do the same. A fully autonomous purchase is the version underwriters read as higher risk. Describe your actual flow, not the marketing version.
What will the underwriter ask a shopping agent?
Whether the agent buys on its own or only after the user confirms each order, and what spending limits exist. Where payment credentials are stored and who processes the payment. Which stores it acts on and whether those stores permit it in their terms. Which model powers it and whether a person reviews unusual orders. How it handles returns, refunds and disputes. Your terms: disclaimers, arbitration and any liability cap. Revenue, user count, countries served, whether you are a merchant of record, prior claims and chargeback rate.
Does the policy cover an AI that picks the wrong product?
A wrong pick is a failure of your service and a tech E&O matter, but the wording on AI output matters. Some 2026 forms exclude or restrict claims that arise from generative AI, and general liability forms may carry a generative AI exclusion. Ask whether AI output is covered, in plain words, and put the answer in the policy rather than an email.
Does the FTC fake review rule affect a shopping agent?
It can. The FTC’s rule, in effect since October 21, 2024, bans fake reviews, including reviews that misrepresent who wrote them such as AI-generated ones, and bans creating or selling them. An agent that summarizes reviews has to be careful not to present synthetic or paid ones as genuine, and an app sold to merchants that writes reviews for them is in the rule’s direct line. Regulatory defense for that sits with the media or tech form.
We make an app that merchants install on their store. Is that this page?
Yes for the app, since the merchant is your customer and relies on your code to take orders, set prices or write product copy. A bug that mispriced a catalog or dropped orders for a day is a merchant’s lost sales, and large merchants will ask for insurance in the contract. The store’s own insurance, as a seller, is a separate matter.
We built the agent with AI coding tools. Is that a problem?
Underwriters ask about it because an agent that spends money should be tested before it ships. Describe how the code is reviewed and tested, particularly around payments. The incidents and the policy questions for apps built that way are in our guide.
Do we need cyber coverage if a payments processor holds the cards?
Probably, because tokens, sessions and order history are still yours, and an attacker who takes over an agent can place orders. The processor reduces the card data you hold but not your liability for what an agent does with a logged-in session. Cyber should be read for payment card fines and regulatory defense.
Could a shopping agent cause bodily injury?
Occasionally, through the goods. An agent that recommends and buys a child’s product, a supplement or a tool that later hurts someone is a thin link in a claim that mostly runs against the seller and maker. Technology E&O forms commonly exclude bodily injury, and contingent bodily injury, an injury someone says came from relying on your output, is available on some forms. It matters most if you recommend safety-related goods.
Do we need coverage if we sell in other countries?
Check where claims can be brought. Many technology E&O policies only defend claims in the US or Canada, and a buyer abroad can sue at home and bring a payments dispute through a local bank.
Which NAICS codes does this page cover?
The 2022 NAICS codes most relevant are 513210 Software Publishers, 541511 Custom Computer Programming Services, 541519 Other Computer Related Services, 519290 Web Search Portals and All Other Information Services and 518210 Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services. A business that sells products itself is classified as a retailer, which is a different page.
Where does this fit among AI app insurance in other industries?
A shopping agent’s distinct claim is spending, which is why this page is separate from advice-giving AI apps. Each industry page names its own gap.
Is there a flagship example for an AI app that gives advice?
Our founder, Kevin Kelsey, owns TrainRec, a fitness app with a built-in AI coach by Manifest Post, and has placed AI apps. The coach is the advice version of the exposure, and the shopping agent is the spending version.
Guides on AI shopping agent.
- Insurance for an App Launching WorldwideAn app store launch puts your app in front of users in dozens of countries on the first day. Your tech E&O may only defend claims filed in the US or Canada, and your cyber policy has to handle European privacy rules, where health and fitness readings are a special category of data.
- Vibe-Coded App InsuranceA vibe-coded app carries the same liability as any other, and the documented failures are security ones: open databases, exposed keys and an agent that deleted data. Here is which policy answers which failure, what the AI tool providers say about who is responsible, and what to check before you buy.
Commercial insurance is 96% of what we do — it is not a department here, it is the whole business. Not ready to talk? The guides answer the questions this page raises in more depth. Already insured with us and need a certificate or a policy change? Ask the service team rather than starting a quote — it is faster and it goes to the people whose job it is. We also write home and auto, which is usually cheaper alongside the business policy than apart from it.
Handed a lease, a vendor packet or a client contract with insurance requirements? Paste it into the COI Requirements Decoder to see what each one means for your business, or read what the requirements mean.