9 Out of 10 People Never Claim the Money They Are Owed. These Agents Do.

ISSUE 03 OF THE FOUNDING FIVE · 8 MIN

Europe's airlines are sitting on about €3 billion of flight compensation nobody claimed; the median class action settlement gets claims from only 9% of the people it covers, and the firms that will chase any of it for you take 35 to 60% of what lands. Four builders put an agent on the paperwork instead, and three of them charge a subscription because the harder model pays better.

Every year money goes unclaimed because claiming it is boring. Flight compensation nobody files, class action settlements nobody signs up for, price drops after the return window opens, and subscriptions that renew because canceling means a phone call. The money is real, and the paperwork is the only thing standing between the owner and the cash.

An agent does not find paperwork boring. Give it the inbox, let it watch for the trigger, and let it file, and the pricing writes itself: take a cut of what lands, charge nothing up front, and the customer never has a reason to say no. Four builders run that model today, and the interesting part is that three of them chose not to.

The problem these builds attack is not that the money is hidden. It is that the cost of collecting it has been priced like a lawsuit. AirHelp, the largest flight compensation firm in Europe, charges a 35% service fee on whatever the airline pays and adds a 15% legal action fee if the case goes to court, so a passenger owed €600 under EU261 keeps as little as €300. Flightright charges 20 to 30% plus a 14% surcharge when a lawyer is involved. Those fees exist because a human has to read the booking, check the eligibility, file the form, and chase the airline for months, and a human's time is expensive enough that the firm needs a third of the money to cover it. As a result, fewer than half of the compensation owed over the last fifteen years has ever been claimed, and airlines are estimated to be holding about €3 billion that belongs to passengers who never asked.

The same shape repeats everywhere money is owed to people who have to file for it. The Federal Trade Commission studied 149 class action settlements and found the median claims rate was 9%, and 4% when weighted by the number of people notified, which means nine out of ten people entitled to a settlement never fill in the form, while the lawyers who ran the case take 20 to 33% of the fund with a median near 24%. Bill negotiation firms charge like collections agencies: Rocket Money takes 35 to 60% of the first year's savings on a bill it negotiates down, and Billshark takes 40% of the savings for up to two years. Public adjusters, the licensed professionals who chase insurance claims, are capped by state law at 10 to 20% of the claim, and even that cap exists because the fee was worth regulating. Behind it all sits a stock of about $70 billion in unclaimed property held by US states, of which $4.49 billion found its way back to owners in fiscal 2024.

That is the terrain: a 35 to 60% toll on recovered money, charged because a human has to do the chasing, and a claims rate under 10% because most people would rather forgo the money than do the chasing themselves. An agent that reads the inbox, recognizes the trigger, and files the form changes both numbers at once. It drops the cost of chasing to near zero, which lets Airfairness charge 30% where AirHelp charges 35 plus 15, and it raises the claims rate because the customer no longer has to do anything except say yes once. The new demand is not the 9% who already file. It is the 91% who don’t, and no incumbent can reach them at a human's cost per claim.

The incumbents know it. AirHelp now sells a membership for $32.99 to $249.99 a year that lets a member keep 100% of the compensation, which is a contingency firm quietly moving to a subscription before a software agent forces the price of the contingency to zero. Watch that move, because three of the four builders below have already made the same choice.

LISTED · FLIGHT CLAIMS

Scans your inbox for flight confirmations, checks EU261 and Canadian APPR eligibility in under 2 minutes, files the claim, follows up, and escalates. No win, no fee, and the fee is 30% of what the airline pays, up to $1,000 per passenger. The tool started life as FlightorFight, built in one weekend by a former Spotify innovation lead, and Airfairness bought it over LinkedIn in June 2025. Founders John Marzo and David Linardi, Toronto.

Reach them · [email protected] · @airfairness · airfairness.com

LISTED · BILL NEGOTIATION

An agent that places real phone calls and sends real emails to negotiate bills, cancel subscriptions, and chase refunds until they resolve. The site reports 150,000+ users, a 93% negotiation success rate, and $3M+ saved. Credit-based subscriptions and annual plans drop the success fee. Calls are placed with the user's approval. Founder Stanley Wei, San Francisco, about 10 people, bootstrapped.

Reach them · @PineAIAssistant · 19pine.ai

LISTED · SUBSCRIPTIONS AND CASHBACK

A browser extension that finds recurring charges across your accounts, then sends its agents into merchant portals, email, and chat to cancel them for you, and activates card issuer cashback offers on its own. The site reports 500,000 users saving an average of $1,200 a year. Freemium, with a Premium tier that waives the $2.99 cashback redemption fee.

Reach them · [email protected] · @joinkudos · joinkudos.com

LISTED · CLASS ACTION CLAIMS

An iPhone app. Tell it which brands you buy, and it matches you to open class action settlements and submits the claim forms. $4.99 a week, $49.99 a year, or $29.99 for life, with 1.6K App Store ratings at 4.6. The settlement money goes straight from the administrator to you, and the app explicitly states it is not a law firm. Publisher Heyoka Bilisim, Turkey.

Reach them · [email protected] · App Store

Four builders, two ways of getting paid, and the choice between them is the product decision. Airfairness takes 30% of the recovery, which means the airline's payment has to route through Airfairness or Airfairness has to be able to see it land, and either way the company is holding or watching somebody else's money for a while. Pine, Kudos, and claimHood charge a subscription and never touch the recovery at all: the customer pays a flat fee out of pocket and the refund, the savings, or the settlement lands in the customer's own account. If you are going to build here, the vocabulary of that choice is worth learning, because it will come up in the first investor meeting.

A contingency fee is revenue you earn only when the money lands, and it is the better business on paper because it scales with the value you create, and the customer never has to decide whether you are worth $12.99 a month. Its cost is timing. Revenue recognition, the accounting rule for when you are allowed to count a sale, says you cannot book the fee until the airline actually pays, and airlines pay in months, so a contingency business can be profitable on every claim and still run out of cash, because the cash arrives long after the work was done. That gap is working capital, the money you need to have on hand to keep operating while you wait to be paid, and it is the reason a contingency business raises money earlier than a subscription business does.

A subscription is the opposite trade: worse economics per customer and far better cash. The fee arrives before the work, so the customer funds your working capital instead of an investor, and the number that matters becomes customer acquisition cost, what you spend to win one subscriber, set against lifetime value, what that subscriber pays before leaving. claimHood's $4.99 a week targets a customer who will churn in a month and still leave $20 behind, while its $29.99 lifetime price bets that acquisition cost is low enough that one payment covers it. Kudos gives the product away and earns on the $2.99 redemption fee and the Premium tier, a freemium model where free users are the marketing budget.

Read the four together, and the pattern is that the harder model, contingency, sits in the one niche where the payout is easiest to confirm, because an airline pays a fixed statutory amount to a known account and there is no argument about whether the money landed. Everywhere the recovery is harder to see, the builders chose the subscription. That is not timidity. It is the correct reading of the one rule of contingency pricing: if you cannot see the money land, you cannot bill for it.

Pick one claim type nobody has automated yet and go contingency on it, as long as the payout is easy to confirm. Hotel rate drops after booking, parking ticket appeals, security deposits that never came back, and warranty claims on things bought eleven months ago all follow the same pattern: watch the inbox for the trigger, file the form, chase, and take 20 to 30% of what lands, which undercuts every incumbent in this issue and still leaves you the best margin in consumer software.

Start with the claim where you can see the money arrive, because the moment you cannot, you are back to $12.99 a month and competing with Kudos. If you price the service at the incumbent's take rate, you win on cost; if you price it at a third of the recovery, you win on margin, and margin is the better business to be in.

browser-use/browser-use · Python, MIT, about 112k stars and more than 10,000 commits. It is the framework most builders in this category actually use to send an agent into a claim form, a merchant portal, or an airline's compensation page, and there is no dedicated open source claims bot worth pointing you to instead, which tells you how early this market is.

Why it is worth an hour: look first at the persistent session and saved browser profile support, because a claim is a multi-step form behind a login, and an agent that loses its session halfway through files nothing. Then look at the custom action and tool architecture, which is where you wire in the claim specific logic, parsing a booking confirmation, filling an EU261 form, or pulling a settlement ID from a notice email, without writing raw Playwright. Before you point it at a real inbox, add a hard list of the sites it may log in to and the one account a payout may be sent to, because the framework will go wherever the page tells it.

Every build on this page holds something more valuable than a card number. An inbox, the login for the merchant portal, and the authority to say "I am the customer" to an airline, a settlement administrator, or a subscription desk. An agent that can file a claim in your name can also redirect where the claim pays out, and the airline has no way to tell the difference, because the request came from the same inbox with the same booking reference it always has.

Before you hand an agent your inbox and your identity, write down the one account a payout may land in, and put that line somewhere the agent reads but cannot rewrite. If you want a machine to confirm that the agent filing the claim was allowed to before the money moves, FLINT Scout does that for a cent, and that is the only pitch you will get from us.

NEXT ISSUE

Agents that check out: the API that lets any agent buy from any store without the store's permission, the travel agents that book, pay, and rebook without asking twice, and the 59% operating margin the card networks earn on every one of those purchases.