Letterbook Wants To Be The Last Support Tool A Founder Ever Sets Up
Two founders, no support agents, and an AI that reads your database before it writes a reply. Inside the YC-backed startup betting that the helpdesk was overdue for a rebuild.
Most founders treat customer support as the tax they pay for having customers. Dawson Chen looked at the same overflowing inbox and saw a company. Letterbook, the startup he co-founded with Ethan Hou in 2023, starts from a plain observation: the software teams use to answer their customers was built for support departments that most young companies do not have and cannot afford.
The pitch is unusually concrete. Connect your support inbox, plug in your database and your Stripe account, and Letterbook's AI reads the actual customer record before it writes a word. It triages the ticket, drafts a resolution, and hands it back to a human to review and send. The setup, the founders claim, takes about fifteen minutes - not the multi-week onboarding that enterprise helpdesks are known for.
01 / WHAT IT DOESAn AI that answers with context, not scripts
The thing that separates a good support reply from a useless one is usually information. Did this customer pay? When did they sign up? What plan are they on? Legacy tools make an agent go dig for that. Letterbook feeds it to the AI directly by connecting to the systems where the answers already live - Stripe for billing, and databases like Postgres, Supabase, MongoDB, and Google Play for everything else.
Support arrives from more than email. Letterbook handles web forms, Discord, API events, and in-app mobile feedback, then routes them into one inbox. A hosted help center gives customers self-service docs, and an analytics layer tracks the numbers support teams actually argue about: first response time and resolution rate.
02 / THE CONTRARIAN BETKeep the human in the loop
Plenty of startups are racing to build a support bot that needs no supervision. Letterbook took the quieter path: the AI prepares the resolution, and a person signs off before the customer ever sees it. It is a design choice about trust. For a founder whose reputation rides on every reply, "review and send" is an easier thing to adopt than "let it rip." The autonomy can come later; the confidence has to come first.
That framing also reveals who Letterbook is really for. This is not software aimed at a 200-seat contact center. It is aimed at the founder answering tickets at eleven at night, or the two-person team that would rather not hire a support manager just to keep up. Letterbook says it has already replaced a stack of incumbents for fast-growing companies inside Y Combinator.
03 / HOW IT'S DIFFERENTRebuilt around AI, not bolted onto it
Nearly every established helpdesk now advertises AI. The difference Letterbook draws is architectural. Adding an AI feature to a ticketing system built in 2011 is one thing; designing the whole product so the AI has live data and does the first pass is another. The former is a feature. The latter is the reason the company exists. Whether that distinction holds up under the weight of well-funded incumbents is the open question every AI-native startup faces - but it is a clear, legible bet.
04 / THE MODELPriced to be a startup's first helpdesk
Letterbook sells the way it wants to be adopted: cheaply, and without a meeting. The entry tier is low enough for a two-person company to expense on instinct, and the plans scale by seats and ticket volume as the customer grows. That is the strategy hiding inside the price sheet - be the first support tool a company ever buys, then grow with it.
Starter covers 300 tickets a month, Growth lifts that to 1,000 with priority support, and Enterprise opens up custom seats, 3,000-plus tickets, and dedicated account management. Every plan comes with a seven-day free trial and no credit card required.
05 / THE FOUNDERSFrom an assistant named Martin to a helpdesk
Chen and Hou did not arrive at customer support by accident. Before Letterbook, they built Martin, an AI personal assistant that reached 100,000 users. Chen studied computer science and economics at Yale before going full-time on startups. The pattern across both products is the same instinct: take a job people find tedious and let an AI do the first draft. The second time around, they pointed it at a problem companies actually pay to make disappear.
06 / WHERE IT FITSA crowded market, narrowed on purpose
AI customer support is a busy category, full of money and incumbents adding assistants of their own. Letterbook's answer to the crowd is to narrow the target rather than widen it: early-stage and fast-growing software companies that care about speed, control, and staying lean. It is a smaller pond, but a specific one - and being the obvious choice for a clearly defined customer tends to beat being an option for everyone.
The company remains tiny, hiring its first employees - a founding AI engineer, a founding product engineer, a founding growth marketer - out of San Francisco. Backed by Y Combinator's Summer 2023 batch and a reported $2 million, Letterbook is still early. But the shape of the bet is easy to read: rebuild the helpdesk for the people who never wanted to run one.