TLDR Founders 2026-10-28

Kiro turns your prompts into specs, designs, and tested code — not just prototypes. Spec-driven development means your AI-generated code actually makes it to production.

Early-stage to Series A startups: get up to 1 year of Kiro Pro+ free.

Terms and Conditions apply, linked HERE.

Requests for Startups is a place for Y Combinator to share the ideas it would like to see founders tackle. Y Combinator is excited about startups building the hardware, infrastructure, financial systems, and specialized software needed to bring AI into the real economy. The ideas listed in Requests for Startups represent just a fraction of what Y Combinator funds. Founders working in other areas are still encouraged to apply to the fund.

The solo founder is one of the archetypes we'll likely soon see a lot more of. AI enables solo founders to do 100x more, allowing them to ship a v1 when it wasn't possible before. People who can create but can't code can now build. Experts can now create their own software to solve niche problems. Kids and teenagers, who already build products on platforms like Minecraft and Roblox, can now build way more with AI. There will likely be a high failure rate, but some will inevitably work out.

AI made it easy to ship every week and just as easy to skip asking customers whether they need what you shipped. This playbook (from a founder who lost about 18 months to two wrong bets at KISSmetrics) walks through how to run early access so you learn from customers as fast as you ship. The core habit is pretty simple, which is to ask people about the last time they did the job and then watch them use your product. People coming back to use it again tells you far more than any interview, and each of the six plays comes with a short template to fill in.

Investors walk into early meetings a bit skeptical, mostly looking for a reason to pass on you. A former YC partner explains how to build your pitch around a few claims that make passing feel a little riskier. Investors still don't know how to judge AI-era companies (and they've all watched a huge one slip away), so one tactic is to pick three or four claims where believing any one of them would make an investor nervous to say no. For example, you can test those claims in investor meetings six to twelve months before you raise and keep the ones people still repeat back to you a quarter later.

Buying established service businesses gives AI builders access to what's hardest to replicate: customer trust, distribution, and employees' hard-earned expertise. Ownership also lets them redesign entire workflows and capture the resulting profit gains, which would otherwise largely accrue to their customers. The challenge is turning hours saved into higher revenue and better service, rather than treating adoption alone as success.

Your business moves quickly. Your money should, too. Stripe Treasury brings payments and financial tools together, helping you pay vendors, send payouts, issue cards, and manage cash flow with less complexity. Build more efficient financial operations on the same platform your business already uses for payments as you continue growing.

See how it works

Supabase Compute allows developers to run agents, background jobs, and APIs right next to their data, in the same project. It is now available in private alpha for free. Each service gets a full Linux environment. A link to the waitlist is available in the article.

Replit's Cloud Agents enable non-engineers to contribute code directly and safely to a company's core codebase. Subject matter experts can describe changes in plain language, with Cloud Agents handling code adjustments and testing, reducing engineering workload. These agents create shared environments, allowing easy setup and immediate contribution from anyone in the team.

The `stripe pay` plugin allows developers to find companies using their Stripe profile handle and send funds for free. It is available now in the Dashboard and CLI. Senders will need a funded financial account, and the recipient needs a financial account linked to their Stripe profile. A dry run command is available for validating flows without creating or sending transfers.

Putting an AI agent on a sales team is mostly a people problem, since reps tend to assume it's coming for their jobs. monday.com's former VP of AI GTM walks through how they got a voice agent making the first call on every new lead without a fight. The trick was to start small, so the agent only got more leads once it was doing as well as the reps on the ones it already had. New leads used to wait about 25 hours for a call back, and now it's under two minutes (the team still spends 20 minutes every morning going over call scores).

Good interfaces allow independent systems to work together without knowing everything about one another. Collaboration problems between teams are often interface problems rather than relationship problems. A good interface makes it obvious who owns what, what they are working on, and how new requests are evaluated. When those things aren't clear, even well-intentioned people end up wasting energy on disagreements and reverse-engineering decisions.

What if your software platform could offer customer websites as an additional product, without building the website infrastructure yourself?

Explore the business case for adding websites to your software offering.

Explore the opportunity →

Let AI handle the analysis and the writing, but keep the high-stakes decisions in ordinary code.

Test AI on your own company's work, since public benchmarks skip most jobs and every competitor reads them anyway.

Identify what your customer is paying for, and hold every improvement against that feature.

Keep ads that still work running in their own campaign instead of turning them off when new ones launch.