Weekend AI Build Exit Checklist: Side Project, Product, or Team

The exit is often the team, not the app

You built an AI workflow agent over a weekend, and now a buyer could pay for the team and IP, not the app. Adobe confirmed on September 2 that it was acquiring Rilo.

Rilo, an India-based AI workflow automation startup, used AI agents to carry out several-step marketing and go-to-market tasks from simple English prompts. It was founded in September 2025 by IIT Bombay co-founders Dhruv Jaglan and Georgi Boby.

Rilo raised $1 million from a group that included Peak XV, DeVC, and Day Zero Ventures at a $10 million valuation. Adobe planned to take on Rilo's six-member team and integrate some of its intellectual property in a deal structured as a technology licensing arrangement.

After the acquisition, Rilo would shut down and stop being available to its customers. Rilo had more than 10,000 users who had built with the platform within months of launch.

The deal was Adobe's second acquisition of an Indian startup after it bought Rephrase.ai in 2023. A small AI workflow product can be valuable even when the app disappears.

For a solo founder exit, the question is whether the buyer is paying for a product or for your ability to keep building. If the product can be turned off, the value shifts to the team, the documentation, and the parts of the system that survive the shutdown.

The IP is the controls, not the model

Demos that call a model are easy to rebuild. Workflows that survive contact with real data are harder. The gap is often the scaffolding around the model.

Pilot agents become production-ready when the surrounding controls and governance are in place, not when model parameters change. Useful AI agent examples require an inspectable operating contract that specifies triggers, approved sources, artifacts, action boundaries, verification, failure states, escalation points, and autonomy limits.

Model authority in an agent workflow is set by the surrounding operating contract rather than by the model's text generation alone. Weekend builds with those controls give buyers something to inspect. Notebooks and prompts make the asset thinner.

Write the contract in plain language. Put the trigger, the allowed sources, the output, the stop conditions, and the human escalation in one page. A stranger should be able to read it, or the IP is harder to value.

Models can draft, summarize, or call tools. The contract decides what it may do, what it must ask for, and where it must stop.

Five questions decide the acqui-hire math

Run this list before you quit, incorporate, or start talking to buyers. Each question should be answerable in a minute.

An acqui-hire is often framed as a job with a bonus. The math changes when the buyer also wants the code, the data pipeline, or the operating contract. If the buyer can keep those things, the payout is not just compensation.

• What can the buyer keep? List the code, prompts, data schemas, integrations, docs, and operating contract. Mostly-demo answers mean weak IP.
• What do users lose? Name the workflows, files, saved runs, and accounts that would stop working. Messy shutdowns mean the product may need a wind-down plan.
• Can the product shut down cleanly? Check whether you can export data, stop billing, archive logs, and tell users what happens. Clean shutdowns protect the team and the IP.
• Is the payout for time or product? Salary-bridge payouts are acqui-hires. Payouts tied to code, IP, or a continuing product are product exits.
• Is full-time justified? Compare the payout to the cost of your evenings, the risk of a shutdown, and the chance the IP keeps working after you leave. No answer means keep it a side project.

Small is the default unless the IP is worth full-time

Most side projects stay side projects, and that is fine. Small AI workflow tools can still be useful without becoming a company.

Full-time is a cost. It removes the spare time that makes a side project survivable, so compare a one-time payout to the value of your time for the next year.

Go full-time only when the operating contract, the team, and the IP are strong enough that a buyer would pay for them. Stay small if the main asset is your attention.

Write the shutdown plan before accepting an offer. Keep the build inspectable, with scope narrow enough to finish before the next deadline.