The episode in short.
Timothy Daniel is the CEO and founder of Rimba, a Y Combinator company from the Spring 2025 batch that deploys AI agents for operations and compliance in heavy industry, starting with renewable fuels. He came to it from law and from compliance work in the biomass business in Indonesia, after a first startup in legal AI that he stopped after seven months. In the first episode of Startup Discovery he explains how Rimba got into YC on his fifth application, what the batch did and did not give it, and how a small team builds fast with coding agents while still passing enterprise review. It is a working founder's view of selling AI into industries that do not run on AI yet.
What we covered.
- 1:10 (watch from here on YouTube)
What does Rimba do?
Rimba deploys AI agents for operations and compliance work in heavy industry. Its first market is energy companies, and within energy the renewable fuels segment; the plan is to win that vertical before expanding. Timothy sees the physical economy (chemicals, raw materials, oil and gas) as the place where AI will change the most, because those companies run processes that agents can streamline and improve.
- 2:46 (watch from here on YouTube)
Is selling to a Fortune 500 company different from selling to a small producer?
The motion is similar, but the buyer changes. At companies under about 200 people Timothy talks to the C-suite or management; at a Fortune 500 company he usually talks to a manager or a director. Larger companies have more layers of checks before a decision, so Rimba plans for a longer sales cycle there.
- 4:36 (watch from here on YouTube)
How did compliance in the biomass business lead to Rimba?
Before Rimba, Timothy worked in compliance for a biomass business in Indonesia that collected agricultural waste in Southeast Asia and exported it to power plants in Japan and Europe as a replacement for coal. Every export depended on checks, certifications and annual audits. His point is that almost every physical product goes through similar testing and certification, which makes compliance a very large field, and renewable fuels was the segment closest to what he already knew.
- 7:36 (watch from here on YouTube)
Why did Timothy leave legal AI?
His first startup built AI for legal research, a natural fit for someone trained in law. After about seven months and many customer interviews in Southeast Asia and the United States, he found that the work did not excite him, traction was hard, and many companies were building the same products: research, contract extraction (document AI) and contract drafting. His advice is to give an idea real time, but to decide early when it is not yours.
- 11:04 (watch from here on YouTube)
Why apply to Y Combinator five times?
Rimba got in on Timothy's fifth application. He does not treat YC as a measure of success, since companies succeed and fail with and without it; the persistence is the point, because a founder's job is to keep going. Coming from outside software, he valued YC for teaching the basics of running a company and for its peer community. His interview advice is to answer briefly and directly, not to overthink, and to know your customers.
- 15:14 (watch from here on YouTube)
What does Y Combinator actually give you, and would he do it again?
Guidance, not labour: Y Combinator founders have to go and ask for the advice themselves. Timothy describes the batch as a pressure cooker of two and a half to three months, with a lot of teaching while you sell and build toward Demo Day. Batchmates arrive at very different stages and leave with very different results, so measuring yourself against the batch is a mistake. He would do it again: YC pushed Rimba out to find customers, and it signed an enterprise contract during the batch.
- 20:26 (watch from here on YouTube)
How do coding agents like Claude Code and Codex change how a startup builds?
Coding agents such as Claude Code and Codex did not change what Rimba builds, but they made the team much faster. Speed comes with a condition: enterprise customers inspect the product closely, so the team needs proper infrastructure, support for enterprise workflows and an understanding of the code it ships. Timothy's rule is no vibe coding: AI output gets reviewed, often more than once, and strong engineers are still needed, which is the argument of our post Coding is no longer the bottleneck. Trust is.
- 22:39 (watch from here on YouTube)
What does running an AI-native company look like?
Here Simon describes how Soba Labs runs as an AI-native company: internal systems are designed for AI agents to read and act on, not only for people, including a company knowledge base that agents can query (knowledge engineering), and scheduled background agents handle recurring daily work. People step in at the decisions that matter, such as conversations with customers (human in the loop). Timothy says Rimba runs few routines internally but sets up daily updates for some of its customers.
- 24:32 (watch from here on YouTube)
Do you need a co-founder?
Timothy applied to YC with a co-founder and likes having someone to think decisions through and move fast with. He thinks the answer differs by person and that a partnership should never be forced: no co-founder is better than the wrong one, and the wrong one is a poor fit rather than a bad person.
- 25:39 (watch from here on YouTube)
How do you find your ICP?
There is no formula, but the loop gets faster the faster you talk to customers, build, iterate and talk to more of them. Who you sell to sets the sales cycle, the contract size, and whether a buyer pays by card or sends a ten-page document for review, so early on a founder has to learn how each kind of company buys software, if it buys at all. Timothy calls this a core CEO job; once capital arrives, the founder hands off the less critical work and keeps what nobody else can do.
- 28:28 (watch from here on YouTube)
Does being in San Francisco matter?
Startups succeed and fail anywhere, but San Francisco helps with hiring, with access to capital, and with a culture where calculated risk is normal. Timothy does not think Rimba would have grown the same way from another country, because the people around you shape what you think is possible. He is equally clear that being elsewhere should not stop anyone from building a company.
- 31:36 (watch from here on YouTube)
Is San Francisco an AI bubble?
Whether or not it is an AI bubble, everyone there talks about AI because most local companies are software companies and most people are engineers, in the same way solar dominated an energy conference he attended in Abu Dhabi and finance dominates New York. Timothy thinks AI companies should not only sell to each other: Rimba's customers work in industries where daily work does not yet revolve around AI, and he sees the larger opportunity there. Simon's takeaway is that developer tools benefit from San Francisco, while healthcare or legal companies can be built elsewhere, Europe included.
- 36:59 (watch from here on YouTube)
What is Timothy's advice for young founders?
Find something you are excited to work on every day and would still work on when you are not, because building a company is hard. If you have not found it, talk to people, look for real problems and keep talking; it can take months or years, whether you are new to the industry or a long-time expert. Then commit the time, and go ahead if you believe in it, whatever others think.