Dili raises $21.7M to bring AI compliance to the infrastructure boom
TechCrunch Russell Brandom
Dili just raised $15M more to use AI for checking construction compliance paperwork. It's proof AI's real money isn't just chatbots—it's messy federal regulations nobody wants to read.
Based on reporting by TechCrunch, Russell Brandom — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Every data center and power plant boom needs someone to check the paperwork, and that paperwork is a nightmare. Dili, a startup that emerged from Y Combinator's Summer 2023 batch, just landed $15 million in Series A funding led by Khosla Ventures, with Allianz, Rebel Fund, Brick and Mortar Ventures' Darren Bechtel, and YC's Garry Tan also chipping in. Add that to a $6.7 million seed round, and Dili has now raised $21.7 million to solve a problem most people have never heard of: making sure infrastructure projects actually follow the rules attached to their federal funding.
The rules in question are genuinely tangled. Davis-Bacon requirements set prevailing wages on certain federal projects. A separate PWA framework governs wage and apprenticeship standards for clean energy work funded under the IRA. Layer in OSHA and EPA requirements depending on what's being built, and you get a compliance maze where a single missed detail can trigger fines in the millions, according to co-founder and CEO Anand Chaturvedi.
What makes Dili's pitch interesting isn't just the AI angle — it's where the AI stops. Chaturvedi is careful to explain that large language models only touch the front end of the pipeline, converting messy unstructured documents like invoices, payroll records, and vendor files into structured data. Once that's done, a deterministic system takes over, applying the actual compliance logic. No LLM guesswork gets near the part where fines are on the line. That's a deliberate design choice, and it's the kind of architecture decision that matters a lot more in construction compliance than it does in, say, a customer service chatbot.
The company says it's already live on roughly 700 projects, spanning manufacturing plants and data centers alike. Split down the middle, about half those clients run Dili as internal software, while the other half hand the whole compliance function over as an outsourced service. Chaturvedi expects that balance to tip toward the software model over time, betting that AI will keep chewing through professional-services work the way it has elsewhere. Whether that prediction holds probably depends less on the technology and more on how fast compliance teams get comfortable letting a deterministic engine make the call.
Dili's timing lines up neatly with the broader infrastructure spending wave tied to AI itself — new data centers, new power projects, new federal dollars, and therefore new compliance headaches. It's a reminder that the AI boom isn't just generating demand for chips and electricity. It's generating demand for someone to keep the paperwork straight.
My take — AI-written commentary, not fact-checked reporting
I like this one precisely because it's boring in the right way — no chatbot theater, just deterministic rule-checking with LLMs kept on a short leash where they can't hallucinate a fine into existence. That's the sober, unglamorous use of AI that actually deserves the hype budget, and it's a pattern I wish more infrastructure-adjacent startups would copy instead of bolting a chat interface onto everything.
Read more about this at: TechCrunch
