Dili raised a $15M Series A led by Khosla Ventures to automate compliance on US infrastructure projects, bringing the AI startup's total funding to $21.7M. The round follows a $6.7M seed and lands as data center construction and IRA-funded clean energy builds push a wave of federally regulated work through the system. Dili says its software is already running on about 700 projects, from manufacturing plants to data centers.
Allianz, Rebel Fund, Brick and Mortar Ventures' Darren Bechtel, and Y Combinator's Garry Tan joined the Series A. Dili went through Y Combinator's Summer 2023 batch. The company's pitch is narrow and specific: the tangle of overlapping wage, safety, and environmental rules that attach to any US construction project touching federal money, and the size of the fines when a contractor gets one of them wrong.
Co-founder and CEO Anand Chaturvedi points to Davis-Bacon rules, which let the Department of Labor set prevailing wages on federally funded builds, as the canonical example. A separate set of prevailing wage and apprenticeship rules apply to clean-energy projects funded under the IRA, layered on top of OSHA and EPA requirements that shift depending on the type of work. A payroll error on a single week's timesheet can compound into a serious enforcement problem.
Key facts
- 01Dili raised a $15M Series A led by Khosla Ventures, bringing total funding to $21.7M after a prior $6.7M seed round.
- 02The software is running on about 700 active projects, spanning manufacturing facilities and data centers.
- 03Roughly half of customers use Dili as in-house software; the other half outsource compliance to the company on a contractor model.
- 04Dili targets Davis-Bacon prevailing-wage rules, IRA clean-energy PWA rules, and OSHA and EPA requirements on federally funded builds.
- 05Backers include Allianz, Rebel Fund, Brick and Mortar Ventures' Darren Bechtel, and Y Combinator's Garry Tan.
The scale of the AI buildout is what makes the market urgent. Hyperscaler data centers, grid interconnects, and IRA-financed solar and battery plants are all being permitted and built at the same time, and most of them carry at least one federal compliance obligation. Traditionally that work has been handled by outside consultants sampling records after the fact. Dili's argument is that sampling is no longer good enough when a single missed classification can wipe out a project's margin.
The company's architecture is deliberately conservative. Contemporary AI models sit only in the data layer, translating unstructured documents — contracts, timesheets, vendor invoices, ERP exports — into structured records. From there a deterministic rules engine applies the actual compliance logic, which keeps the LLM out of the final judgment. A task that used to take a full day of professional-services work can be turned around in minutes.
Adoption is running on two tracks. Chaturvedi says roughly half of Dili's 700 projects use the product as an in-house software tool, with the customer's own team operating it, while the other half hand the entire compliance workflow to Dili on a contractor model. Dili is willing to run either arrangement, but Chaturvedi expects the mix to tilt toward software as buyers get comfortable with the model.
That trajectory is consistent with a broader pattern across enterprise AI: work that used to be sold as a professional service — legal review, tax prep, audit sampling, compliance monitoring — is being repackaged as software with a much thinner human layer on top. The customer keeps the workflow in-house and pays a fraction of what an outside firm would charge.
The risk with any AI-based compliance product is that a hallucinated field or a misread document ends up in a federal filing. Dili's deterministic back end is meant to address exactly that failure mode, but the industry has not yet seen a large enforcement action against an AI-assisted compliance workflow, and the legal question of who is liable when the model misreads a payroll record is untested. Buyers writing multimillion-dollar checks for federally funded builds will want that answer before they retire the outside consultants entirely.
For Khosla and the rest of the syndicate, Dili is a bet that the AI infrastructure cycle generates its own second-order software market. Every new data center, every new IRA-funded plant, and every new transmission upgrade is a compliance customer, and there is no incumbent that owns the workflow end-to-end. If Dili can hold reliability at 700 projects, the same playbook extends into permitting, environmental reporting, and vendor onboarding — a much larger surface than compliance alone.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
