Dili Wins $15M Backing for AI Regulatory Compliance Tools

Federal money for infrastructure arrives with a documentation bill attached. A contractor building a data centre or a clean energy facility with federal support must evidence wage rates, apprenticeship ratios, safety records and environmental conditions across its own payroll and every subcontractor beneath it. Dili, a New York company selling software that automates that evidence-gathering, has raised $15 million in Series A funding led by Khosla Ventures.

The round closed on 30 July and follows a $6.7 million seed, taking total capital raised to $21.7 million. Allianz, Rebel Fund, Darren Bechtel of Brick and Mortar Ventures and Garry Tan of Y Combinator also participated.

Dili said the proceeds will fund hiring across engineering, product and go-to-market.

Dili was founded in 2023 by Anand Chaturvedi, chief executive, and Brian Fernandez, chief technology officer. It went through Y Combinator’s Summer 2023 cohort.

The regulatory surface it targets is narrow and unusually dense. Davis-Bacon rules allow the Department of Labor to set prevailing wages on federally funded construction. A separate set of prevailing wage and apprenticeship requirements, known as PWA rules, applies to clean energy projects claiming tax credits under the Inflation Reduction Act. OSHA and EPA obligations layer on top depending on the nature of the work. Each regime demands a different evidentiary record from the same underlying payroll and vendor data.

The penalty structure is what makes the category commercially interesting. Projects that fail PWA requirements forfeit a 5x bonus multiplier on their IRA tax credits, reverting to one-fifth of the headline value. A compliance gap becomes a project financing problem.

Demand for the product is tied directly to the volume of federally funded construction. The compliance obligations Dili automates exist because specific statutes attach them to specific funding programmes. Changes to IRA credit eligibility, or a shift in Department of Labor enforcement posture, would alter the size of the market the software serves.

“Non-compliance can result in millions of dollars of fines for those projects,” Chaturvedi said. “So it’s really powerful to be able to check all the information as it comes in, instead of just sampling data.”

Sampling is how the work has historically been done. Auditors and consultants pull a subset of payroll records and vendor invoices, then extrapolate. The Dili argument is that continuous checking across the full record is now cheap enough to replace the sample.

The architecture is the more interesting part of the pitch. Large language models are confined to the data layer, where they convert unstructured documents into structured fields. A deterministic engine then applies the compliance rules to that structured output.

That split is a direct response to the reliability problem that has stalled AI adoption in regulated workflows. A model that decides whether a project complies is a model that can hallucinate a compliant answer. A model that only reads a timesheet and hands the numbers to a fixed rule set has a narrower failure mode.

Chaturvedi described the ambition in terms of coverage rather than accuracy. He pointed to reading across internal documents, vendor documents, ERP records and payroll systems to draw out the specific data a reporting or compliance obligation requires.

The company says its software is running on approximately 700 projects spanning manufacturing facilities and data centres. Roughly half of those projects use Dili as an in-house software tool. The other half outsource the compliance process to the company entirely, on a contractor model.

That split matters for how the business should be valued. Services revenue carries different margins and different scaling characteristics from software revenue, and a company that is half services is not a software company priced at software multiples. Chaturvedi expects the mix to shift.

“Software and AI are going to start eating a lot of those professional services workflows, so I think more and more people will start to bring those in-house,” he said. “The interesting thing will be how the market itself evolves and where the customer needs go as AI develops.”

To be sure, the services half of the business is also a distribution advantage in a market where buyers are construction firms rather than software buyers. Selling an outcome is easier than selling a platform to a general contractor with no compliance software budget line.

The company intends to expand beyond prevailing wage and apprenticeship monitoring into broader audit, assurance and waste-detection workflows currently handled by large consulting firms. That is a considerably larger addressable market and a considerably harder one, since it puts Dili in competition with incumbents that hold the client relationships.

Investor interest in the category reflects a broader pattern. Compliance is document-heavy, rule-bound and expensive to staff, which makes it an obvious target for extraction and matching technology. It is also crowded. “AI for compliance” is a common pitch, and Dili differentiates on the specificity of the rule set it has encoded rather than on the underlying technique.

Allianz is the notable name on the Dili cap table. Insurers have a direct interest in construction compliance data because it bears on the risk they underwrite, and an insurance-adjacent distribution channel would be a meaningful advantage over generalist vendors.

Dili has disclosed no revenue, contract value or customer retention figures. The $15 million gives it runway to prove whether continuous compliance monitoring converts to recurring software revenue before the next raise.