Nathan Thomas

A construction-fintech platform: AI document analysis

Current. Fractional design lead, via Toptal.

This engagement stays anonymized as "a construction-fintech platform" until the product launches and I have written approval to show it. The full case study, with screens and outcomes, follows then.

Situation

A small team is building a product that reads construction-loan draw packages, the stack of documents a lender reviews before releasing the next round of funding, and surfaces what a reviewer needs to approve or flag. The domain is genuinely hard and the buyer is conservative with money.

Constraint

Pre-launch, small team, no instrumentation yet. The product has to feel trustworthy to a reviewer who is responsible for real dollars, and it has to be buildable by a small engineering team without thrashing.

The decision

Anonymized until launch and written approval. The short version: I led the core workflow simplification, brought a recommendation with the options and the trade-off behind it, and we built from that call rather than from a wall of directions.

Outcome

To be verified at launch. I'm capturing baseline timings and demo-session observations now, so the next version of this page carries real adoption and time-saved numbers rather than estimates.

The honest scoreboard

What I can verify

I own the core workflows and bring one to two directions with a recommendation in every review. The client renewed past the MVP.

What I couldn't measure yet

Post-launch adoption and time saved. Not live long enough. I'm capturing baselines now so the next version of this page has real numbers.

← Back to work