We embed for six weeks, sit with your operators, walk your workflow, and find where AI and automation actually pay back. You own the deliverable regardless of next step.
The deliverable is a workflow map, a decision-point inventory, a prioritized opportunity list with effort, payoff, and AI-fit scores per workflow, and a written 90-day plan with sequenced builds. PDF, ~25-40 pages, with an appendix of what we observed but didn't recommend acting on and the honest reason.
The document is yours. Take it to your in-house team. Take it to a different firm and get a fixed bid against it. Use it to justify headcount internally. We don't structure the diagnostic to lock you into the next step.
The work we looked at and chose not to recommend. The integrations we considered and decided weren't worth the effort. The workflows where the AI model isn't ready today and probably won't be for 12-18 months. Operators trust the plan more when they see what we left out.
Two on-site days at the start. We shadow operators, sit in on standups, watch the actual workflow happen. We don't ask "what does your day look like" — that gets us the sanitized version. We watch.
Recorded calls with each functional area. We map the data flows: where work enters the system, where it gets routed, where it gets approved, where it gets paid out. We catalog every integration, every spreadsheet, every workaround.
Each workflow gets scored on five criteria (see below). We draft the 90-day plan, sequence the builds, and run the draft past operators to pressure-test it. The plan changes based on what they push back on.
Two on-site days at the end. 60-minute final readout with your full team, then a working session with the operators who'd own each build. You leave that day with the PDF and an answer on every question your team raised.
Engineering weeks, dependencies, integration risk. We score this the way the team that would actually build it scores it, not the way a sales deck scores it.
Operator hours returned per week. Error rate reduction. Throughput lift. Tied to the metric your team is already accountable for.
What's the cost if the automated path fails on an edge case? Some workflows are fine to fail-soft. Some are not. We score that.
Does the team have the slack to adopt this? Is there a champion? Has the org tried and failed at this before? Workflows score lower if the human side isn't ready.
Is the LLM ready for this task today, or is it still too early? We track which model classes solve which problem shapes and we won't recommend AI on a workflow where the model is going to fail two out of ten times. The build shape this implies lives on our agentic AI work.
A freight brokerage we worked with had eight dispatchers who collectively spent ~4 hours/day parsing inbound RFP emails from shippers into their TMS. Copy-paste from email body, fix the format, route to the right agent. Every day. Nobody had asked anyone to score it.
During the workflow walk-through we watched it happen. We pulled six months of inbound RFP samples, ran them against a structured-extraction prompt, and got 94% field-level accuracy on the first pass. We scored the workflow: effort 3/10, payoff 9/10, risk 4/10 (fail-soft, dispatcher reviews before commit), readiness 8/10, AI fit 9.2/10.
That workflow became the first build engagement. The plan in the diagnostic told the team in plain terms what would be built first, why it ranked above the other twelve opportunities, and what we expected to ship in the first 90 days. No slide deck. A written plan they could fund.
Operators who know there's friction in their workflow but can't articulate where the highest-impact fix is. Companies who've been burned by agencies promising AI without understanding the work. Ops leaders who need a credible written document to get internal buy-in.
Anyone with a clear, scoped project — you should skip to Build. Anyone who wants a deck (we don't make those). Anyone who wants sales pitches around AI without an honest read on where it fits and where it doesn't.
If you already know what to build, the diagnostic is over-engineered for your situation. Skip to Build. If your operation runs on Procore and the friction is between Procore and the rest of your stack, our Procore page covers that pattern specifically.
Fixed fee, paid up front, varies by operation size and number of workflows in scope. Most diagnostics land between the smaller fixed-scope audit on /services and a quarter of a typical build engagement. We share the number on the scoping call after we understand what's in front of us.
We tell you on the scoping call before you pay anything. We're not the right fit for consumer apps, ad-driven products, gambling, or work outside operational software. If your problem is closer to a different studio's wheelhouse, we'll point you there.
Read-only access to the tools your team uses every day. Usually that's your project tool (Jira, Procore, Salesforce, Monday), your accounting system (QuickBooks, NetSuite, Sage), and any custom dashboards or spreadsheets your operators rely on. We sign mutual NDAs before we touch anything.
Yes. If you engage us for a build within 60 days of diagnostic delivery, the diagnostic fee credits 50% against the first invoice. After 60 days the credit lapses, but the document is still yours.
PDF, ~25-40 pages. Workflow map, decision-point inventory, prioritized opportunity list with effort/payoff/AI-fit scores per workflow, written 90-day plan with sequenced builds, and an appendix of what we observed but didn't recommend acting on (and why). You can hand it to any other firm and get a fixed bid against it.
Some of it. The two on-site days are valuable because operational reality is different from what any leader says it is. We've done remote-only diagnostics for clients we already work with on retainer; for new engagements we'd rather travel.
You decide. Some clients move straight into a build with us. Some take the plan to their in-house team. Some take it to a different firm. We don't structure the diagnostic to force a next step. The honest read is the deliverable.
Usually 2-4 weeks from scoping call to kickoff. If you need it faster, tell us; we'll be honest about whether we can move that fast without dropping quality.