Applied AI · Strategy, build and run
We help growing companies get AI out of demos and into the work: strategy, build, and the ongoing run, from one small senior team.
§ 01The thesis
Most companies have tried AI by now, and the model is rarely what stops it. What stops it is everything around the model. We keep meeting the same three walls, whatever the industry.
Wall I
What the firm knows lives in people’s heads, inboxes, and documents nobody can find.
How we work on itWe build a working memory of the firm from its own documents, so answers come back with their source.
Wall II
The AI can’t see how work actually moves, so it can’t help with the work.
How we work on itWe map how the work really runs, then place AI at the steps where it changes the outcome.
Wall III
Staff don’t rely on output they can’t verify, so adoption stalls after the demo.
How we work on itEvery answer shows where it came from, and a person approves what leaves the building.
We’ve met the same three walls in law firms, executive search, public relations, design studios, packaging, industrial services and agriculture. The pitfalls don’t care about your industry.
§ 02Method
We start before you buy anything, build on what the diagnostic finds, and stay to run it. One team the whole way, so nothing is lost between the plan and the work.
We map where knowledge and time leak, find the two or three places AI changes your margins, and hand you a plan you can run with or without us.
A margin map and a build plan. Fixed fee, credited toward the build.
We deploy the pattern on your documents, your workflows, and your people. Your team uses it on real work from the first weeks.
Working systems, with sources your team can check.
We tune what’s live, extend it to new teams, and take on the next use case as it surfaces. Continuing is the default, not the obligation.
One monthly retainer. Cancel anytime.
§ 03The alternative
Hiring an AI lead and an engineer means two senior salaries and months of ramp before anything ships. We bring the team, and a pattern that has worked before, from the first week.
| Consideration | Hiring in-house | Thoughtspan |
|---|---|---|
| Time to first result | A search, then months of ramp | Working on real material within weeks |
| Who you get | One or two people, learning your business | A senior team that has met these problems before |
| Scope | Strategy or engineering, rarely both | Strategy, build, and the ongoing run |
| Commitment | Salaries and benefits, whether or not the work ships | A monthly retainer you can cancel anytime |
Tools are chosen to fit the work, and we work with the ones you already have.
We build and run what we recommend, and we stay accountable for whether people use it.
The walls are the same everywhere. What’s particular is your workflow, so that’s what we design around.
§ 04Field notes
Short accounts from our engagements and from the field. Client names are withheld; each note is marked as an engagement or an observation.
For a Houston law firm, the answer was spread across a decade of recorded annexations. We assembled the full chain from the county records, turned the legal boundary descriptions into a checked map, and set out what the lien does and doesn’t secure. The firm’s attorney checked it against what they knew. It caught the annexations that are easy to miss.
Two queries disagreed on a game studio’s monetization. We traced it to the definition of a paying user: later buyers were being counted in earlier periods. It’s why our AI analysts start from a company’s definitions, not just its schema, and show the query behind every answer.
For the owner of a global packaging company, we’re building an AI chief of staff that keeps a living record of his guidelines and relationships and drafts the replies he would write. Threads that go quiet come back to him before they slip.
In search firms, the candidates who matter are often buried in years of call notes, along with the notes that say who you can’t approach. No field in the database holds that. It’s the clearest case of the first wall we know.
§ 05The team
The people who diagnose your business are the people who build and run what comes next.
Co-founder · Strategy & product
More than ten years in product and data across 30+ titles. Leads the diagnostic and stays on the work through build and run.
Co-founder · Engineering & infrastructure
Data engineer and data scientist. Builds the pipelines, warehouses and AI systems underneath every engagement.
§ 06Questions
Good. The tool isn’t the hard part. Getting your firm’s knowledge and workflows into it, and getting people to trust what comes out, is. That’s the work we do, with the tools you already have where they fit.
Yes. Start with the diagnostic: two to three weeks, a fixed fee credited toward the build, and a plan you can run with or without us.
The diagnostic is a fixed fee. Build and run are one monthly retainer, prorated, that you can cancel anytime. We compare it to the cost of hiring an AI lead and an engineer, not to a software subscription.
We work across industries on purpose. The three walls look the same in a law firm, a search firm and a packaging company. What differs is the workflow, and the diagnostic is where we learn yours.
The founders. Jeremy and Eric run every engagement directly, from the diagnostic through the ongoing run.
You decide what we connect to. Access is agreed in writing before we build, and the systems we build for you keep their sources visible so your team can check every answer.
§ 07Begin
Two to three weeks. A clear map of where AI changes your margins, and a plan you can run with or without us.
Thoughtspan