Title Resource Group
I mapped the title-search workflow and put AI on the repetitive exam work. The examiner still verifies each result.
Already 30% of the annual cost target.
Discovery
Find the AI + UX bet that pays for itself, before anyone builds it.
For product and operations leaders with a full AI backlog and no way to rank it.
Email me
Tell me what’s broken and where you’re stuck.
What you get
When the AI backlog is full
You can name ten places to add AI. You cannot yet say which one pays for itself. This engagement answers that before anyone builds.
A shortlist you can fund
You leave with a few opportunities scored on reach, impact, confidence, and effort, the same pass growth teams at Dropbox and Airtable run every planning cycle, and a clear place to start the roadmap.
Mapped from the real workflow
I interview the people who do the work, then map the process and the customer journey so the leaks in time, cost, and drop-off are visible. Opportunities are written as needs and pain points, never as features. Most AI backlogs are solutions looking for a problem.
For complex products
Enterprise, fintech, regulated workflows, and SaaS. Old systems and unclear ownership are a normal place to start.
Aimed at the next 15–30%
The recommendation is the lever that moves efficiency, conversion, retention, or cost. Sometimes that lever is AI. Sometimes it is a workflow or UX change.
Examples of this work
Swiss Re
I joined mid-build, learned the governance domain, and modernized an 18-year-old platform while daily approvals kept working.
18 years of workflows, still in daily use.
BC Tool
I replaced the analyst spreadsheet with a workspace that reads brand-tracking waves across three markets and four years.
Results without opening the spreadsheet.
How it works
- 01
Brief
You tell me what is leaking: time, cost, drop-off, or a backlog of AI ideas no one can rank.
- 02
Map the work
I interview the people who do the job and map the process and the customer journey.
- 03
Score the bets
Each opportunity gets a quick score on reach, impact, confidence, and effort. The top few get a detailed pass with real numbers. The ones that do not pay stay off the list.
- 04
Name the first move
You leave with a shortlist and a recommended start. Sometimes that start is AI. Sometimes it is not.
The approach
Outcome first, opportunities second, solutions last.
The AI backlog is a list of solutions. Discovery turns it back into a tree: one outcome at the root, the real opportunities under it, and only then the bets. Teresa Torres's opportunity solution tree and the RICE scoring from Lenny Rachitsky's archive, run against your workflow.
- Outcome
- One number the engagement serves: cost per case, time to resolution, conversion, retention. If leadership cannot name it, that is the first week's work.
- Opportunity map
- The current workflow as an experience map, with the needs and pain points under each step. Written as problems, not features. Teresa Torres puts it bluntly: most teams write their opportunities as solutions.
- Score
- A 30-second RICE score on everything, then a 30-minute detailed estimate on the top few with real reach and real cost. Dropbox teams that added the detailed pass moved their metric twice as much.
- First move
- A recommended start with the evidence behind it. Sometimes that is an agent. Sometimes it is a workflow change or UX on the product you already have.
Fit
This is for
Product and operations teams
Enterprise, fintech, regulated workflows, and SaaS. You have a real process, unclear ownership is fine, and you need a decision before a build.
This is not for
A demo, not a decision
Not a fit if you already know the build and only need extra hands, or if you want a chatbot shown to leadership this week.
FAQ
A scored shortlist, the map of the current workflow, and a recommended first move. Enough to fund the next step or to stop a bad one.
Most discovery sprints run two to four weeks, depending on how many teams I need to interview. I will tell you the range before we start.
Reach, impact, confidence, and effort, scored fast on the whole list and in detail on the top few. Confidence is where most AI ideas lose: nobody has checked whether the model can do the job on real cases, so we check before you fund it.
No. The job is to name the lever that moves the number. That is sometimes an agent, sometimes a workflow change, sometimes UX on the product you already have.
Someone who owns the number, and the people who do the work today. I keep the interview load light. Leadership only needs to be in the room for the decision.