Brilliant discovery calls. Hours lost writing them up.
Transformation consultancy · AI discovery engine · three-month proof of concept
The client
A transformation consultancy automating discovery.
Seeking to automate the journey from sales calls to strategic proposals.
The challenge
A universal consulting problem, at scale.
Discovery calls produced unstructured data that reps spent hours summarizing, and by the time proposals were written the sharpest insights had gone missing.
- Discovery calls producing unstructured data that reps spent hours summarizing
- Client problems never systematically mapped to the firm’s capabilities
- Key insights lost by the time a proposal was written
No scalable way to learn from hundreds of client conversations.
The solution
Leo, AI-powered discovery intelligence.
Over three months we built a proof of concept that listens to client calls, identifies problems, and maps them to the firm’s solutions. The pivotal innovation wasn’t the AI.
The solution-mapping matrix
We formalized the firm’s Delivery Framework into a matrix that became Leo’s intelligence foundation.
Transcript to structure
Fireflies and Otter ingest the call, OpenAI models extract the problems, Make and Steamship orchestrate the run.
Airtable interfaces
Custom interfaces where the team reviews, corrects, and approves every summary before it leaves.
The results
Proof of concept, delivered.
- 26 discovery summaries generated from real client calls
- 5 golden examples created as quality benchmarks
- 100% problem-capture rate, against a ~60% manual baseline
Why it matters
We didn’t promise magic. We built a system, measured honestly, and created a clear path to excellence — iteratively and transparently.
Keep reading
Related work.
Start here
Show us how the work moves today.
Thirty minutes. You walk us through how the work moves today; we tell you where it's breaking and what a system would take.
Pick a time and we’ll be there.