How Telepath Thinks
Telepath reads your entire closed-won history, extracts the multi-dimensional patterns your won deals share, and uses those patterns to score your open deals against what actually worked at your company, not a generic benchmark.
The Fundamental Insight
Your CRM already contains the answer to "what does our ideal customer look like." It's sitting in your closed-won deals: every company that bought, the size they were, the industry they were in, the technology stack they ran, the problem they described in discovery. The signal is there. The problem is that humans can only hold so many variables in their heads at once.
When a sales leader builds an ICP, they typically recall the last five deals they were involved in, bias toward the largest or most memorable, and anchor on the two or three attributes that feel salient. AI doesn't have those constraints. It can read every deal simultaneously, weight signals by how often they co-occur with wins, and surface combinations that no human would have thought to look for. That is the job ideal customer profile software exists to do, and the rest of this page is how Telepath does it.
Six Stages. No Black Box.
Every step is traceable. You can see what Telepath found, why it weighted it, and how it changed your scores.
Connect
Upload a CSV of your closed-won deals or connect your HubSpot account via OAuth. Telepath reads your historical pipeline data: deal size, industry, company size, sales cycle, and every custom field you track. Only closed-won deals build the model: open deals show who you are talking to now, not what your ideal customer looks like. Personal columns Telepath detects are removed on our server before analysis.
Enrich
On paid plans, Telepath adds Companies House records for UK companies to fill some of the gaps your CRM has. Everything else comes from the fields you already record.
Extract
Telepath reads across every closed-won deal and finds what your wins have in common across the fields you record, such as industry, company size, region, deal value and lead source. The pattern comes from your wins, not from a workshop.
Weight
Not all signals are equal. Telepath weights each trait by how concentrated your won deals are on it and how many of them record it, computed from your own deals each time the model is built, not from a fixed rule. A simple checklist treats every criterion as equal; this does not. The model is calibrated to your data, not a generic dataset.
Score
Every open deal with enough data to score gets a T-Score (your 1,000 most recent open deals) from 0–100 based on how closely it matches your winning pattern; low-data deals are left blank, not guessed. The score is not a black box. Every T-Score comes with a full breakdown: which signals fired, which were missing, and what the gap means for that deal.
Deliver
The T-Score is written back to HubSpot as a single custom property on each deal, refreshed nightly against your current model as your pipeline changes. In Slack, /telepath-focus shows the three open deals most recently updated in HubSpot and where each stands now, and /telepath-pipeline describes how your open pipeline splits by stage and group and what data is missing. Neither names a rep or says what to do next.
Or ask it in Claude: the Ideal Customer Profile MCP answers pipeline questions in plain English, with a receipt on every number.
What Your T-Score Means
What Telepath Is Not
Four things people sometimes assume, and why they're wrong.
Not lead scoring
Lead scoring assigns points based on behaviour: email opens, page visits, form fills. T-Score ignores behaviour entirely. It measures fit: how closely does this company match the profile of your closed-won deals? Two very different questions.
Not a generic AI model
Telepath is trained on your data, not a shared dataset. The patterns it finds are specific to your market, your product, and your sales motion. A competitor using Telepath would get completely different signals.
Not a prediction engine
Telepath doesn't predict whether a deal will close. It tells you whether it looks like your best wins. The deal still has to be sold. What changes is where your team chooses to invest their time.
Not a replacement for good selling
A high T-Score doesn't mean "close it automatically." It means "this deserves your best effort." A low T-Score doesn't mean "walk away." It means "go in with eyes open." Telepath gives your judgement better inputs, not a substitute for it.
Telepath Gets Sharper Over Time
The model changes only when you choose to refit it. Every night Telepath checks whether the won deals it was built from have changed. When they have, you can refit against your latest closed-won deals: signals that stop appearing in your wins lose weight, and new ones get surfaced. The longer you use Telepath, the more wins the model has to learn from.
See It For Yourself in a Few Minutes
Upload your closed-won deals and get an ICP analysis, computed in code from your deals, in minutes. No commitment, no credit card. The report is yours to keep, regardless of whether you become a customer.
Get Your Free ICP Report