nikam.dev ← All cases Telegram →
nikam.dev / Cases / AI customer intelligence for a B2B team
Enterprise · NDA · Customer intelligence + AI

Customer intelligence: a verified base and AI rehearsal of negotiations.

A closed platform for the B2B team of a large international player: a verified base of 600+ companies and almost 700 key people across four target segments - plus an AI engine where a salesperson rehearses the conversation with a specific decision maker before the call: pitch, objections, discovery. Segment debates, a 42-country cultural playbook, natural-language smart search. Not a single person enters the base without live-source verification. Under NDA.

What
Internal web product
Audience
B2B sales, analytics, leadership
Volume
608 companies · ~700 decision makers
Status
Live · NDA
- The problem

A B2B sale is prepared blind: who decides, what hurts them and how to talk to them - all discovered on the call itself.

The team sells an expensive product to companies across countries. The client picture lives in the salesperson's head, an outdated spreadsheet and a dozen LinkedIn tabs. Every failed pitch burns a live client that took weeks to warm up.

Pain 01
Company and people data goes stale - and lies
Ask ChatGPT «who is the CEO of company X» - it answers confidently and wrongly: the model remembers a world one and a half years old. An internal audit found ~10% factual errors in quickly-assembled profiles - including people long gone from their companies.
Pain 02
Pitches are tested on live clients
The hypothesis «will this framing land» is tested in combat. A failed call isn't minus one attempt - it's a burned contact that took weeks to find and warm. There is no cheap way to rehearse the conversation.
Pain 03
Objections are a surprise every time
Each segment has its own pressure points: an asset manager objects differently from a media platform. The salesperson learns them by collecting bruises - and the knowledge stays in their head and leaves with them.
Pain 04
Cultural differences kill deals
The same framing works in opposite ways in Zurich and Riyadh: one culture values directness and numbers, another - hierarchy and relationships. The team learns this on lost deals.

A base you can trust - and an AI you can rehearse on.

The platform holds two things at once: a verified picture of the market (companies, people, priorities, communication styles) and a simulation engine where a salesperson safely plays out the conversation with a specific decision maker - before calling the real one.

A verified base across target segments
608 companies across four target segments: profile, size, stack, pain points, product-fit score. Each with its key people: role, priorities, communication style, quotes. One search - the whole picture.
Not a single person without a checked source
After an internal quality audit (found ~10% errors in «lean» profiles), a hard rule: no person enters the base without live-source verification. Departed people are flagged with dates, contested facts get a «conflict» flag. Trust in every row is the base's main asset.
Conversation simulations with a decision maker - 5 modes
Survey, discovery, pitch, objection handling (the counterpart «pushes back hard» - by design), feature prioritisation. The AI plays a specific person from the base - with their priorities and manner. Sessions are saved, insights extracted automatically, the hypothesis gets a verdict: confirmed / denied / needs work.
A collective segment persona
The «voice» of up to 15 real decision makers of a segment in one counterpart: priorities are aggregated, and the salesperson effectively talks to the whole segment at once. A fast way to test the message before name-by-name rehearsals.
Debates: segment vs segment
Two segments (or two specific people) argue about the product for several rounds - the AI produces a synthesis: where arguments align, where they split, what to take into the pitch. A surprisingly cheap way to find positioning blind spots.
Cultural playbook: 42 countries
42 country profiles across 21 parameters: negotiation style, hierarchy, attitude to risk and to data. Plus organisation archetypes and buying scenarios. A simulation can be run «inside the culture» of a specific country.
Natural-language smart search
«Who in the base is the hottest fit for our new module X?» - the search runs over the entire base and returns a ranked list with the reasoning for each match.

Four layers: data, verification, simulations, enrichment.

The core of the product isn't «yet another company database» - it's data-quality discipline plus an engine that turns the base into a rehearsal room for the salesperson.

/ 01
The data model
Company → key people → segment. Per person: role (decides / influences / uses), priorities, communication style, quotes. Departed people carry a marker and a date, contested facts a flag.
/ 02
The verification protocol
Every fact with a source and a date; a two-source rule, otherwise «unverified». Regular reconciliation rounds: weekly news scan, monthly «is this person still there?» check.
/ 03
The simulation engine
Plays a specific decision maker, a segment, or a culture. Every session's insights are extracted and accumulated: objections, reactions, phrasings - knowledge no longer leaves with the salesperson.
/ 04
Enrichment with scoring
Every profile has a 0-100 completeness score; thin profiles are enriched first. A company suggested by an employee goes through a workflow and lands in the base with a team notification.
- Outcome

What changes for the team.

608 companies
In the verified base across four target segments - with profiles, key people and a product-fit score.
5 simulation modes
From discovery to hard objection handling - plus segment debates and a 42-country cultural playbook. The pitch is rehearsed before the call, not on the client.
0 unverified people
The rule after the internal audit: no person enters the base without live-source verification. Errors are caught in reconciliation rounds, not in front of a client.

Is your sales prep running blind?

If your team sells an expensive product and every pitch is tested on live clients - let's talk. In 30 minutes we'll map what a verified base and a negotiation rehearsal room would look like for your pipeline.

Message on Telegram → nick@nikam.dev