I watched a rep build a stakeholder map recently. It took two and a half hours.
They logged into Sales Navigator. Searched for the first job title. Tried different keyword variations. Scrolled through results. Found the right person. Clicked a Chrome extension to push them into a B2B database. Pulled the phone number and email. Synced to the CRM. Then did the whole thing again for every other persona they needed to find at that account.
And at the end of it, all they had was a list of names and contact details. No context on what each person cares about. No pain points mapped to their role. No messaging angles ready to go. They still had to sit down and think through all of that themselves.
That’s 4-5 tools, 2-4 hours per mid-market account, and message quality that depends entirely on how much effort the individual rep puts in that day. When you’re busy and you’ve got too many accounts on your plate, the research suffers. And when the research suffers, the outreach suffers.
So I built something better.
One click in the CRM
The rep hits an “Enrich Account AI” button in the CRM. The system handles the rest.
It pulls firmographics for the account and checks it against the ICP criteria. It surfaces a brief fit assessment. In one of the examples I walk through in the video, the system noted that a company sat at the upper boundary of the ICP on headcount and was post-Series D when the sweet spot is Series A to B. But it also flagged that multi-geography knowledge scaling was a real pain point given their office locations. That’s nuance. Not just “fit” or “no fit” but “fit with caveats.” The kind of judgement call a rep would make if they had time to think it through.
It then searches for stakeholders matching the client’s target personas. In the demo, I used a fictional example (Trainual targeting Paddle) with personas across executive leadership, HR, operations, and sales.
For each stakeholder it finds, the system generates an interactive card showing their ICP persona tag, role-specific pain points, and 2-3 ready-to-use messaging angles.
The messaging is where it gets good
The pain points and messaging angles aren’t generic. They’re tailored to the individual.
A Chief People Officer gets angles around multi-country compliance and the gap between formal onboarding and role-specific training. A People Operations Manager gets angles about replacing the “doc graveyard” of scattered Notion pages, or stopping being the “human router” for every “where do I find X” question.
The messaging even pre-empts likely objections. For example, addressing the “we’ve already got an HRIS” pushback before the rep even picks up the phone. That’s not something most reps would prepare. But when it’s already there on the card, the quality of the first conversation goes up significantly.
Why it’s not fully automated (on purpose)
This is a deliberate design choice, and it’s the part I feel strongest about.
The AI surfaces potential stakeholders, but the rep reviews them before any contact enrichment happens. This matters for two reasons.
First, AI makes mistakes. It might pull through someone who isn’t the right fit, or surface two similar people where the rep (after checking their LinkedIn profile) would clearly prefer one over the other.
Second, enrichment costs money. Every contact reveal uses credits from the B2B database. By letting the rep choose which contacts to enrich, we avoid wasting spend on people who were never going to be prospected.
This is exactly how sales teams should be thinking about AI. Use it to eliminate the 90% of work that doesn’t require human judgement. Keep the human in the loop for the decisions that matter. Not everything needs to be fully automated. Sometimes the best workflow is one where AI does the heavy lifting and a human makes the final call.
The before and after
What used to take 2-4 hours per account now takes minutes. Reps go from juggling 4-5 tools to working inside a single workflow. And because every rep gets the same AI-generated research, there’s consistency across the team. The quality of stakeholder mapping no longer depends on who’s having a good day.
Scale that across 10 reps each handling 10-20 accounts per week and the time savings are enormous. But the bigger win is the messaging quality. When the pain points and angles are already there, reps spend their time personalising and selling rather than researching and tab-switching.
That’s the shift. Less time in spreadsheets. More time in conversations.
How I built it
I built this using Claude Code with MCP integrations. For the client’s instance, I connected it to their existing B2B database and CRM. In the video walkthrough, I demonstrate the concept using Claude Chat with Amplemarket’s MCP and HubSpot.
The design choice that took the most thought was where to put the human in the loop. For the pre-call research dashboard (which I wrote about in a separate post), I made it fully automated because the rep is consuming the research, not acting on it. They read and decide how to use it in the conversation.
For stakeholder mapping, the rep needs to make a decision before money gets spent (which contacts to enrich). That’s a judgement call. So the human stays in the loop at that specific point, not before, not after.
Thinking through where AI stops and where the human starts is the most important part of building these workflows. Get it wrong and you either waste money on bad data or slow down the rep with unnecessary approval steps.
And as always, the context layer is what makes it work. The ICP criteria, persona definitions, pain points, and messaging frameworks all need to be built in and kept up to date. Without that context, the AI is just guessing. With it, the output is genuinely useful from the first click.
The point
Both of these workflows (pre-call research and stakeholder mapping) follow the same principle. Automate the grunt work. Keep humans on the judgement work. And build the context layer that makes AI actually useful rather than generically impressive.
If your team is still building stakeholder maps manually across multiple tools, I can build a workflow like this for you. The setup, the context layer, the integrations, all tailored to your ICP, your personas, and your tech stack.
Happy to chat: yellowo.co.uk/schedule
You can watch the full demo walkthrough here: