I was on a call with a client recently and asked how their reps prepare for discovery calls. The answer was what I expected: LinkedIn, Google, the CRM, the company blog, the careers page, maybe Sales Navigator for the org chart. Dozens of tabs. Heavy context switching. And about 2-3 hours of manual work per meeting.
The problem isn’t that reps don’t want to do the research. It’s that when you’ve got three calls stacked up in a day, something gets cut. And it’s always the research.
So I built something to fix it.
What the rep actually gets
When a new meeting is booked, the rep receives an email on the morning of the call with a link to a research dashboard. No manual research required. Open the link, read for 10-15 minutes, and walk into the call fully prepared.
The dashboard is an HTML page designed to sit on the rep’s screen during the call, with a navigation menu at the top so they can jump between sections. Here’s what’s in it and why each part matters.
ICP fit verdict. Before the rep even scrolls down, they know whether this company is a strong fit, borderline, or weak. The system checks the company against ICP criteria and persona definitions and gives a clear recommendation. This matters because not every meeting deserves the same energy. A strong fit gets full discovery. A weak fit gets a different conversation (more on that below).
Company snapshot. Industry, headcount, HQ location, funding stage, estimated revenue, key customers, markets they serve. Everything a rep needs to understand who they’re talking to, pulled automatically. This used to be the first 20 minutes of manual research. Now it’s just there.
Prospect profile. Job title, tenure, prior roles, location. But here’s where it gets interesting. The system reads their LinkedIn bio and pulls out personality insights. In one of the examples I show in the video, it flagged that the prospect “sees himself as a coach and enabler, not just a revenue driver.” That kind of insight changes how you open a conversation. You wouldn’t pitch that person the same way you’d pitch someone whose bio is all about hitting targets and scaling teams.
CRM status. The system checks the CRM for any existing relationship. If there’s a deal in the pipeline, it surfaces the lifecycle stage, associated contacts, deal amount, and suggests how to position the conversation. If there’s nothing in the CRM, the rep knows they’re starting from scratch. Simple, but most reps forget to check this until they’re already on the call.
Sales organisation breakdown. Department headcounts across inside sales, business development, sales ops, sales engineering, and enablement. Essential context when you’re selling into the revenue org.
Tech stack. What CRM are they using? What marketing, analytics, and ops tools are in place? This helps reps understand the existing infrastructure and spot gaps. If they’re using a competitor, the conversation is different than if they’ve got nothing in place.
Recent news and partnerships. The system searches the web for external data (product launches, partnerships, podcast appearances, press mentions). In the demo, it surfaced that the company had launched three new products in November. That became a natural discovery angle: “How are your reps getting up to speed on the new product lines?”
Fit signals and watch-outs. Positive indicators alongside things to be mindful of. For example, the system flagged that the prospect was based in Alabama while the client was UK-based, meaning time zones could be challenging for workshops. It also spotted that the company already had a four-person enablement team, something the rep would need to navigate carefully. It even pulled through sales methodology and quota expectations from job descriptions. That’s the kind of detail most reps would never find manually because they wouldn’t think to look at job postings as a research source.
Tailored discovery questions. This is where the research really earns its keep. The questions aren’t generic “tell me about your current process” filler. They’re built from the company’s specific situation. Launched a new product? The system asks about enablement for reps selling that product. Selling into four different verticals? It asks whether BDRs specialise by vertical or prospect across all of them. Using MEDIC as a sales methodology? It factors that into the questioning framework.
CRM notes
As well as the dashboard, the system writes a summarised version of the research directly into the CRM as notes. So even if the rep doesn’t open the dashboard, the key information is right there in the contact or deal record. I also included a link to the full HTML dashboard within the CRM notes so reps can easily jump between the summary and the detailed view.
What happens with weak fits
Not every meeting is with a perfect prospect. In the video, I show two examples (one strong fit and one weaker one). The weaker fit was a marketing contact, which for this particular ICP isn’t the strongest persona to sell to.
Most teams would ignore this and take the meeting blind. The rep would spend 30 minutes discovering what the system already knew: this person isn’t the right buyer. The dashboard flags it clearly, not as a warning to cancel the meeting, but as something to be mindful of. Reps can adjust their positioning and expectations accordingly. Maybe this meeting becomes a referral conversation rather than a discovery call.
How I built it
I built this using Claude Code with MCP integrations. Amplemarket’s MCP handles the company and prospect data (firmographics, org structure, tech stack, headcount breakdowns). Web search pulls in the recent news and partnership data. The CRM integration reads existing deal history and writes the notes back in.
But the tools aren’t the interesting part. The interesting part is the context layer. The ICP criteria, persona definitions, product positioning, and discovery frameworks all need to be built in. Without that context, the AI produces generic research that reps won’t trust and won’t use. With it, the output is specific enough that reps actually open it before every call.
I spent more time building the context layer than I did wiring up the integrations. That’s always the way. The plumbing is easy. The thinking is hard.
The real shift
The point of this isn’t to remove humans from the process. It’s to remove the grunt work so humans can focus on the conversation. A rep who walks into a call having read a 10-minute brief is a fundamentally different seller than a rep who spent 20 minutes Googling and still doesn’t know what sales methodology the company uses.
Automate the research. Master the conversation.
If you want to see how this would work for your team, happy to chat: yellowo.co.uk/schedule
You can watch the full demo walkthrough here: