How Salespeople Should Set Up Context for Their AI Tools
Salespeople use AI to research prospects, draft outreach, prep for calls, and write follow-ups
Salespeople use AI to research prospects, draft outreach, prep for calls, and write follow-ups. It's a real time-saver — until the output reads like a template, because the AI doesn't know your product, your ICP, or the deal you're actually working. Generic outreach is worse than no outreach, and generic is the AI's default.
Here's how a salesperson sets up context so AI stops sounding like a mass email.
Why does AI write generic sales outreach?
Because effective outreach is specific, and specificity is exactly what the AI lacks without context.
Ask for a cold email with no context and you get the email everyone gets — vague value prop, interchangeable phrasing, nothing that proves you understand the prospect. The model can't reference your actual product strengths, your ICP's real pain points, or where this prospect fits, because it doesn't know any of it. So it writes the safe, average version, which is the version that gets ignored.
Salespeople live or die on relevance, and relevance is what's missing when the AI is working blind.
What context should a salesperson give their AI?
The things that make outreach land instead of bounce:
- What you sell — your product, its real strengths, the proof points you can stand behind
- Your ICP — who you sell to, their pain points, their triggers, common objections
- Your positioning — why you over the alternatives, in language that isn't marketing fluff
- Deal context — for a given prospect: their situation, prior touches, where the conversation stands
With those in context, "draft a follow-up to this prospect" comes out referencing the right pain point, the right proof, the actual state of the deal — a message you'd actually send, not one you'd rewrite.
Why deal context has to stay live
Sales context isn't a profile you write once. A deal is a moving thing — a discovery call surfaces a new pain point, a champion goes quiet, a competitor enters, the prospect replies with an objection that changes everything. Outreach written against last month's understanding of the deal is outreach that misses.
That's why a static "about our product and ICP" doc only gets you halfway. The product and ICP layer is relatively stable; the deal layer changes constantly, and it's where relevance is won or lost. What helps is context that updates as the deal moves — pulling from your call recordings, your CRM, the back-and-forth — so when you ask the AI to draft a follow-up, it's working from where the deal actually is today, not where it was three calls ago. Live context is the difference between a follow-up that references what was just discussed and one that sounds like it wasn't paying attention.
Isn't that what AI sales tools are for?
Sales AI tools and second-brain apps for sellers exist, and some are good. But here's a practical reality worth leaning into: most sales tooling already plays well with Claude and other AI — CRMs, call recorders, and sales platforms increasingly expose their data to the AI you use. The infrastructure to use your sales context in AI is largely already there.
What's missing isn't another sales app — it's the context itself, structured and current, ready for those tools to pull. A context layer fills exactly that gap: it gathers your product, ICP, and deal context from the sources you already use and serves it to whatever AI you draft and research in. The sales tools do the selling motions; the context layer makes sure every one of them, and every AI alongside them, actually knows the deal.
How should a salesperson actually set this up?
Keep your product, ICP, and deal context in one place and serve it to every tool you use.
A context layer pulls from where this already lives — your CRM, your call recordings, your sales materials — keeps it current as deals move, and delivers it to any AI tool through MCP. Set it up once, and every tool you open already knows what you sell, who you're selling to, and where each deal stands, without you pasting a product one-pager into every chat.
For a seller working a pipeline across research, outreach, and follow-up tools all day, that's the difference between AI that writes templates and AI that writes messages specific enough to get replies.
→ What to put in it: What Information Should You Add to Your AI Context?
→ How delivery works: How to Deliver Personal Context to AI Tools
→ Set up your context layer with Unabyss →
