12 Claude Prompts for B2B Account Executives
Most sales tooling in 2026 is a CRM with an AI sidebar bolted on. The sidebar can rewrite your subject line. It can't tell you that the VP of RevOps at your stalled deal quietly left two weeks ago, or that one of your closed-lost champions just took the same role at a hotter logo. The prompts below do. Each is a workflow an AE can run today from Claude Desktop with the LinkFetch MCP server connected. Copy, paste, swap the bracketed bits. That's the whole motion.
LinkFetch is a compliance-first LinkedIn data API for AI agents, native to Claude and MCP rather than raw HTTP. Every prompt below assumes you have it wired into Claude Desktop and a working API key. The setup walkthrough is in the daily outreach build post; skip that step if you've already got an MCP server humming.
Why AEs lean on agents differently than SDRs
SDRs work the top of the funnel: volume, list quality, first-touch personalization. AEs work the middle and bottom: stakeholder mapping, champion movement, deal-stage signal, expansion plays. Sixty-one percent of AEs say their biggest time sink is researching named buyers before a call rather than prospecting net-new ones (coffee.ai, 2026). That research is what these prompts compress.
Two of these prompts are real-time (you run them in the 90 seconds before a meeting), four are weekly cadence (Monday morning radar), three are on-demand (triggered by a deal moving stages), and three are quarterly (forecast prep, QBR, territory planning). Run them all and you spend roughly 1,100 credits a quarter per active book of 40 accounts. That's an order of magnitude less than most teams spend on the same workflows in legacy sales-intelligence tools (crispy.sh, 2026).
What separates a useful prompt from a buzzword soup
The prompts above share a shape. They name a specific tool from the LinkFetch surface, they return a structured output (table, ranked list, or row count), and they tell Claude what to do when the data isn't there ("skip private profiles", "say so explicitly"). Without those guardrails Claude will improvise. With them, the output is something an AE can paste straight into the CRM note field.
A second pattern: every prompt names the cadence. Real-time prompts are fired from a call prep doc. Weekly prompts run as part of a Monday-morning routine. Quarterly prompts are forecast-prep. Cadence is not a nice-to-have, it's what turns a clever one-off into a repeatable workflow. AEs who pin these as Claude Desktop saved prompts and re-run them on schedule are the ones who actually compound the value.
The credit math by prompt type
| Prompt type | Typical cadence | Credits per run | Realistic monthly load |
|---|---|---|---|
| Pre-call brief | Real-time | ~10 | 200–300 (20–30 calls / week) |
| Champion-change radar | Weekly | ~30 | 120 |
| Renewal-risk early warning | Weekly | ~120 | 480 |
| Hiring-spike scan | Weekly | ~200 | 800 |
| Inbound triage | Daily | ~10 | 200–400 |
| Account-list build | Quarterly | ~250 | ~85 amortized |
A working AE running this kit at full cadence consumes roughly 1,900 to 2,400 credits a month. LinkFetch charges a flat per-request rate at every tier, so the spend stays predictable as your book scales (connectsafely.ai, 2026). No overage multiplier, no cache discount cliff, no surprise at the end of the quarter.
Where AEs misuse these prompts (and how to not)
Two failure modes show up reapeatedly when AEs first wire up an LLM-driven workflow on top of a LinkedIn data surface. The first is running the radar prompts every day instead of weekly. Champions don't change roles often enough to justify daily polling, and the noise floor of "no relevant change" output will fatigue the muscle that's supposed to read the actual alerts. Stick to Monday cadence for radar work.
The second is treating the inbound-triage prompt as a hard disqualification gate. The score is a signal, not a verdict. AEs who read the recommended response track and then check the strongest signal column manually will calibrate the prompt to their own ICP much faster than AEs who let the rank order drive the queue. The prompt is a force multiplier on judgment, not a replacement for it.
FAQ
Are these prompts safe to run on a LinkedIn-restricted browser?
Yes. LinkFetch operates inside the AE's own LinkedIn session through the browser extension, which means every request runs against profiles the AE can already see when logged in. The platform does not rotate sessions, scrape with proxies, or use synthetic accounts. The LinkedIn account behind the session is the principal, and the compliance posture follows from that.
How do I share these saved prompts with the rest of my team?
Claude Desktop supports saved prompts that sync to your workspace if
you're on a team plan. Paste each block above into a saved prompt,
name them by cadence (weekly-radar, pre-call, inbound-triage),
and your whole AE team can run the same ones with their own LinkFetch
keys. Credit usage is per-key, so each AE owns their own budget.
Will these break if a buyer has a private LinkedIn profile?
The prompts include explicit instructions for what Claude should do when a profile is private. For most prompts that means returning the surface fields only (current company from the profile URL, public title) and noting the gap in the output. The exception is the champion-change radar, which falls back to gracefully skipping the private row rather than guessing at a change.
Can I add my own custom prompts to this kit?
Yes, and you should. The 12 above cover the recurring AE workflows that show up across most B2B teams, but every account-list, every ICP, and every deal shape has its own quirks. The fastest way to extend the kit is to take a prompt that's close to what you need, run it once, edit the bracketed sections to be more specific, and save the new version. After three or four iterations you'll have a prompt-library that fits your motion exactly.
Do I need to be technical to set these up?
No. The setup is a one-time install of the LinkFetch MCP server in Claude Desktop, walked through in the daily outreach post. After that, every prompt above is plain English copy-paste. The hardest technical step is pasting your account list into the bracketed placeholder, and Claude will tell you if the format is off.
Last updated: 2026-05-25 · Author: LinkFetch team