gtmskills.com

Reference file

Person

person.md

Instructions

Run this only when personalization matters more than volume. For routine outreach, the Company brief is enough — don't pay the cost of a deep person research pass.

Step 1 — Check what you already have

Cheap before expensive:

  1. swan-search-companies and pull any existing contact records in Swan for the person.
  2. If a CRM is connected, search the CRM for the contact by email or LinkedIn URL — prior reps may have notes.
  3. swan-get-memory (org + company memory) — past briefs on this person may exist.

If a prior dossier exists and is < 30 days old, use it as the base. Refresh only what's stale.

Step 2 — Enrich the structured profile

swan-enrich-contact with the LinkedIn URL or email. Returns role, tenure, work history, location, education. Cheap, fast, structured.

If the contact has no LinkedIn URL yet, run swan-enrich-email first to find the work email, then enrich.

Step 3 — LinkedIn presence and recent activity

swan-linkedin-social-media-presence on the profile. From the returned posts, decide in 3–5 lines:

  • Active or dormant? (volume in last 90 days)
  • What do they post about? (themes, one line)
  • What kinds of posts get traction? (engagement skew)
  • Any clear signals: hiring announcements, product launches, public opinions on competitors, problem statements.

Don't dump every post — keep the take tight.

Step 4 — Mention pattern on the web

swan-fetch-scraped-url on a Google search like "<full name>" <company> to find press, podcasts, talks, GitHub, personal blog. Aim for 1–3 high-signal sources, not exhaustive.

What you're looking for:

  • Public talks → topics they care about, recurring themes
  • Press quotes → official positioning
  • Podcasts → unfiltered priorities and pet peeves
  • Personal site or substack → strongest signal of how they think

If nothing useful surfaces after 2 fetches, stop — they're a low-public-profile person and you've learned that.

Step 5 — Work history pattern

From the enrichment payload, look at the last 3–4 roles. Flag:

  • Tenure — long stints (3+ years) suggest a builder; short stints suggest a fixer or mover.
  • Industry stickiness — same vertical for 10+ years signals deep domain identity.
  • Company size trajectory — startup → enterprise or vice versa signals the environment they choose.
  • Co-workers — prior managers or peers who may be mutual connections.

One line per pattern. Don't theorize beyond what the data says.

Step 6 — Mutual connection check

swan-get-org-senders to list connected senders. Surface obvious overlap — same prior company, same school, geo, shared LinkedIn 1st-degree connection. Often the warmest entry point.

Step 7 — Compose the dossier

<Full name> — <Title> at <Company>

ROLE
  - one line on scope and reports.

CAREER ARC
  - one line per pattern from Step 5.

RECENT ACTIVITY
  - 3–5 bullets from LinkedIn + web.

WHAT THEY CARE ABOUT
  - inferred priorities, evidence-backed.

WARM PATHS IN
  - mutual connections, shared background, recent engagements.

OUTREACH ANGLE
  - one sentence on the highest-signal hook.

End with "Confidence: high / medium / low — based on signal density."

Rules

  • MUST cite evidence per claim. "Active on LinkedIn" alone is not enough; "12 posts in last 30 days, avg 200 reactions" is.
  • MUST stop at Step 4 if web research surfaces nothing after 2 fetches. Don't burn credits on a low-public-profile person.
  • NEVER invent quotes, beliefs, or priorities not present in the source.
  • NEVER use this for volume outreach. Use the lighter Company brief instead.
  • If a tool result is truncated, read the JSON from files/tool-outputs/<toolName>_<callId>.json in swan-execute-code.