Pre-Call Research Agent — Enterprise Briefing Skill

Enterprise Tier Automatically researches meeting participants before every call. Saves 20-30 minutes of manual prep every time.

⚡ Two modes of operation:

ON-DEMAND Send a name + company → agent researches 3 public sources → structured briefing in under 2 minutes.
CALENDAR SCAN Cron job scans your calendar every hour → auto-briefs on external participants → delivered 20 min before each call.

What It Does

Research Pipeline

When you trigger a research request, Hermes queries 4 data sources in priority order:

Source	What It Extracts
🏢 Company Website	About page, blog, team page — company size, products, recent announcements, leadership
📰 News Search (90 days)	Articles mentioning the person or company — product launches, funding, partnerships, speaking events
👤 Public Profiles	LinkedIn, Wikipedia, GitHub — role, career history, education, publications
📁 Your CRM / Wiki	Past interactions, meeting notes, purchase history (optional, one-time setup)

Privacy: Only public data is searched. Hermes never scrapes behind login walls. If no info is found, it honestly says so rather than fabricating.

How to Use It

On-Demand Triggers

Send any of these to your Hermes Telegram bot:

Trigger	Example
`research [person] from [company]`	`research Sarah Chen from Stripe`
`brief me on [person] at [company]`	`brief me on John Doe at Acme Corp`
`prep me for [time]`	`prep me for 3pm` — reads your calendar
`who am I meeting tomorrow?`	Scans calendar, briefs all external participants
`scout [person] from [company]`	`scout Satya Nadella from Microsoft`

Hermes responds immediately: "Researching Sarah Chen (CTO, Stripe) — sending briefing in ~2 minutes."

Calendar Auto-Scan

Set it once — tell Hermes:

Set up a daily scan of my Google Calendar at 7am weekdays.
For every external meeting, research the participants and
send me a briefing 20 minutes before the call.

After that, Hermes:

Scans your calendar every hour (8am-6pm weekdays)
Detects external participants in your events
Researches new people (skips recently-researched ones — 7-day cache)
Delivers briefings to your Telegram 20 minutes before each call

You do nothing. It just works.

What You Get

Sample Briefing Output

📋 PRE-CALL BRIEFING
━━━━━━━━━━━━━━━━━━━━━━━

Sarah Chen · CTO · Stripe
📅 Your call: Today, 3:00pm

👤 ABOUT THEM
─────────────
• CTO at Stripe (since 2022) — previously VP Engineering at GitHub
• Background: Distributed systems, payments infra, developer tools
• Education: MIT CS (MEng), Stanford MBA

🏢 COMPANY CONTEXT
──────────────────
• Stripe: ~$65B valuation, 8,000+ employees
• Recent: Launched Stripe Connect in Brazil (Mar 2026)

📰 RECENT
─────────
• Spoke at Fintech Conf London — focus on real-time payments
• Company blog: "Building for a multi-currency world" (Apr 2026)

💡 YOUR HISTORY
───────────────
• Last contact: Email thread re: API integration, Nov 2025

🎯 SUGGESTED TALKING POINTS
───────────────────────────
1. Real-time payment infrastructure in LATAM
2. Multi-agent automation for developer workflow
3. Follow up on API integration from Nov 2025

⚡ KEY INSIGHT
──────────────
Sarah prioritizes engineering efficiency metrics.
Frame in developer hours saved, not features.

━━━━━━━━━━━━━━━━━━━━━━━
🟡 Medium confidence — company + news data confirmed
Sources: linkedin.com · stripe.com/about · fintechconf.io
AI-generated — verify critical facts before use

The briefing includes background, company context, recent news, your history, 3-5 talking points, a key insight, cited sources, and a confidence score.

Sensitivity Dial

You control how deep the research goes:

Level	What It Includes	When to Use
Professional (default)	Role, company, industry news	Most calls — safe, non-invasive
Standard	+ Publications, talks, industry awards	Important meetings, pitches
Deep	+ Social media, personal interests, broader public records	Key deals — use sparingly

To change: research Sarah Chen from Stripe, Standard sensitivity or set a permanent default with Always use Professional sensitivity for pre-call research.

    ⚠️ Deep mode warning: Deep mode searches broader public records including social media. Hermes will ask you to confirm before generating a Deep briefing.
  

Edge Cases

Situation	What Happens
No public info found	"Limited public information available for [name]." Hermes never fabricates. You still get company context.
Multiple people in one meeting	Primary person gets full briefing. Others get 3-5 bullet points each.
Common name (e.g. "John Smith")	Include the company: `research John Smith from Stripe`
LinkedIn profile is private	Falls back to company website + news. Briefing notes where data is limited.
Briefing arrives late	Pre-cached for calendar mode. On-demand: < 2 min in 95% of cases.

Setup Checklist

✓ Your Hermes agent is configured (done during Enterprise onboarding)
✓ Telegram bot is connected (done during onboarding)
✓ Calendar integration: gcalcli installed and authorized (trainer assists, ~15 min)
✓ CRM export: one-time CSV from HubSpot/Salesforce (optional)
✓ Test: send research Satya Nadella from Microsoft to your bot

Tips

Research ahead: For important calls, trigger the night before. Briefings are cached for 7 days.
Group meetings: Calendar auto-scan handles multiple participants automatically.
Verify key facts: Every claim includes a source link — click to verify if something seems surprising.
Sensitivity: Stick to Professional for routine calls. Reserve Deep for critical deals.
Give feedback: If a briefing is ever wrong: "That briefing on [name] was incorrect — they're actually [correction]"

📋 Pre-Call Research Agent