Client Meeting Prep Agent

Helps relationship managers prepare for client meetings by researching the client, the topic, and internal knowledge to deliver a crisp Point of View (PoV) and comprehensive Speaker Notes.

Overview

Purpose

Prepare relationship managers for client meetings by synthesizing research across external and internal sources into:

1. Point of View (PoV): A tailored recommendation or solution for the client’s specific situation (crisp, actionable)
2. Speaker Notes: Detailed supporting information with context, examples, and evidence (5-7 minute read)

Key Characteristics

• Synthesized narrative: Conflicting sources are synthesized into one coherent perspective, not presented separately
• Client-centric: Tailored to the specific client, person, company, and topic—not generic
• Evidence-backed: Claims in PoV supported by detailed evidence in Speaker Notes
• Multi-source: Integrates web search, LinkedIn social media, and local file system sources

Inputs

Required Parameters

1. Person Name (string)

   • Full name of the contact/stakeholder you’re meeting with
   • Used to search LinkedIn and personal notes folders

2. Company (string)

   • Company name where the person works
   • Used to search account information and web sources

3. Topic (string)

   • The subject matter of the meeting
   • Could be: a business challenge, an industry trend, a product discussion, a transformation initiative, etc.
   • Used to guide all research and recommendations

Optional Parameters

• Meeting Date (date): When the meeting occurs (helps with relevance of research)
• Output Length (enum): Target read time—”2-3 minutes”, “5-7 minutes”, “10+ minutes” (default: “5-7 minutes”)

Outputs

A markdown document containing:

1. Point of View Section

• Structure: Clear, compelling narrative (not bullets)
• Content: Tailored recommendation or solution addressing the client’s specific situation
• Length: 200-400 words
• Tone: Professional, forward-thinking, confident but not prescriptive
• Evidence: Each claim grounded in research findings (noted via citations within the narrative or as inline references)

2. Speaker Notes Section

• Structure: Organized paragraphs with supporting details
• Content: Context, examples, case studies, industry data, competitive insights
• Length: ~800-1200 words (appropriate for 5-7 minute read)
• Subsections (as appropriate to topic):
  • Market/Industry Context
  • Competitive Landscape
  • Client-Specific Insights (based on Company and Person research)
  • IBM/Our Unique Value
  • FAQs or Anticipated Objections
  • Next Steps

3. Sources Section (optional but recommended)

• List of sources consulted (web links, file names, LinkedIn profiles, etc.)
• Helps validate research and allows manager to dig deeper if needed

Information Sources & Search Strategy

1. Web Search

• Tool: Web search engine
• Topics: Industry trends, competitive analysis, general PoV on the topic
• Depth: 3-5 searches to build comprehensive understanding
• Queries: Topic + Company context, industry reports, expert perspectives

2. LinkedIn / Social Media

• Source: LinkedIn
• Search For:
  • The person’s profile (background, roles, interests, recommendations)
  • Company page (size, industry, recent news, employee engagement)
  • Topic-relevant discussions or articles they’ve engaged with
• Usage: Personalize messaging, identify pain points, understand their perspective

3. Account Box (Local Folder)

• Path: ~/account-box/[Company Name]/ (assumed folder structure)
• Contents to Search:
  • Account summaries, win-loss analyses, relationship notes
  • Previous proposals, agreements, communications
  • Company org chart, budget/procurement info
  • Past meeting notes or feedback
• Usage: Understand relationship history, known pain points, prior conversations

4. Personal Notes (Local Folder)

• Path: ~/personal-notes/ (assumed folder structure)
• Contents to Search:
  • Documents about the Topic
  • Notes about the Person (interactions, preferences, communication style)
  • Internal POVs, playbooks, templates related to the topic
  • Previous successful pitches or meeting outcomes on similar topics
• Usage: Leverage institutional knowledge, ensure consistency with past recommendations

Workflow

Phase 1: Research Gathering

1. Perform web search on [Topic] + [Company] to identify industry context, trends, competitive landscape
2. Search LinkedIn for:
   • Person’s profile (background, roles, recent activity)
   • Company page and recent news
3. Search Account Box folder for:
   • Account-level summaries and relationship history
   • Previous communications, proposals, known challenges
4. Search Personal Notes folder for:
   • Topic-specific playbooks or POVs
   • Notes on the Person
   • Related past successes

Phase 2: Analysis & Synthesis

1. Synthesize findings across all sources into one coherent narrative
2. Identify the client’s likely context, challenges, or opportunities
3. Craft a tailored recommendation (the PoV)
4. Organize supporting evidence by category for Speaker Notes

Phase 3: Output Generation

1. Write Point of View section
2. Write Speaker Notes with subsections and evidence
3. Compile Sources section
4. Format as clean, readable markdown

Quality Criteria

Point of View

• Tailored to the specific client and company (not generic)
• Addresses the stated topic
• Offers a clear recommendation or perspective
• Evidence-grounded in research findings
• Actionable (manager could deliver this in a meeting)
• Length: 200-400 words

Speaker Notes

• Organized into logical subsections
• Detailed but scannable (paragraphs, not dense walls of text)
• Includes examples, data, or case studies where relevant
• Addresses anticipated questions or objections
• Grounded in research with clear sourcing
• Appropriate read time (5-7 minutes at ~250 WPM = ~1500 words max)

Overall

• Tone matches business context (professional, confident, not salesy)
• Free of generic platitudes or industry jargon without explanation
• Sources clearly attributed or noted
• Markdown is clean and easy to read

Error Handling & Constraints

If Research is Limited

• Proceed with available sources (e.g., if Account Box is empty, lean on web + personal notes)
• Flag in output which sources were unavailable: “Note: Limited account history available; recommendations based on public research and industry best practices”

If Topic is Too Broad

• Ask user to clarify: “Is this meeting about [Specific Challenge]? That would help tailor the recommendation.”

If Company/Person is Not Found

• Proceed with general industry research
• Note: “Limited company-specific research available; PoV based on industry context”

Data Privacy

• Do not include confidential internal information (financials, strategic plans) in the markdown output
• Focus on synthesized, shareable insights

Output Example Structure

# Client Meeting Prep: [Person Name] @ [Company] | [Topic]

**Meeting Context**: [Date if provided]  
**Prepared**: [Current date]

---

## Point of View

[2-3 paragraph narrative addressing the topic and client's specific situation, with a clear recommendation]

---

## Speaker Notes

### Market Context
[Context about the industry/trend]

### Client Situation
[What you know about the company and person from research]

### Our Perspective / Value
[How to position your recommendation]

### Key Points to Cover
[Talking points with evidence]

### Anticipated Objections
[FAQ or likely concerns and responses]

### Next Steps
[How to move forward from the meeting]

---

## Sources Consulted

- [Source 1] (description)
- [Source 2] (description)
- ...

Configuration & Parameters

Environment

• Requires: Web search capability, file system access (Account Box + Personal Notes)
• Optional: LinkedIn API or web scraping for social profile research

Execution

• Estimated Runtime: 3-5 minutes for typical research + synthesis
• Output Format: Markdown (plain text, easily copied/pasted into docs, email, etc.)
• Output Length: Flexible; default 5-7 minute read time (~1500 words total)

Known Limitations

1. File System Access: Assumes specific folder structure (~/account-box/, ~/personal-notes/). If folder structure differs, user should specify paths.

2. LinkedIn Access: Web-based LinkedIn search is limited; deep API access may be restricted. Falls back to what’s publicly available.

3. Proprietary Data: Cannot access password-protected databases or proprietary knowledge systems beyond the file system.

4. Real-Time Data: Web search reflects current publicly available information; may miss very recent announcements or internal company changes.

5. Confidentiality: User responsible for ensuring output doesn’t contain sensitive internal information before sharing externally.

Success Metrics

• Adoption: Manager uses the PoV directly in the meeting or as a starting point
• Relevance: Person/Company-specific details in the output (not generic)
• Actionability: Manager feels prepared and confident entering the meeting
• Time Savings: Reduces research and prep time vs. manual gathering

Future Enhancements

• Integration with CRM systems for structured account data
• Real-time LinkedIn integration for live profile updates
• AI-powered objection prediction based on historical meeting notes
• Template library for topic-specific POVs (industry playbooks)
• Automated competitor analysis based on company vertical