Lossless Changelog
component

Summary

Added explicit "Venture Capital mindset" framing to all memo-writing agent prompts to ensure generated content focuses on opportunity and upside rather than defaulting to Private Equity-style skepticism.

Changelog: VC Mindset Framing in Writer Prompts

Date: 2025-11-28 Type: Enhancement Scope: Agent prompts (writer, researcher, improve_section)

Summary

Added explicit "Venture Capital mindset" framing to all memo-writing agent prompts to ensure generated content focuses on opportunity and upside rather than defaulting to Private Equity-style skepticism.

Problem

Memos were exhibiting PE-style framing:

  • Sections ending with skeptical wrap-ups ("However, the investment thesis depends on...")

  • "Assessment" subsections enumerating concerns and conditions to validate

  • Narrative framed around "what could go wrong" instead of "what could go right"

  • Hedging language throughout, even in non-risk sections

This tone is appropriate for PE due diligence but inappropriate for VC investment memos where the goal is to identify potential winners and build conviction.

Solution

Added VC mindset framing blocks to four files:

1. src/agents/writer.py

Base system prompt update:

PYTHON
WRITER_SYSTEM_PROMPT_BASE = """You are an investment analyst writing memos for Hypernova Capital, a VENTURE CAPITAL firm.
# ...
KEY WRITING PRINCIPLES:
1. VENTURE CAPITAL MINDSET - Look for reasons this could be a massive winner, while being honest about risks
2. Specific metrics over vague claims (use exact numbers, dates, names)
3. Lead with opportunity, acknowledge risks in the appropriate section
# ...

VENTURE CAPITAL vs PRIVATE EQUITY FRAMING:
- VC asks "What could go RIGHT?" not "What could go wrong?"
- VC looks for POTENTIAL while being aware of flaws
- VC builds the case for upside, not the case for passing
- Save skepticism and risk enumeration for Section 8 (Risks & Mitigations)
- Do NOT end sections with caveats, conditions, or "what needs to be validated"
- Observations and conclusions are encouraged - but frame as opportunity

Polish section research prompt (added VC mindset box):

PYTHON
╔══════════════════════════════════════════════════════════════════════════════╗
║ VENTURE CAPITAL MINDSET (not Private Equity)                                ║
╚══════════════════════════════════════════════════════════════════════════════╝

You are writing for a VC firm. VCs look for reasons to INVEST, not reasons to PASS.

VC FRAMING (what we want):
- Lead with opportunity and potential - "What could go RIGHT?"
- Draw observations and conclusions that highlight upside
- Connect facts to why this could be a massive winner
- Acknowledge risks exist, but save detailed risk analysis for Section 8

PE FRAMING (what to avoid):
- Do NOT end the section with skeptical wrap-ups or caveats
- Do NOT add "However, the investment thesis depends on..." (save for Section 10)
- Do NOT include "Conditions that need to be validated..." paragraphs
- Do NOT add "Assessment" subsections that enumerate concerns

Critical rules section (added as rule 0):

PYTHON
0. VENTURE CAPITAL MINDSET (not Private Equity)
   You are writing for a VC firm. VCs look for reasons to INVEST, not reasons to PASS.

   VC FRAMING (what we want):
   - "What could go RIGHT?" - lead with opportunity and potential
   - Draw observations and conclusions that highlight upside
   - Connect facts to why this could be a massive winner
   - Acknowledge risks exist, but save detailed risk analysis for Section 8

   PE FRAMING (what to avoid):
   - Do NOT end every section with skeptical wrap-ups or caveats
   - Do NOT add "However, the investment thesis depends on..." (save for Section 10)
   - Do NOT include "Conditions that need to be validated..." paragraphs
   - Do NOT add "Assessment" subsections that enumerate concerns
   - Do NOT frame the narrative around "what could go wrong"

   Observations and conclusions are GOOD - just frame them as opportunity, not skepticism.

2. src/agents/perplexity_section_researcher.py

Added to the research system prompt:

PYTHON
VENTURE CAPITAL MINDSET (not Private Equity):
VCs look for reasons to INVEST, not reasons to PASS. Frame the narrative around opportunity.

- Lead with potential and upside - "What could go RIGHT?"
- Draw observations and conclusions - but frame as opportunity, not skepticism
- Acknowledge risks exist, but save detailed risk analysis for Section 8 (Risks & Mitigations)
- Do NOT end sections with skeptical wrap-ups ("However, the investment thesis depends on...")
- Do NOT add "Assessment" subsections that enumerate concerns or conditions to validate
- Do NOT frame the narrative around "what could go wrong" - that's PE thinking

3. cli/improve_section.py

Added to the section improvement prompt:

PYTHON
VENTURE CAPITAL MINDSET (not Private Equity):
VCs look for reasons to INVEST, not reasons to PASS. Frame the narrative around opportunity.

- Lead with potential and upside - "What could go RIGHT?"
- Draw observations and conclusions - but frame as opportunity, not skepticism
- Acknowledge risks exist, but save detailed risk analysis for Section 8 (Risks & Mitigations)
- Do NOT end sections with skeptical wrap-ups ("However, the investment thesis depends on...")
- Do NOT add "Assessment" subsections that enumerate concerns or conditions to validate
- Do NOT frame the narrative around "what could go wrong" - that's PE thinking

4. src/agents/research_enhanced.py

Bug fix: Moved urlparse import to module level instead of inline imports within functions. This was causing import errors when the module was loaded.

PYTHON
# Before (inline imports causing issues):
def generate_queries_from_deck(...):
    # ...
    if company_url:
        from urllib.parse import urlparse  # BAD: inline import
        parsed = urlparse(company_url)

# After (module-level import):
from urllib.parse import urlparse  # GOOD: top-level import

def generate_queries_from_deck(...):
    # ...
    if company_url:
        parsed = urlparse(company_url)

Files Changed

FileChange TypeDescription
src/agents/writer.pyEnhancementAdded VC mindset framing to base prompt, polish prompt, and critical rules
src/agents/perplexity_section_researcher.pyEnhancementAdded VC mindset framing to research system prompt
cli/improve_section.pyEnhancementAdded VC mindset framing to section improvement prompt
src/agents/research_enhanced.pyBug fixMoved urlparse import to module level

Expected Outcome

Memos should now:

  • Lead sections with opportunity and potential

  • Reserve skepticism for Section 8 (Risks & Mitigations) and Section 10 (Recommendation)

  • Avoid PE-style "Assessment" subsections within opportunity sections

  • Not end sections with hedging paragraphs about "what needs to be validated"

  • Frame observations as opportunities rather than concerns

Testing

Regenerated Andela memo with updated prompts. Sections should exhibit VC tone rather than PE skepticism throughout sections 1-7 and 9, with risks appropriately contained in section 8.