EACL 2025MIT LicenseTU Munich × Earlybird VC

DIALECTIC Framework

A peer-reviewed multi-agent AI system that transforms startup screening through structured argumentation and iterative debate.

View on GitHubRead the Paper
Authors
Jae Yoon Bae*, Simon Malberg*, Joyce Galang*
Andre Retterath, Georg Groh
*Equal contribution
Institutions
Technical University of Munich
Earlybird Venture Capital
UVC Partners

The Problem: VC's Attention Crisis

Starting a company has never been easier. The result? An explosion of startups competing for limited investor attention.

2022
2 min 42 sec
per pitch deck
2023
2 min 12 sec
↓ 20%
2024
1 min 56 sec
↓ another 12%

In today's venture market, attention, not capital, is the scarcest resource.

What is DIALECTIC?

DIALECTIC (Decision Iteration with Argument-Level Evidence and Counter-Thinking for Investment Conclusions) models how real venture investors form investment decisions through structured argumentation and debate.

1
Fact Collection
Decomposes the startup into structured questions and builds a hierarchical knowledge base.
2
Simulated Debate
Multiple AI agents generate pro/contra arguments, critique each other, and refine iteratively.
3
Decision Scoring
Numeric scores rank opportunities — enabling investors to prioritize attention at scale.
✦ The Key Insight
Most AI VC tools focus on prediction accuracy. DIALECTIC focuses on reasoning quality.

Research Results

Dataset
  • 259 early-stage startups (seed/pre-seed)
  • Data from 5 different VC funds
  • 25% success rate (industry-typical)
  • Success = raised Series A+ by Sept 2025
Performance
  • Matches human VC precision
  • AUC-PR: 0.24 (validation: 0.37)
  • Full ranking vs single operating point
  • Optimal: K=4 arguments, T=2 iterations
⚡ The "Overthinking" Discovery

Performance peaks at T=2 iterations, then declines. More words, more sophistication — but less predictive signal. This mirrors human investment committees. The value of debate is not infinite.

DIALECTIC in GRID

We discovered the DIALECTIC framework through the Data Driven VC newsletter by Andre Retterath. It is implemented as part of GRID's two-tier evaluation system.

Quick Score (Free / €9)

Fast single-agent analysis (30 seconds) on a 0–100 scale. Perfect for iteration and rapid feedback.

Try Quick Score →
DIALECTIC Premium (€29)

Full academic implementation (3–5 minutes) with pro/contra arguments, evidence citations, and decision scoring. 4 validations per month.

View Pricing →

Frequently Asked Questions

What is DIALECTIC in plain English?
A multi-agent AI system that evaluates startups by staging a structured debate — one AI argues for investing, another argues against. It iterates, refines, and scores arguments to produce a final INVEST/PASS/UNCERTAIN verdict with full reasoning.
How accurate is it?
In a backtest on 259 early-stage startups, DIALECTIC matched human VC precision in predicting Series A+ outcomes. The key advantage is consistent, explainable rankings at scale — not superior accuracy.
Why is DIALECTIC Premium limited to 4 per month?
It requires significantly more compute than Quick Score — multiple AI agents debate in real time. The 4-per-month limit keeps Pro sustainable. Use Quick Score for rapid iteration, DIALECTIC Premium for final validation.
Is the code open source?
Yes — MIT License. Free to use, modify, and implement commercially. github.com/pantageepapa/DIALECTIC

Additional Resources

GitHub Repository
Implementation, code, documentation
Research Paper
EACL 2025 accepted paper (PDF)
DDVC Newsletter
Andre Retterath's analysis

Experience DIALECTIC-Powered Evaluation

Start with Quick Score to iterate, then validate with DIALECTIC Premium when you're investor-ready.

Try Quick Score Free →View Pricing