Hi friends,
I am excited to share a new release from our open-source project, Omen.
What Is Omen?
Omen is an open-source strategic reasoning and phenomenon simulation engine.
It is designed to help founders, product teams, and strategy practitioners model complex competition with interpretable outputs, instead of relying on opaque conclusions.
GitHub: https://github.com/StrategyLogic/omen
What We Shipped: Founder App
This release introduces the Founder App (deployed on Streamlit Community).
Its mission is very specific:
- Build a founder as a Strategic Actor
- Extract decision style and strategic signals
- Provide structured artifacts for downstream strategic simulation
In other words, we model the decision-maker first, then run broader strategic replay on top of that foundation.
Why This Matters
Most strategy tools jump directly to market-level outputs.
We found that many key outcomes are highly sensitive to founder cognition, constraints, and decision preferences.
If the actor model is weak, simulation quality drops quickly.
The Founder App is our attempt to make this layer explicit, inspectable, and reusable.
Core Capabilities in the App
1. Founder Persona
- Founder narrative
- Key traits and evidence-backed profile
- Consistency-oriented signals for interpretation
2. Strategic Chain
- Formation analysis
- Why-analysis
- Combined strategic path from perception to action
3. Insight and Gap Analysis
- Process gaps
- Outcome gaps
- Strategy-vs-reality divergence signals
4. Visual Graphs and Timeline
- Founder influence graph
- Timeline cards with event context and strategic markers
- Step-oriented analysis workflow for better observability
Metrics and Artifacts
The app outputs structured JSON artifacts under output/case_replay/<doc-id>/, including:
strategy_ontology.jsonfounder_ontology.jsonanalyze_status.jsonanalyze_formation.jsonanalyze_why.jsonanalyze_insight.json
This keeps results reproducible and easy to integrate with later stages.
Multilingual UI
The Founder App currently supports Chinese and English switching in the UI.
This helps us collaborate across mixed-language teams and community contributors.
Runtime Constraints (Current Phase)
A few transparent notes:
- The app is LLM-powered.
- Each founder case currently runs in a single pass.
- The app is still evolving; after restart/redeploy, run records in session state may be reset.
We prefer explicit constraints over hidden assumptions.
CLI Companion (New Shape)
We also aligned founder analysis CLI into a dedicated path:
-
omen analyze founder --doc <name>for full flow omen analyze founder persona --doc <name>omen analyze founder strategy --doc <name>omen analyze founder insight --doc <name>
--doc resolves from cases/founder by default (with .md inferred).
Tech Stack
- Python 3.12+
- Streamlit
- Typed JSON artifacts + ontology-style modeling
- LLM-based analysis pipeline
- Modular CLI workflows for reproducible runs
What Is Next
We are iterating on:
- Better persistence for analysis records
- Richer strategic actor abstraction
- Cleaner bridge from founder analysis to multi-actor simulation
If this direction resonates with you, I would love your feedback.
Issues, critiques, and ideas are all welcome.
Thanks for reading, and hope you enjoy trying Omen.
Chen Jiaxing (Janus)
Top comments (0)