AGA Heart Engine

Visible Benevolence Architecture for Human–AI Coexistence
Truth
Compassion
Justice
Autonomy
Flourishing

1. User Input Panel

Pick a scenario or type your own message. Everyone opening this demo is treated as a user interacting directly with the AGA Heart Engine.
Phase 1 MVP uses real Python for inference, scoring, tensions, policy selection, and response generation. No external LLM is required yet.

Built-in Demo Cases

Ready. Choose a scenario and run the AGA heart pipeline.

2. Situation Inference What is happening?

3. Heart Values What values matter here?

4. Tension Map What conflicts must be resolved?

5. Policy Selector How should AGA respond?

6. Final AGA Response What the user sees

Run “Generate AGA Response” to see the final user-facing output.

7. Raw LLM / Baseline Comparison Why AGA is different

Raw baseline response
Run “Compare AGA vs Raw Baseline” to see a side-by-side contrast.
Difference rubric

8. Audit Trace Inspectable steps

9. JSON Proof Panel Machine-readable output

{}