Pick a story
Explore illustrative scenarios for context portability, application-enforced hooks, and advanced AI coordination. These demos are educational simulations, not deployment evidence.
The Driver's Seat
An AI agent sees the situation before choosing
Sit in an AI agent's seat for one small, fixed task. See what it knows and what it doesn't before it acts, then step it through acting, recovering from a failure, and learning from the result.
Gentian's guitar journey
Context travels across platforms
A factory worker sets his constraints once, and three demo music services adapt to them. Change a constraint and watch their cards update.
Campion's corporate training
Work and personal profiles for one person
A senior engineer has work and personal profiles. Step from morning to evening and see how a different persona and constitution change the guidance.
Marta's responsibility journey
Current state shapes the guidance
A project manager weighs her cousin's request for a loan. Her values and her state in the moment shape the guidance she gets.
Relay: agent handoff
Values context travels between agents
Two procurement agents hand off work. Without VCP, the receiving agent is blind to upstream constraints. With VCP, it checks that the sender's constitution is compatible with its own before acting.
Go deeper: niche scenarios
Economic Governance
Fiduciary mandates govern agent spending
Set a spending limit, a risk profile, and blocked categories, then see a simulated check allow, block, or escalate each transaction under VCP/E, the draft Economic Governance layer. The SDK does not implement VCP/E yet.
ExploreRen's energy cooperative
AI agents negotiate with private context
Ren's solar cooperative sends three AI agents into an energy market, where one bids for the co-op against rivals and all three put its sustainability values ahead of pure profit.
ExploreNoor's school district
Values become enforceable hooks
A school coordinator turns "don't do homework for students" from an aspiration into VCP hooks that the AI homework tool checks on every response.
ExploreDr. Hana's diagnostic tool
See how confident each claim is
A rural GP sees how confident the AI is in each claim, so a 92% judgment and a 35% guess no longer look the same.
ExploreRelated projects exploring VCP
Creed Space
Creed Space builds constitutional governance and practical AI safety infrastructure. How deeply it uses VCP, and how that works in operation, is for Creed Space to show; this demo site proves neither.
MillOS
An AI-native organizational operating system that is exploring VCP context for coordinating agents. Claims about its persistence, adaptation, or runtime guarantees need evidence from MillOS itself.