SOVEREIGN · AGENTIC α-AGI

Authority follows
evidence.

Give the agent a bounded mandate. Inspect what it chose, how the work fits together, and why the evidence supports it. You decide what advances.

PLAN → EXECUTE → VERIFY → REVIEW

One mandate. Three linked missions.
Every approval binds an exact result.

Private local workspace. Unlock with the access code printed in your terminal.

Loading examples…

YOUR LOCAL CONSOLE

Unlock your workspace

No wallet or token balance is needed for local research. Your access code stays in this browser tab’s memory.

02 / EXECUTION & REVIEW

A mandate becomes accountable work.

NOT STARTED

Downstream work uses approved outputs. A rejected result stops this mandate.

Start with a clear boundary.

Choose a scenario or import your own public-input mandate. The agent can calculate and propose; it cannot approve its own work.

One explicit step at a time.

Create a mandate to begin. Execution stops for your review.

03 / ACCOUNTABILITY

Your local signing identity

Identity is a local Ed25519 key. It does not establish ENS ownership or independent validator approval.

Unlock the local workspace to inspect its public key.

A COMPLETE, BOUNDED LOOP

Autonomy with
accountability.

Actual computation

The native runtime performs evolutionary portfolio search, independently checks feasible allocations, builds resource schedules, and verifies quotations against supplied sources. These are planning tools, not measured business returns.

Reviewed memory

Only approved prior allocation and scheduling results can seed later searches. Reuse is recorded in the research trace; it is not proof of general learning or improved future performance.

Explicit authority

Local operator review is distinct from an independent validator quorum. Wallet control, ENS identity, on-chain jobs and $AGIALPHA settlement use separately configured and validated runtime integrations. This console never spends funds.

⬅️ Back to Gallery