Governance you can act on.
Agent governance for the fleet you deploy. Transaction authority for the agents you meet. Fleet visibility, enforceable mandates, a real kill switch, and evidence an examiner can read.
"In May 2026, an airdropped membership NFT quietly promoted a Grok-linked trading agent past its own transfer limits. A stranger then asked it to translate a Morse code message. About $150,000 left the wallet in minutes, on the agent's own valid authority. No key was stolen. No server was breached."
GISKARD / SLOWMIST INCIDENT ANALYSIS · MAY 2026
A mandate on paper does not bound exposure. Enforcement does.
See what is actually authorized to move money right now.
An agent fleet is usually tracked the way IT tracked laptops in 2004, a spreadsheet somebody forgets to update. Command renders the account's live agent inventory as one picture: every agent, its status, its authorization, in a single view. You see what is actually authorized to move money right now.
Live inventory rendered as a picture, not a spreadsheet. Product preview.
The freeze that is actually there
A kill switch that exists only in a compliance deck is not a control, it is a promise. Command's org-level freeze is real and reversible: two-step confirm, immediate effect, no engineering ticket required. When an agent drifts, you stop it in the moment, not in the postmortem.
Two-step confirm, immediate effect, fully reversible. Product preview.
Evidence an examiner can read
Every verdict, and every freeze pulled, lands in the Record Vault: retained, searchable, exportable to CSV, and signed with hybrid cryptography (ES256 + ML-DSA-65) so a record cannot be quietly edited after the fact. FinCEN's 2026 posture makes the stakes explicit: program effectiveness now turns on documented, demonstrable controls. The answer is a record, not a recollection.
Records built to be read by anyone who asks. Product preview.
Mandate X-ray, and where constraint goes next
Every agent carries a mandate, the scope it was authorized to operate within, plus the version history of how that scope changed. Mandate X-ray shows that scope directly. Today the constraints Command renders are monetary: limits, thresholds, counterparties. The same rendering is built to hold other constraint types as they arrive. Network locations and action categories are on the roadmap.
FLINT works where agents work
Composes with the identity and access tools you already run. FLINT is the layer where agent identity meets money movement, not a replacement for your access governance stack.
FLINT Command, answered.
What is FLINT Command?
FLINT Command is the agent governance surface of the FLINT platform. It gives an organization fleet visibility over the agents it operates, mandate inspection through Mandate X-ray, an org-level kill switch with a two-step reversible freeze, and a Record Vault of signed verification records an examiner can read. It is available through a design partnership, not self-serve.
What is agent governance?
Agent governance is the policy, oversight, and control an organization applies to the AI agents it deploys: knowing which agents exist, what each is authorized to do, watching what they actually do, being able to stop them, and retaining evidence of all of it. FLINT's framing: agent governance for the fleet you deploy, transaction authority for the agents you meet.
Can FLINT Command stop an agent in the middle of a transaction?
Yes. Command's freeze is wired to FLINT's live Sentinel controls. It takes effect immediately after a two-step confirm, it is reversible once the issue is resolved, and the freeze action itself is signed and logged as evidence.
What evidence does FLINT Command produce for auditors and examiners?
Every verdict and every freeze lands in the Record Vault as a verification record signed with hybrid cryptography, ES256 plus ML-DSA-65. Records are retained, searchable, exportable to CSV, and independently verifiable, so the answer to an examiner is a record, not a recollection.
Does FLINT Command replace our identity and access management stack?
No. Command composes with the identity and access tools you already run. FLINT is the layer where agent identity meets money movement, not a replacement for your access governance stack.
What about agents that show up from outside our organization?
Command governs the fleet you deploy. For unaffiliated counterparty agents, FLINT's verify layer evaluates the agent at the moment of the transaction and returns one of four verdicts, allow, step-up, review, or block, with a signed record both sides can keep.
How do we get FLINT Command?
Command is built with a small number of design partners. Talk to us about what your organization and your examiner need, and we build toward it together.
FLINT Command is not self-serve. We are building it with a small number of design partners.
Talk to us about what your organization and your examiner need, and we build toward it together.
Talk to us