Decide what agents can do before they do it
Set the rules once: which models, which data, which tools, and which actions need a person. Methos enforces them on every run.
GUARDRAILS
Rules that live outside the prompt
Prompts can be argued with. Methos keeps guardrails in the platform, so an agent can't talk its way past a limit you've set.
CONTROL MODES
Four modes. Move up as trust is earned.
COMING SOONShadow
Works alongside your team and records what it would have done.
Use when: proving value before anyone relies on it.
Shadow
Works alongside your team and records what it would have done.
Use when: proving value before anyone relies on it.
Suggest
Drafts the answer or action for a person to use.
Use when: early rollout, high-judgment work.
Approve
Prepares the action and waits for sign-off.
Use when: actions that change records or reach customers.
Autonomous
Acts on its own within its rules.
Use when: proven, low-risk, high-volume work.
Move each workflow up one mode at a time, as the record earns your trust. Approval steps in workflows are available today.
ROLES AND PERMISSIONS
The right people hold the right keys
Owners, admins and members each see and change only what their role allows. Admins decide who can build agents, who can publish workflows and who can see the record.
MODEL GOVERNANCE
An approved model list, enforced
Choose which AI models your company permits. Agents can only run on models from that list, and you can add new ones as they are approved.
USAGE AND COST CONTROLS
Spend you can see and cap
Track model usage by agent and team, and set limits so a runaway workflow can't run up a bill.