Find access that doesn't make sense.
1Commando maps every employee, service account, and AI agent to what it can reach, scores that access against similar roles, and flags anything irregular for review.
1CommandoTrailby T. Brooks
from Okta
env: prod
Sarah reaches GitHub prod-deploy through eng-oncall, a group she joined while she was in Engineering. None of her 4 accounting peers hold it and it has not been used since November 2022. Recommend removal.
What a review looks like in 1Commando.
A working recreation of app.1commando.ai running on Marlow Industrial, a demo company with 2,324 identities on its graph: 2,253 current employees, contractors, service accounts, and AI agents, plus 71 offboarded accounts. Create an account and you land in the same demo.
Access piles up.
People change jobs and keep what they had. Nobody goes back to take it away.
Reviews are rubber stamps.
A quarterly spreadsheet with a row for every grant gets approved in an afternoon.
Many machines have no owner.
Service accounts, API keys, and AI agents carry standing access. When no owner is on record, nobody is asked to review it.
Everything sits on one knowledge graph.
We read your Okta directory over its API and import admin exports from your HR system, cloud, code, and data platforms. Every identity, group, role, permission, and resource becomes a node. Every grant becomes an edge that records who granted it and when, wherever the source system keeps that.
Because every identity sits on the same graph, access that looks unlike its neighbors stands out.
Every identity gets a score per domain.
A high score means access in that domain is expected for this identity. A grant is flagged when its score is low for how sensitive the resource is, or when few peers hold that same resource. The score rates the fit between a permission and a role, and says nothing about how the person performs.
When a reviewer approves a flagged grant, the decision and any reason given are recorded, and the flag stays closed for one year. After that it comes back for review.
Read about the Clearance ScorePick an identity and a resource, and Commando draws the path.
The trail page starts from one identity, lists only what it can reach, and draws the path to the resource you pick.
Pick any account and see what a compromised version reaches.
Each ring is a hop count from the account: one hop for its direct grants and groups, two for what those groups carry, and so on. Restricted resources are outlined in red. It works the same for a person, a service account, or an agent.
Every grant on one timeline.
Each grant will appear on the date it was made, with job changes marked. Access that stayed after a job change will be drawn in red. The grant dates and job history it needs are already on the graph.
Agents and service accounts are scored and flagged like people.
Service accounts, AWS IAM users, Slack bots and apps, GitHub App installations, Google OAuth apps, and AI agents live on the graph with their owner, their reach, and their last use. A machine with no owner on record and a restricted resource in reach is flagged the same way a person would be.
Owner: none on record
Last used: 2026-09-21
Owner: Nicholas Bradley
Last used: 2026-09-21
Owner: Maya McCarthy
Last used: 2026-09-22
Owner: Travis Matthews
Last used: 2026-09-22
Owner: none on record
Last used: 2026-03-02
Commando will answer questions about the graph in plain English.
Each answer will name the identities and groups involved, with a link to each one's trail in the graph.
Quarterly access reviews built from the graph.
Each review lists flagged grants first. Reviewers see each outlier with its trail and peer comparison, then approve it for one year or request its removal, with an optional reason. A removal request stays open until the access is gone from the source system. The review and the decision log export as CSV for auditors.
1Commando only reads from your systems.
1Commando needs to read a lot of your company to build the graph. This is how that access works.
Every connector only reads. When a reviewer requests a removal, your team makes the change in the source system, and the next import shows it.
The Okta connector uses a read-only API token, encrypted at rest, that you can revoke in Okta at any time. Other sources come from admin exports you upload.
Everything is served over HTTPS, and connector tokens are encrypted at rest. Hosting, retention, and deletion details are available on request.
Security documentation and our current audit status are available on request.
See your own access graph.
Start with Okta or Google Workspace, plus a Workday export. 1Commando only reads from them.