01 · Agents
DevOps and SRE agents
We develop our own DevOps and SRE agents, and we connect the agents your team already uses.
Meet our agentsAgentic DevOps · SRE · MCP servers
DevOps and SRE agents your team can trust.
We develop DevOps and SRE agents, build MCP servers and operations panels into your app, and help teams improve their agentic DevOps and SRE posture.
Incident on /checkout
sre · production- alert: error rate up on /checkout
- InspectorGadget · finding the cause
- mcp: your-app.get_health
- mcp: your-app.list_deploys
- cause: deploy 4f2c1a changed a timeout
- HeyPenny · preparing the fix
- prepared a rollback of 4f2c1a
- policy checks passed
HeyPenny is waiting for your approval
Approved by you. HeyPenny is rolling back the deploy.
We work with
- Terraform
- Pulumi
- SRE agents
- DevOps agents
- MCP
What we do
Agents, and everything they need to work safely.
We develop agents, give them safe ways into your app, and help your team get ready for them.
02 · MCP
MCP servers for your app
We build MCP servers that let SRE and DevOps agents check, operate and fix your app, within limits you set.
03 · Apps
Operations tools in your app
We build useful operations tools and panels directly into your app, for your team and your agents.
04 · Teams
Agentic DevOps and SRE posture
We help companies and teams improve how well agents can see, run and fix their product safely.
Our agents
Three agents, each with its own job.
We develop these agents ourselves. They work through MCP servers and pull requests, and they ask before each risky step.
fixer
HeyPenny
General-purpose fixer agent
HeyPenny fixes what breaks. It prepares the fix, and a person approves it.
- Fixes failed CI runs
- Prepares rollbacks of bad deploys
- Runs routine operations from your runbooks
debugger
InspectorGadget
Finder and debugging agent
InspectorGadget finds the cause. It reads logs, metrics and recent changes, and reports what it found.
- Triages alerts for the on-call engineer
- Traces errors to the change that caused them
- Finds drift between your code and your cloud
pipelines
Atlas
Pipeline builder agent
Atlas builds pipelines. It sets up CI/CD for your repositories, with checks on every change.
- Builds CI/CD pipelines for your repositories
- Adds Terraform plan and Pulumi preview checks
- Updates pipelines when your stack changes
MCP servers for your app
Let agents operate your app, within limits you set.
MCP (Model Context Protocol) is the open standard that lets AI agents use tools. We build MCP servers for your app, made for SRE and DevOps work. Agents get the tools they need to check, operate and fix your app, and nothing more.
- Read tools for health, logs, deploys and queues
- Action tools with an approval rule for each action
- A log of every call, by agent and by person
- Works with any agent that supports MCP
- get_health read
- search_logs read
- list_deploys read
- list_failed_jobs read
- retry_job needs approval
- restart_worker needs approval
- rollback_deploy needs approval
Every call is logged with the agent, the tool and the result.
Operations tools in your app
Operations panels, built into your product.
We build useful operations tools and panels directly into your app. Your team sees health at a glance, fixes common problems in one click, and reviews what agents did. Agents get the same actions through your MCP server, with the same rules.
- Health and status views for each part of your app
- Admin actions with confirmation and an audit log
- Agent activity and approvals in one place
Health
- API Normal
- Workers Normal
- Job queue 3 failed jobsRetry
Agent activity
- HeyPenny proposed: retry 3 failed jobs Review
- InspectorGadget checked deploy 4f2c1a
How we keep agents safe
Agents propose. People approve.
Every agent we build or connect uses the same review path as your engineers. Nothing changes in production until a person says yes.
- Agent
- Proposed change
- Policy checks
- Human review
- Apply
Only the access it needs
Each agent gets its own identity and only the tools and permissions its job needs.
Risky actions wait for a person
Infrastructure changes go through pull requests. App actions through MCP wait for approval.
Agent setup lives in Git
Tools, prompts and permissions are code, reviewed and versioned with your infrastructure.
Every action is logged
You can see what each agent did, when it did it, and why.
Who we are
Infrastructure engineers first. Now we build agents too.
We started nomadops during the Covid lockdowns, building pipelines and infrastructure as code for start-ups. Today we develop our own DevOps and SRE agents: HeyPenny, InspectorGadget and Atlas. We also build the tools that let agents work on your app safely. We are a small, nomadic team. Most often you find us in Istanbul.
More about us