Works with any Linux server and AWS
DevOps that
never sleeps.
Your servers and AWS
Linux boxes and cloud infra, managed in one place.
One DevOps agent
Ask in plain English. It SSHes in, reads the logs, and checks AWS.
Cheap enough to leave on
Awake at 3 a.m. Start free, then $19 a month for Pro.
Your passwords and keys stay encrypted on our server. The AI sees command output, not your credentials.
No card. Two servers free. Nothing to install on your box.
Disk is filling up on prod. Find what’s safe to delete.
$ df -h /
Root disk is at 94% (47 GB of 50 GB). Checking the usual suspects.
$ du -sh /var/lib/docker /var/log /var/www/releases
- Unused Docker images: 14.2 GB
- Rotated nginx logs older than 30 days: 6.1 GB
- App releases in use: keep.
Needs your approval
docker image prune -a --force
Removes images no container uses. Running containers are untouched.
Freed 18.4 GB. Root disk is now at 52%. Both commands are in the Activity log.
How it works
Connect, ask, approve.
Three steps, no agent to install, no YAML to learn.
- Step 1
Connect
Add a server with an SSH user and a password or key, or connect AWS with IAM keys. We test the connection before saving. Secrets are encrypted and never leave our server.
- Step 2
Ask
Type what’s wrong, or open a tab: Services, Logs, Docker, PM2, Nginx, Packages, Users, Scheduled. Everything runs in the browser. Nothing to install on the box.
- Step 3
Approve
Read mode can only look. Execute mode shows the exact command before it runs. Destructive commands always ask. Every action lands in the Activity log.
Features
Everything you would SSH in to do. In one place.
Built for the box you already have, not for a fleet you are about to buy.
Browser terminal and file manager
No PEM files on laptops. Open a shell, edit .env, upload, download, zip, chmod. Ask the AI sidebar while you work.
One tab per job
Services, Logs, Docker, PM2, Nginx, Packages, Users, Scheduled. Restart, follow logs, prune, upgrade, add a key, fix a cron job.
Uptime and health checks
HTTP monitors every 1 to 10 minutes. Every server checked every 15 minutes: disk, memory, load, failed services, pending reboots.
AWS, explained
EC2, S3, IAM, CloudWatch, cost and FinOps recommendations, GuardDuty, Inspector. Ask why the bill jumped and get an answer, not a chart.
Activity log
Every command, who ran it, when, and whether it worked. Tabs, terminal and bot alike. Cannot be edited. Kept for a year.
Reports on demand
Ask for a PDF: what changed this week, what is pending, what to fix next. Send it to a client or a manager.
Who it is for
People who have a server, not a DevOps team.
Your developer left
You own one to three VPS. The site is down and you cannot read a log. A freelancer costs $25 to $80 an hour and needs the setup explained again.
You do DevOps by default
Backend or full-stack, no DevOps hire. Infra is 15% of the job and 100% of the stress. You already have EC2, Lightsail or a VPS.
Small shop with client servers
One person knows the servers. Juniors need to see logs without root. Clients ask what was done and when.
Pricing
Start free. Upgrade to Pro for $19 a month.
Every account starts on Free, with no card. Upgrade to Pro any time in Settings → Subscription, and cancel any time. For a team, or to run it on your own server, contact us.
Free
Your servers, your terminal, the bot with a small allowance. Three uptime monitors.
$0forever
- 1 project
- 2 SSH-connected servers
- 1 seat
- $2 AI credit · fast AI model
- 3 uptime monitors · health check every 15 minutes
- 1 managed database
- 1 auto database backup, newest 3 kept
- 100 MB file storage
- Activity log for 7 days
- Docs support
Pro
One person, unlimited servers. An email when a site goes down.
$19per month
or $150 / year, save $78
- Multi-cloud support (on request)
- Unlimited SSH-connected servers*
- 1 seat
- Unlimited uptime monitors* · 1 / 5 / 10 min
- Health check every 15 minutes, down alerts by email
- $5 AI credit a month, or $60 a year on yearly
- Fast and advanced AI models
- 2 auto database backups, newest 7 kept
- 1 GB file storage
- Activity log for 1 year
- Email support, 2 business days
* Fair use applies.
Start free, upgrade any timeTeam
Everything in Pro, for a team. Unlimited servers, pooled AI credit.
Customfor teams
- Multi-cloud support (on request)
- Unlimited servers*
- Seats for your whole team, with roles
- 100 uptime monitors
- Pooled AI credit · top-up
- Activity log for 1 year + export
- Priority email, 1 business day + 30-min onboarding
* Fair use applies.
Business
Everything in Team, self-hosted on your own server, with your own OpenAI key.
Customannual licence
- Runs inside your own network
- Your own OpenAI key · data stays with you
- Everything in Team
- Multi-cloud support (on request)
- Annual or lifetime licence
- Install and onboarding included
- Priority support
Every tab, the approval gate and the Activity log are on every plan. Limits are the five numbers above. Pro is $19 a month or $150 a year.
Contact us
Talk to a person.
Email support, or message the founder directly on LinkedIn.
FAQ
The questions people actually ask.
I can build this myself using AI. Why would I use you?
+
You can. A weekend of Cursor plus an SSH library gets you a chat that runs commands. Then you still have to encrypt keys, never send them to the model, ask before every change, log who ran what, idle the session, keep a file manager and twelve tabs working, and run uptime checks while you sleep. That is the product. Most people paste a root password into ChatGPT and stop there.
Are my credentials safe on your server?
+
SSH passwords, keys and cloud secrets are encrypted at rest (AES-256-GCM) and decrypted only on our box to open a session. The model never sees them — only command output. Read mode cannot change anything. Execute shows the exact command and waits for Yes. Destructive commands always ask, even with auto-approve. Everything lands in the Activity log and cannot be edited.
What happens if AI credit runs out?
+
The bot pauses. Every tab, the terminal and the file manager keep working. A new Free account starts with $2 of credit on the low-cost gpt-5.6-luna model; Pro gets $5 of credit each month ($60 a year on yearly) and gpt-5.6-terra; it starts over at every renewal. Upgrade any time in Settings → Subscription.
Can I get more credit, or custom pricing?
+
Yes. Email support@devops-agent.io for a free one-time credit upgrade if you hit the wall mid-fix. If you need more than that on a regular basis, we will set custom credit pricing for you. There is no self-serve top-up yet.
Will this break my site?
+
Nothing runs until you press Yes. Silence counts as No. We do not deploy, we do not rewrite your app, and we do not touch the box unless you approve the command. If you are not sure, start in Read mode and only look.
Do I need to install anything on my server?
+
No. It uses SSH, the same way you would. Nothing to install, nothing to update, nothing left behind if you leave.
What if I get locked out of SSH?
+
We are locked out with you. Use your host’s web console (DigitalOcean, Hetzner, AWS, and so on) to reset the password or open port 22, then come back. We cannot bypass a firewall that only allows your laptop.
Which servers and clouds does this work with?
+
Any Linux VPS you can SSH into: DigitalOcean, Hetzner, Hostinger, Contabo, Vultr, Linode, OVH, bare metal. AWS with IAM keys. Google Cloud is coming. Shared hosting, cPanel, Wix, Vercel, Render and Netlify are not a fit — there is no SSH root.
Is there a free plan? How do I pay for Pro?
+
Yes. Every account starts on Free, with no card. When you want more, upgrade to Pro in Settings → Subscription: $19 a month or $150 a year, paid by card through Stripe. Cancel any time in the same place and you keep Pro until the end of the period you paid for.
Who is watching your server at 2 am?
Connect a server in two minutes. Ask it what is wrong. Approve the fix.