Diego Castronuovo Book a call

AI agents for engineering teams. For teams of 2 to 10 developers, on any cloud with Kubernetes.

AI agents that see all of production and change nothing on their own.

I build agents in your cloud that investigate incidents, review pull requests and write code, with a shared memory and tests that check how they behave. In the end, it is all yours.

Book the free call

A train only goes where there are rails, and it leaves when the signal goes up.

Engraving of a country railway line: a steam train stopped at the semaphore signal, next to the signal box

1

Four pieces. The pilot has one, the project all of them.

What I build

No.PieceWhat it does
1Operations agents

They investigate an incident by reading the cluster, logs, metrics and code, read-only. When a change is needed they write a plan: a person approves it in an app with no AI in it, and a separate service runs it.

2Pull request reviewer

Comments on every pull request.

3Coding agents

A cloud workstation where Claude Code, Codex and similar tools work on their own and hand back a branch, a pull request and a verified preview environment.

4Shared memory

A Git repository of skills, runbooks and decisions, read by the agents and by the developers' own tools. Every addition goes through review, like code.

Every delivery includes evals: automated tests of how the agents behave, run again on every model or prompt change. The agents work with Slack or Microsoft Teams, GitHub or GitLab, and the models you already use.

2

I have built it once already

I designed and built it, almost on my own, for a live-streaming SaaS platform, where a small development team uses it every day: five specialist agents, 19 read-only tool servers.

I don't reuse that code for you: I start from scratch with the same design, on your cloud and your tools.

3

From the call to the project

It starts with a single agent, in your cloud.

  1. 30 minutes

    Call

    You tell me about your infrastructure and what takes up your team's time. I tell you which agent I would start with.

  2. 2–3 weeks

    Pilot

    One agent of your choice, in your cloud, with its evals.

  3. 6–8 weeks

    Project

    The four pieces together. In the end the code, permissions, memory and tests are yours.

  4. Every month

    Ongoing support

    I update the agents, rerun the evals when a model changes and review additions to the memory.

4

Prices in euros.

Pricing

Fixed price, agreed before we start.

ServiceDurationPrice
CallYou tell me about your infrastructure and what takes up your team's time. I tell you which agent I would start with.30 minutes€0
PilotOne agent of your choice among the four, in your cloud, connected to your tools and with its evals. If you go on to the project, I deduct it from the price.2–3 weeks€5,900
ProjectThe four pieces together, in your cloud. In the end the code, permissions, memory and tests are yours.6–8 weeksfrom €18,000
Ongoing supportEvery month I update the agents, rerun the evals when a model changes and review additions to the memory.Monthly, cancel any timefrom €2,000 per month
Managed by meThe same agents, run by me, without installing them in your cloud. Coming soon.Coming soonWaitlist

It starts with the call: 30 minutes, free. Book the free call

5

About me

I'm Diego Castronuovo, a platform engineer with 15 years of experience in cloud systems and application scalability. My job is keeping production systems running: Kubernetes, Terraform, PostgreSQL, release pipelines, monitoring.

If you built your app with AI and it is not ready for real data yet, start with the production readiness audit.

For any other engineering work, by the day or at a fixed price, see freelance platform engineering.

My CV

6

Frequently asked questions

Does my data leave my cloud?

The agents run in your account. Data only goes to the model provider you chose.

Can I give the agents write permissions?

Yes. We start from read-only, with every change approved; from there, you decide.

Do I need Kubernetes?

Yes, on AWS, Google Cloud or Azure.

What do I own at the end?

Everything: code, configuration, memory and tests, in your repositories.

7

Start with the free call.

30 minutes on a video call. You tell me about your infrastructure and what takes up your team's time, and I tell you which agent I would start with.

Book the free call