Free pilot now open — limited spots. Full access, hands-on onboarding.Apply →
Letco logo
Governed AI development

Your team is already using AI. Put it under control.

This is the half of Letco that has nothing to do with autonomy. It is about running the AI coding tools your developers already use inside infrastructure you own, with a record of everything that happened — without banning the tools or waiting for an enterprise contract.

Where most teams are today

  • Developers are running AI coding tools on their own laptops, with their own accounts.
  • Nobody can say which code or context was sent to which provider, or when.
  • There is no record to hand an auditor, a client, or a security questionnaire.
  • Every per-token API plan makes the bill grow with the amount of work the AI does.

What Letco changes

Every session is recorded

Conversations with the AI are stored encrypted, session by session — not just the final diff, the whole exchange. Each session gets its own encryption key, and that key is wrapped for the organisation as well as for the user who ran it, so the record belongs to the company by design rather than to one developer's account. Individual transcripts can be exported, and all of them can be backed up at once, including scheduled recurring backups into your own S3 bucket.

What it gives you: a reviewable record of how a result came about — for audit, for incident review, and for working out which prompts actually produce good work. It stays with the company when the person who ran it moves on.

Execution moves off the laptops

The AI developer tools — Claude Code, Codex CLI, Gemini CLI, Grok CLI — run inside environments you control instead of on personal machines: Letco Cloud sandboxes, your own local Kubernetes cluster, or your server-side cluster. Each cloud session is its own Kubernetes pod with user-namespace isolation, no privileged access, and default-deny egress: AI providers, git hosting and package registries are reachable, internal networks and cloud metadata endpoints are blocked.

What it gives you: the work happens on company infrastructure instead of on a laptop nobody can see into — and developers still reach it exactly as they would a local machine.

Or run the whole platform yourself

For teams whose requirements do not permit a hosted service, the entire platform deploys on-premise into your own infrastructure. Sessions, stored content and encryption keys stay inside your environment — Letco has no access to them.

What it gives you: the same product, with the trust boundary drawn at your own perimeter.

Certification in progress

We are working toward independent security certification, including ISO 27001 and a SOC 2 audit. The controls are built with those frameworks in mind, and attestations will be published on the security page as they complete.

What it gives you: a straight answer to the certification question — including the part that is not finished yet.

Network topology

Three kinds of runtime, one transport. Whether the AI runs on a laptop, in a Kubernetes pod or in a virtual machine, the wiring is the same — and every runtime opens the connection outwards, so adopting Letco does not mean opening a hole in your network.

Network topology — The whole pictureYOUR NETWORKDeveloper machinesthe AI tool and your repositorySandbox clusterone isolated environment per sessionVirtual computerspersistent Linux and Windows machinesEncrypted channeloutbound TLS onlyLetco platformhosted, or on your own serversThe channel carries messages, it does not store them: an interruption pauses traffic without losing work, becauseruntimes buffer locally and redeliver once the link is back.
The whole pictureOne encrypted channel, three kinds of runtime. Every runtime opens the connection outwards, so nothing in your network has to accept inbound traffic — and the platform itself runs either in Letco Cloud or on your own servers.

Remote Access — moving the work off the laptop does not put it out of reach

Whichever of those runtimes a session lands in — a device, a cloud sandbox or a virtual computer — Remote Access reaches the services running inside it the same way. Every route is mediated by the platform and ownership-checked, links expire and can be revoked, and administrators can switch the whole capability off for a deployment.

Scan and open

Remote Access lists the ports actually listening inside a runtime and opens an HTTP app in your browser, authenticated as you — no firewall change, no tunnel to build by hand.

Share without an account

A guest link lets a colleague or a customer open the running app with no Letco account at all. You set the expiry — an hour, a day, a week — plus an optional password, and you can revoke it at any moment.

Connect privately

SSH and databases go over a private stream instead of a public URL: a copy-paste ssh ProxyCommand or connection hints you run from your own machine. Databases never get a guest link.

Not gated behind an enterprise deal

Recorded sessions, isolated execution and on-premise deployment are normally the part of a product you only reach after signing an enterprise contract. In Letco they are how the platform works, for a five-person team the same as for a division.

Runs on subscriptions, not per-token APIs

Letco drives the AI developer tools on flat subscription plans rather than pay-per-token APIs, so the cost of governing AI work does not scale with how much work the AI does. Usage is still metered per session, so you can see where it went.

Start with control, add autonomy later.

Teams usually begin here — getting AI work onto their own infrastructure with a record of it — and move on to the full planning and verification pipeline once that is in place. It is the same platform either way.