AI writes the code. We make sure it lasts.

AI makes code cheap. Code you can still understand and change in 7 or 17 years is not. Larv IT helps software teams get the speed of AI without losing code that lasts.

Get in touch

Services

  • AI-assisted development

    We help you set up an AI workflow, with your conventions, review agents and hooks. Share it between developers without suffocating them with must-follow practices.

    Claude Code 路 OpenCode 路 plugin marketplaces 路 agentic loops

  • Test automation and CI/CD

    We take you from manual testing to a merge gate: unit, integration and end-to-end tests that start the whole system with seed data and prove it works before anything reaches production.

    Playwright 路 Gitea Runners 路 GitHub Actions 路 Bitbucket Pipelines

  • Technical leadership

    An interim tech lead, architect, CTO or engineering manager while you grow, or while you recruit the permanent one. We can run that recruitment too.

    Architecture 路 team setup 路 recruitment

  • Observability

    Logs, traces and metrics that show what your system does in production, from the server to the mobile app, built on open standards.

    OpenTelemetry 路 Grafana 路 Loki 路 Tempo 路 Prometheus

  • Integrations and messaging

    APIs and message flows between your systems and the outside world, from REST and GraphQL to queues and SMS gateways.

    REST 路 OpenAPI 路 GraphQL 路 RabbitMQ 路 SMPP 路 OAuth2

How we build

Our own work runs through a development pipeline that drives AI coding agents with very little supervision. We are building it to:

  • Prove quality. A delicate army of AI reviewers with different roles looks at every change before it merges. Deterministic tests are always front and center. If it can be tested, it should be tested.
  • Keep code easy to understand. The pipeline measures how hard each project is to read, and works to make that number better with every change.
  • Ask the human only and exactly when it's their decision. Each project's written goals settle the everyday choices. Every answer the human gives adjusts the project's direction, so the same question does not come up again.
  • Spend tokens only where they pay off. Cost goes in where it raises quality, and is cut everywhere else.

Projects

Open source, free to use, MIT licensed, with no runtime dependencies.

  • @larvit/log

    TypeScript 路 npm

    Structured logging with a tiny API, focusing on OpenTelemetry export of logs, spans and traces. Runs the same in Node, Bun, Deno, browsers and React Native.

    Sourcenpm

  • fejkdata

    Go 路 library and CLI

    Locale-aware fake data for tests and fixtures: Swedish, Norwegian and US persons, valid test ids, addresses and more, streamed as JSON, CSV or SQL.

    Sourcepkg.go.dev

  • @larvit/adf-codec

    TypeScript 路 npm

    Converts Atlassian Document Format, the rich text of Jira and Confluence, to and from markdown, with a lossless round-trip. It also writes simpler markdown that GitHub, GitLab and Obsidian render, at the cost of some formatting.

    Sourcenpm

  • @larvit/smpp

    TypeScript 路 npm

    SMPP 3.4 client and server for Node, with keepalive, reconnect, long messages and delivery receipts built in. Tested against Kannel, Jasmin and five other SMPP stacks.

    Sourcenpm

Find all our code on GitHub and on our own Gitea.

Mikael "Lilleman" G枚ransson

About

Larv IT is Mikael "Lilleman" G枚ransson, a tech lead, architect and engineering manager with more than 20 years of building for the web. He has been CTO of an e-commerce start-up, UI architect for a video advertising platform and tech lead for a SaaS business system, in industries from telecom and e-commerce to video on demand.

Larv IT started in 2011 as an open source consultancy, and has turned the general parts of its client work into open source libraries ever since.

We believe the best results come where tech meets people. People are very different, and a team that leans into that, instead of fighting it, gets extraordinary results.