Stacks/Coding agent
Lovable's app-building agent
Lovable's co-founder describes rebuilding its agent several times: from an autonomous subagent design back to one predictable tool chain, from planned fine-tuning to retrieved 'knowledge files', and from full rebuilds to stateful sandboxes with incremental type checks. Its Camille framework now forks and merges whole running computers, app and database included.
- Use case
- Chat-driven creation, hosting, and iteration of web apps for people who can't code; users have created more than 70 million projects
- Category
- Coding agent
- Team size
- not stated
- Orchestrator / framework
- After an early async design with task creation and subagents got stuck, Lovable went back to one user request driving a predictable chain of tools with chat and a live preview. By late 2024 the main editing flow was a single response with file changes, parsed after the agent finished, with retries and error handling but no open-ended loop. It later added stateful sandboxes and Camille, a framework for forking and merging entire running computers.
- Memory
- 'Knowledge files': short notes engineers wrote from chat logs of broken apps, selected by a retrieval step and added to the model's context (RAG instead of fine-tuning). A later system retrieved problems and solutions for the agent; an early version cut the rate of users getting stuck by 5%.
- Evals
- not stated
- Observability
- not stated
- Cost (stated in writeup)
- Not stated
- Cost per output (stated)
- Not stated in the writeup
- WhatWeRun estimate
WhatWeRun estimate
Not enough info to estimate
The post doesn't name the coding model or give tokens or volume per edit.
Reviewed Nov 13, 2026
- Observability tools
- not stated
- Other tools
- Stateful sandboxes that checkpoint memory, CPU state, and diskPersistent dev servers for live previewIncremental TypeScript checksCamille (fork and merge running computers)
- Source
- https://lovable.dev/blog/how-we-built-lovable-and-why-we-keep-rebuilding-it
Anton Osika · Published Nov 4, 2026 · Last verified Nov 13, 2026
Related
More like this
corezoid.comCorezoid's ticket-routing coding agent
A coding agent triggered by Jira labels that fixes tickets and opens merge requests, with a digital twin of the engineering org deciding which repository, process, or service each ticket belongs to
Stated: ~$2.50 per code change+1
aboutamazon.comThe agent teams that build AWS Emulator
Autonomous agents that write, test, and fix every line of AWS Emulator, a downloadable local emulator of AWS services, with no human-written or human-reviewed code
blog.duolingo.comDuolingo AI STR Fix bug agent
Investigating and fixing Shake-to-Report bug tickets: gathering context (relevant code, tracking events, experiment states, recent PRs, repro steps) and proposing a fix for engineers to verify