Lamoom

About Lamoom

Lamoom is a marketplace where agent apps run inside the user's own Claude or ChatGPT client, not on Lamoom's servers. Each app is a self-correcting loop that scores its own output and iterates until it meets a quality bar, leaving a trail of reasoning and intermediate files on disk. Anyone can publish a loop and set their own price per run or monthly.

Review

Lamoom launched this week to fill a gap: people build useful agent workflows in Claude that stay locked on their machine, with no way to share or charge for them. The platform lets those builders publish loops as apps, while users install them and run everything locally using their own files, API keys, and data. Lamoom only records that a run happened - it doesn't see the content.

Key Features

• Apps execute locally inside the user's Claude or ChatGPT environment, accessing their own files, data, and connectors. The platform does not view run contents.
• Every app is a loop with judges. The agent scores its own output and continues refining until the score passes a threshold, rather than returning the first draft.
• Runs leave a paper trail on disk: reasoning steps, intermediate data, and final outputs. If a run fails partway, the work done so far is preserved.
• Publishing is open. Publishers retain their loop, set the price per run or as a monthly subscription, and the marketplace lists the app.
• New users get $20 in free credit to test apps without upfront payment.

Pricing and Value

Lamoom provides $20 in free credit upon sign-up. App pricing is set individually by publishers - they can charge per run or offer a monthly subscription. The reference material does not show specific price points or any platform commission structure; those details are not yet defined publicly.

Pros

• Execution stays on the user's machine, so data never leaves their local environment.
• The loop-with-judges structure means outputs have passed an internal quality check, not just a single generation.
• Publishers control their own pricing, which could lead to a range of cost options.
• Compatible with both Claude and ChatGPT, widening the potential user base.

Cons

• The app catalog contains eight apps at launch, all built by the platform's creator, so variety is minimal.
• The judge mechanism uses an LLM that may converge on a confidently wrong answer if the scoring criteria are flawed; publishers lack built-in visibility into how often the judge overrules itself versus rubber-stamping the first pass.
• Not well suited for users who prefer a fully hosted cloud service. Apps require the user to run Claude or ChatGPT locally and have sufficient resources on their own machine.

Lamoom fits developers and tinkerers who already build agent workflows in Claude and want to share or sell them. It's also useful for users comfortable running agent apps on their own machine and who value keeping data local. The small app library and reliance on self-evaluation mean early adopters will likely need to experiment and possibly build their own loops, rather than relying on a large selection of pre-built, tested apps.

Open 'Lamoom' Website