We raised our first round.
From ourselves.

Our first investors are the people writing the code. LangSearch is funded by our own team, with our own money and time.

We’re putting those resources behind a question we think deserves a serious answer: what does an AI system need from the web? This is the connection we want for our own agents, and we want other developers to be able to build with it too.

The model is not the whole system.

A capable model can reason, write and plan. But it still needs information to reason about. A new document, a changed product, a fact outside its training data: the world doesn’t wait for the next model release.

Search is one of the connections between a model and that moving world. Done well, it gives an application useful context and sources people can inspect. Done poorly, it gives the model more text without making its answer any better.

We want to build that connection well.

The World Engine for AGI starts with a useful API.

“The World Engine for AGI” is our long-term ambition. It describes the role we want to play: helping AI systems find and use information about the world.

Our first step is more concrete. LangSearch offers a Web Search API with structured results, source URLs and snippets or full webpage text. Developers can choose a result count, a text length limit, dates and domains, then decide how to use the response in their own application.

We’re focusing on Web Search. We believe doing that job well is a better starting point than presenting a long list of products.

We’re funding the connection we want our own AI systems to have.

Why make it free?

Developers should be able to test whether search improves their application without first decoding a list of paid feature switches.

Our supported Web Search features are free. Full webpage text is included. Requesting more results within the supported range does not add a per-result fee. Our Free Plan includes a daily free quota, with all supported search features included.

Self-funding gives us a way to start building around that choice. It doesn’t make infrastructure cost nothing. It means we’ve chosen to put our own resources behind this work.

Conviction is a start. Evidence is the work.

Search quality is about what happens in your application: whether a result is relevant, whether the context is usable, and how long your agent waits for it.

We look at accuracy and latency together. Our homepage comparisonsgive you a starting point; your own queries and workflows are the next test. Follow the sources, inspect the context and measure whether it helps your application.

Our first milestone is what you build.

This announcement doesn’t end with a valuation. It ends with an invitation.

If you’re building an agent, a research tool or an application that needs information from the web, try LangSearch. Follow the sources. Inspect the context. See whether it helps your application do something useful.

That is the return we’re working toward first.