Latere AI

Put Latere's agents to work through our applications, build your own on the Latere Platform, or run the open source cores behind it on your own infrastructure.

Latere is Latin for "to be hidden." What is hidden is human intelligence. In increasingly autonomous systems, human judgment does not disappear. It recedes behind every layer of decision-making, invisible but indispensable. Latere exists to ensure that this hidden human intelligence remains present, remains effective, and is never engineered away, so that superintelligence stays trustworthy and people stay in control.

AI can now do real work: run a workflow from start to finish, turn an idea into a working product, build and ship software. As it grows more capable than the people it works for, the question that matters is whether it can be trusted. Our mission is trustworthy superintelligence. Every system we ship follows one principle: the human stays in the loop. Agents act within the rules people set, record every step, and seek human judgment when it matters.

Latere Chat takes a piece of work in your own words. An agent does it on a computer of its own, shows you each step, and asks before it reaches anything outside. It can come back to a conversation on its own, to check whether something you were waiting for has arrived, and keep work running on a schedule.

Latere Studio is Latere's app for the Mac. A shortcut opens an ask over any app, and what needs you arrives as a notification you can answer in place. When you choose, a conversation works in folders on your own Mac, with every command it runs kept on record.

Both are views of the same conversations on the Latere Platform, and neither holds agent logic of its own.

One sign-in, one console, and one API address across every capability. Each capability except Apps is built on an open source core you can run yourself.

Identity is the sign-in and access every capability below shares: single sign-on through OpenID Connect, with federated login, organizations, teams, and token issuance. One account covers every Latere product, for people and agents alike.

Agents runs hosted agents. Define an agent as a versioned configuration, start sessions with it yourself or from a trigger on a schedule or a push, and follow every step in its log. Each session works in a sandbox of its own or on your own computer, survives the process that runs it, and waits for a person where it should. Check the open source core: Topos.

Oversight learns, from a person's own Allow and Deny answers, which of an agent's actions can run on their own, which should wait for the person, and which should not run. It also measures how often each source of those decisions is wrong: your rules, a model that reviews actions, and Oversight itself. Check the open source core: Limen.

Models puts one gateway in front of your model providers. Issue and revoke access without handing out provider keys, set spend limits, and track usage through one API. Check the open source core: Lux.

Environments is where agents and applications run code: a sandbox the platform runs, your team's machines, or your own computer, each behind its own boundary. Run commands, attach a terminal, and move files in and out, with credentials swapped in at egress so they never enter the workload. Check the open source core: Cella.

Connections lets agents work in the services you already use. Connect a service once, such as a calendar, your mail, or an MCP server, and agents use it without ever holding your credential: each call goes through Connections, which checks it and adds the credential on the way out. Check the open source core: Pons.

Repos hosts Git repositories over HTTPS and SSH. Give each project, sandbox, or agent run its own Git remote, with S3 compatible storage as the source of truth. Check the open source core: Origo, and origo-web for browsing its repositories.

Storage keeps files and workspaces durable for people, agents, and sandboxes, with version history and permissioned sharing over S3 compatible storage and Postgres. Check the open source core: Arca.

Apps hosts web applications from a repository. Push to build a preview, tag a version to release it, and reach each app at its own address on latere.site. A new app is private until you open it to others, and apps their owners list appear in the gallery. The contract its clients share is public in apps.

Parsing turns a file into pages, blocks, tables, and the fields of a schema you supply, each tied to its place on the page. Check the open source core: Lectio.

Evals compares agents, and programs built on models, on the same tasks. Each setup tries every task several times, every attempt is graded, and Evals reports which did better, by how much, how sure that is, and what else differed between them. Check the open source core: Examen.

Developer documentation for the platform lives at platform.latere.ai, and the Go SDK reaches every capability from code.

Two capabilities for running models on compute you control, each built on an open source core.

Model Serving deploys and operates open-weight models, from a single GPU host to a Kubernetes fleet. Freeze and verify model weights, start an inference engine, and expose OpenAI- and Anthropic-compatible endpoints with health checks. These endpoints can sit behind Lux for model routing and access control. Check the open source core: Fornax.

Model Execution runs open-weight LLMs in pure Go, with no Python or vendor runtime. Check the open source core: Forma.

ReplicHAI audits whether a research paper actually reproduces. Give it a paper: it finds what the authors released, implements and re-runs the work, records every decision the paper left unwritten, and returns a verdict you can take apart. What reproduced, what did not, and what could not be judged, component by component, published with the full transcript beside it. Finished audits are public in the registry, and its source is open in replichai.

Tools and libraries to build with the platform and beyond.

AI as an Infrastructure is our open book on AI as an infrastructure, from systems to agents: history, design decisions, and foundations, written for software engineers new to machine learning. Also in 中文.

• Human judgment is irreplaceable. AI can execute, even reason. But deciding what matters, what is worth doing, and what trade-offs are acceptable still requires a human in the loop.
• Transparency over magic. People should be able to understand what an agent is doing and why. Progress, decisions, and results should be inspectable, with opportunities to intervene and correct course.
• Autonomy is a spectrum. Full AI autonomy and full manual control are both valid. Where to draw the line should always be a human decision.
• Build for the long term. AI capabilities will keep evolving, paradigms will keep shifting. But the principle that human judgment belongs in the loop will not. We build for that to remain true.

Found a vulnerability in a Latere service or repository? Read the security policy and report it through GitHub private vulnerability reporting or to security@latere.ai. It covers scope, testing rules, response times, and safe harbor for good-faith research.

latere.ai · Docs · Open Source · Blog · Contact