Human intelligence in the loop.
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.
Autonomous agents can now act across tools, models, data, and environments. The question is no longer whether they can move, but where judgment remains. Every system we ship follows one principle: the human stays in the loop. Work is visible. Authority is bounded. Output is reviewable. AI executes, humans decide.
Seven systems keep autonomous agents legible, bounded, and reviewable, with one identity tying them together.
Wallfacer is an autonomous engineering platform. It works across multiple levels of abstraction: chat for exploration, specs for design, tasks for parallel execution, and code for precise edits. Full autonomy when it earns trust, full control on demand. Open source, runs locally, bring any LLM provider.
Topos runs managed agent workspaces. Run Codex, Claude Code, or custom agents in governed cloud sessions with safe model access, tools, and resumable state. Its Adversarial Review capability puts an independent critic on an agent's output before you see it, vendor-neutral and auditable, so only the disputes that survive reach a human.
Lectio turns any document into data. Drop in almost any file, PDF, Word, spreadsheets, slides, images, or web pages, and it reads every page, resolves the layout, and returns readable text or structured data, with every detail sourced back to where it came from on the page.
Drive is the user data plane. It persists user-owned state across products: personal files, org shared drives, Cella-mountable workspaces, agent-produced code repos, and file-shaped agent memory. Both a Drive-style web app and the storage API that Cella, Topos, and Wallfacer mount.
Cella is a sandbox runtime for agents, developers, and MCP background jobs. One API call spins up a named sandbox, ephemeral or persistent. Fast enough to throw away, durable enough to keep.
Lux is a model gateway. One URL routes to OpenAI, Anthropic, Gemini, OpenRouter, and Ollama with sealed credentials, virtual keys, spend caps, hooks, and audit.
Identity is single sign-on for Latere. One identity authenticates people and agents across every product, with scoped tokens, central revocation, and a full audit trail.
- 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. Every AI action should be visible, auditable, and reversible. AI that cannot be inspected cannot be trusted.
- 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.
