Start
Overview
Lemma is an operating-system platform for business teams: pods combine data, agents, workflows, functions, connectors, files, and human surfaces around real work.
What you are building
A Lemma pod is a bounded operating system for one business use case. It owns the state, logic, automation, permissions, and operator surfaces for that use case. The goal is not to create a loose collection of AI helpers; the goal is to give a team a system that can run a repeatable operating loop.
Core resources
- Tables and records store durable structured business state.
- Files store unstructured context such as briefs, contracts, manuals, transcripts, and attachments.
- Functions perform deterministic typed backend actions.
- Agents handle judgment-heavy work such as drafting, classification, extraction, research, and summarization.
- Workflows orchestrate multi-step processes, waits, branching, schedules, and app-triggered runs.
- Apps provide full operator interfaces for repeatable human workflows.
- Conversations provide user-facing or agent-scoped interaction with instructions, metadata, tools, and streaming output.
Mental model
Models provide intelligence. Lemma provides rails: state, permissions, connectors, deterministic tools, human review, auditability, and interfaces.
When to use each layer
| Need | Use |
|---|---|
| Single-table create/update | Record API |
| Validation, coordinated writes, external calls | Function |
| Drafting, research, classification, extraction | Agent through task or conversation |
| Stages, branching, waits, schedules, triggers | Workflow |
| Primary user interaction is chat | Conversation / assistant surface |
| People need queues, forms, detail views, and actions | App |
| Documents, attachments, source material | Files |