Glossary
Definitions for the Dots vocabulary.
Get Started
- Agent EnvelopeAgent Envelope — the YAML-frontmatter-plus-Markdown row block shared by CLI, MCP text, and docs export.
- DIIICE GlossaryDIIICE — the six knowledge types a DOT is classified into: Data, Intelligence, Instructions, Ideas, Context, Examples.
- DOTDOT — the atomic unit of knowledge in Dots: a single typed row that is also a node in the knowledge graph.
- EdgeEdge — a typed, directed link between two rows, drawn by an author or discovered by a behavior.
- Dotabase Operation Manual (DBOM)Dotabase Operation Manual (DBOM) — the per-Dotabase purpose, authority, and judgment that tells a person or an agent how that Dotabase is meant to be used.
- ViewView — a saved lens over a Dotabase's rows, with its own type, filters, sorts, and grouping.
- AccessThe per-tenant identity and authorization plane — the RBAC module covering principals, roles, grants, credentials, approvals, and audit within one company.
- BehaviorA bound, per-Dotabase extension that runs automatically on a row's lifecycle — the platform's mechanism for making a Dotabase *do* something rather than merely store rows.
- Chip (AI co-worker)The AI co-worker inside Dots — the product's in-app assistant persona, with hands. Chip works inside the user's actual workspace through the `dots` CLI: it searches Dotabases, writes rows, manages tasks, and cites what it retrieved.
- Context CabinetThe foundational Dotabase holding every captured file and document — raw, searchable, and the default destination for anything with no better home. Rows are `CTX-N`; it is usually just called "the Cabinet".
- DIIICE: ContextThe DIIICE type for background, circumstances, preferences, or situational framing that situates other knowledge.
- DIIICE: DataThe DIIICE type for raw facts, records, measurements, or datasets with no interpretation layered on top.
- DIIICE: ExamplesThe DIIICE type for concrete instances, demonstrations, samples, or cases that illustrate a point.
- DIIICE: IdeasThe DIIICE type for proposals, hypotheses, concepts, or possibilities not yet acted on or verified.
- DIIICE: InstructionsThe DIIICE type for directives, procedures, steps, rules, or commands telling someone how to act.
- DIIICE: IntelligenceThe DIIICE type for synthesized insight, analysis, conclusions, or judgments derived from data.
- DotabaseA structured, searchable collection of DOTs — the platform's core container, and the single foundational data primitive Dots is built on. Roughly "a database of DOTs", but with an AI-native shape: schema, saved views, bound behaviors, an embedding path, and its own Operating Manual telling agents how to use it.
- DotsThe company and the platform — a knowledge platform building the infrastructure layer for human intelligence, built on a single foundational data primitive: the Dotabase.
- HIMHuman Intelligence Model — the proprietary large language model Dots is building, trained on human intelligence rather than on the public web.
- HUMAN.mdA document describing who a person is in the context of the work — their role, what they own, what they defer, their voice, their strong suits, and how to operate with them. It is read at session start so an agent works with the person rather than at them.
- Memory BankThe foundational Dotabase holding episodic memories — durable observations, decisions, rationale, and failures worth keeping across sessions. Rows are `MEM-N`.
- Power-Up PlaybookA curated knowledge index that scopes what a capability retrieves. Abbreviated **PUP**; rows are `PUP-N`.
- SpaceA container that groups Dotabases inside a workspace — the organizing layer one level above the Dotabase.
- SuperpowerAn invokable capability stored as a row and callable from Chat, the CLI, MCP, or the API. Rows are `SPWR-N`.