Tattva Docs
Reference

Glossary

Every term used across Tattva, in one place.

The platform-wide glossary. Each project also has its own data-specific glossary under Ontology → Glossary.

Core

Workspace — Your company's container in Tattva. Holds projects and members.

Project — One scoped environment with its own data, ontology, history and decisions. Bound to a function and a vertical pack.

Project function — The team or domain a project serves. One of: Sales, Marketing, Growth, Customer, Product, People & Org, Decision Intelligence, Other.

Vertical pack — Industry-specific defaults applied to a project: CPG D2C, Marketplace, Subscription SaaS, Generic.

Source — Any data the platform reads — a file, a BigQuery table, a connected app, a document, a call transcript.

Source version — A snapshot of a source at a point in time. Every re-upload creates a new version.

Trust level — Whether a source can be cited freely (trusted) or is in exploration mode (sandbox).

Ontology

Ontology — The typed map of your business: metrics, entities, concepts, conventions, glossary terms, and the relationships between them.

Atom — A single thing in the ontology. Five kinds: metric, entity, concept, convention, glossary term.

Metric — A measurable quantity with a formal definition (e.g. Net Revenue = Gross − Returns − Discounts).

Entity — A noun: Customer, Order, Campaign, Channel, Rep, SKU.

Concept — A recurring abstract thing: New Customer, Activated User, At-Risk Account.

Convention — A rule you operate under: financial year, currency, time grain, segment definitions.

Glossary term — A team word that isn't a formal metric (e.g. "T1 city", "premium tier").

Candidate — A proposed new atom learned from data, awaiting human accept/defer/reject.

Conflict — A disagreement between two sources or two definitions that needs human resolution.

Canonical project — A workspace-wide project holding shared atoms that have graduated from individual projects.

Reasoning

Reasoning — The page where you ask questions and see answers.

Composer — The input box on the Reasoning page.

Run — One question-and-answer cycle.

Phase — One of the four steps of a run: Analyze, Select, Execute, Reflect.

Intent — The kind of question (lookup, ranking, trend, RCA, forecast, attribution, optimize, scenario, etc.).

Thinking depth — How hard the platform reasons: Off, Low, Medium, High.

Specialist — A dedicated analyst flow: RCA, Forecast, Attribution, Optimize.

Skill — A reusable, named, versioned analysis workflow.

Tool trail — The expandable list of every tool the platform ran to produce an answer.

Citation — A link from a claim in an answer back to the source row/query/definition it came from.

Clarification — A question the platform asks you when your question is ambiguous.

Decisions

Decision — A logged call, with question, evidence, hypotheses, tradeoffs, actions and horizon.

Hypothesis — An assumption a decision rests on. Validated, invalidated, or unresolved.

Action — A concrete step that follows from a decision, with owner and deadline.

Tradeoff — A cost or downside accepted by a decision.

Horizon — When a decision's outcome will be known.

Decision Sign — A team profile derived from decision history. Includes an archetype and five reasoning traits.

Archetype — Calibrated Optimizer, Patient Strategist, Iterative Operator, Hypothesis Hunter, or Balanced Generalist.

Trackers & Foresight

Tracker — A saved question that runs on a schedule. Cadences: hourly, daily, weekly, manual.

Tracker classification — Green / watch / alert, applied to each run based on thresholds.

Goal — A measurable target a tracker (or several) can be linked to.

Link role — How a tracker relates to a goal: primary, leading indicator, lagging indicator, guardrail.

Foresight — Forward-looking research. Modes: risk, opportunity, tradeoff, scenario, monitor.

Pinned URL — A web source you explicitly add to a foresight query.

Trusted domains / Excluded domains — Project-level lists that weight or block web sources for foresight scans.

Reports

Report — A shareable, structured summary of decisions, trackers, and answers, with citations preserved.

Slide — A panel inside a report.

Export — A static (PDF or slide-deck) version of a report, with citations preserved as footnotes.