1. Home
2. Help
3. Getting started
4. Key concepts

Key concepts

Projects, sessions, engines, tools, chains, trackers, quotas. The vocabulary that shows up throughout the rest of the docs.

Project

A project is the top-level container. Everything else — documents, engine sessions, vendors, trackers, tasks, schedule, analytics — is scoped to a project unless you explicitly view across projects.

Create a new project for each job. Most users have one project per active bid/contract. Archive a project when a job closes; you can still read it, but it won't show in the default project picker.

Engine vs tool

An engine is an AI-powered analysis that takes document input and produces structured output. Running one costs one run from your monthly quota. There are 19 engines.

A tool is a utility surface that doesn't call the AI. Tools don't consume your runs quota and are available on every plan including the free trial. There are 4 tools: Schedule Tracker, Email Classifier, Task Manager, Doc Chat.

Session

A session is one engine run. Every time you click Run analysis and get a result, that's a session. It's saved to your project with:

• The engine that was run
• The structured JSON output (what gets displayed in the results view)
• Timestamps and the user who ran it
• The document names used as input (the documents themselves aren't always persisted — see Data storage below)

Re-opening a saved session from the Results or History view does not consume a run — only new analyses do. You can re-export, re-read, and chain into new engines freely.

Chain

A chain is a suggested or automatic next step between engines. E.g., after running Contract Review, chain into Insurance Tracker (which inherits the insurance requirements clause), or into PM Handoff (which uses the contract analysis as its primary input).

Some chains carry data forward (dashed edges in the Platform Atlas graph) — the source engine's output auto-populates the target's input. Others are just suggestions (solid edges) that tell you "this is a logical next step" without auto-loading anything.

See the Engine chains page for the full graph.

Trackers

A tracker is a live table of items you manage over time — distinct from engine sessions, which are point-in-time analyses. There are 7 trackers: Procurement, Submittal, RFI, COI (Insurance), Punch List, Tasks, Schedule.

Several trackers can be seeded from an engine run. For example, Submittal Extractor extracts a list of required submittals from a spec section; you can import that list into the Submittal Tracker with one click, then manage each submittal through architect review.

Runs quota

Quotas reset monthly on the 1st of each calendar month. Workflow tools (Schedule, Email Classifier, Task Manager, Doc Chat) never consume runs — they're unlimited on every plan.

Data storage

What Trueleveler persists, scoped to your organization:

• Engine session outputs — the structured JSON results, searchable and re-readable indefinitely
• Tracker state — every row in every tracker
• Vendor records — names, contacts, ratings, notes, global to your organization
• Project audit log — uploads, runs, edits, exports; forensic trail per project
• Documents you upload to the Document Library — the actual file, retained

What isn't persisted:

• Documents sent directly as engine input without going through the Document Library — read once, analyzed, discarded. The structured output is kept.
• Inter-organization data — your project data never mixes with another org's. Enforced at the database layer via Supabase row-level security.

Organization and team

When you invite your first teammate, Trueleveler creates an organization container around you. All your projects, vendors, and engine runs belong to the org, not your individual user.

Four roles: Owner (founder, billing), Admin (manage team + all projects), Member (run engines + edit trackers), Viewer (read-only). See Team & roles for details on what each can do.

Exports

Every engine result can be exported to PDF (formatted report), XLSX (Excel with preserved structure), or CSV (raw data). Reviewer sign-off fields are included in the PDF format when applicable. Full details on the Exports & reports page.

Where things live