Core Concepts
A short mental model of how InaiBridge works.
Sources
A source is a project tool you connect — Jira, Asana, or Azure DevOps. InaiBridge reads your work items from each source.
One canonical record
Every source maps into a single, consistent internal representation. That means your intelligence spans all your tools at once — a risk in Jira and a slipping initiative in Azure DevOps are reasoned about together, not in silos.
Different sources expose different levels of detail, so InaiBridge records how trustworthy each signal is. A tool that only reports "done / not done" is labeled as lower-confidence than one with rich status and dates — and InaiBridge never inflates a signal beyond what the source actually supports.
What that record holds
Every initiative is held as three things kept deliberately apart: what you planned, what is actually happening, and the gap between them. Keeping them separate is what lets InaiBridge tell you why something moved rather than only that it did.
Status is two questions, not one. Where the work is — not started, active, done — is tracked separately from how it's going against the date — on track, at risk, late. Most tools collapse these into a single field, so "in progress" has to mean both, and a late project and a healthy one look identical. InaiBridge keeps them distinct.
And the baseline you're measured against is itself recorded for how much it can be trusted. A late project must never look on-track because the baseline was quietly reset to today. Where InaiBridge had to infer a baseline rather than read one, it says so — a deadline that moved is a deadline that moved, and no amount of re-snapshotting will launder it into "on track".
Workflows and nodes
Work runs as a workflow — a graph of nodes connected together. A node might pull in a source, run an agent, pause for a person, or send a result onward.
InaiBridge ships with the workflows you need rather than asking you to build them: the Inaibridge-Lumina suite is seeded into your organization, ready to run. You can read any workflow, and watch it run, in the Customer app.
Organization scoping
Everything belongs to your organization — the tenant boundary that scopes your connected data, secrets, workflows, and intelligence. See Connecting your data.
Scoping goes one level further, inside your org: what you see depends on the programs you own. Counts and totals across the apps describe the programs visible to you, not necessarily your whole organization.
InaiAgents
An InaiAgent is InaiBridge's grounded intelligence — reasoning over your connected data to produce briefings and answers anchored in real records. It's a capability with several surfaces, not a single button: it shows up as the Project Briefing (a grounded multi-tier briefing), Mira (a grounded conversational companion), and as a grounded InaiAgent step inside a workflow.
Don't confuse that step with the plain Agent step. An Agent is a general-purpose LLM step (a model and a prompt); an InaiAgent is grounded — it reasons over your data, cites its sources, and abstains when the data can't support an answer. The distinction that matters is grounded (InaiAgent) versus plain LLM (Agent). See InaiAgents.
Proposals, not changes
The last concept is the one that shapes everything else. What the intelligence concludes is a proposal: a Suggested risk, finding, lesson, or close-out. It does not enter your record until a person decides it in the Inbox, and that decision is signed. An engine conclusion is never an organizational position on its own.