Glossary
The terms used across Productivity Workbench, defined plainly.
The platform
Productivity Workbench
Productivity Workbench is a product delivery platform that offers two ways of working: the Product Workspace for team-led agile delivery, and the AI Product Studio for AI-native delivery under human review.
It is one platform rather than two products so that an organisation can run both without maintaining two sets of tooling, two backlogs and two reporting models.
See Productivity Workbench in context
Tenant
A tenant is the isolation boundary for customer data: every read and every write is scoped to a single tenant, at both the frontend and the backend.
Isolation is enforced server-side rather than by the interface, so a request that escapes the boundary fails at the data layer rather than relying on the screen not offering it.
Product Workspace
Product Workspace
The Product Workspace is the team-led way of working: people plan and do the delivery, and the tool supports them with vision, outcomes, epics, features, stories, sprints, capacity and reporting.
It uses Scrum for smaller teams and SAFe for larger enterprises. AI is optional here and arrives through bolt-ons.
See Product Workspace in context
Bolt-On
A Bolt-On is an optional add-on that brings AI assistance to the Product Workspace, such as drafting help or coding-agent integration.
Bolt-ons are deliberately modest in scope. They are how a human-led team gets some AI support without changing how it works; the AI Product Studio is the option for teams that want AI to run delivery.
Vision
A Vision is the statement of what a product is for, sitting at the top of the delivery hierarchy above outcomes, epics, features and stories.
It exists so that every item beneath it can be traced back to the intent it serves, which is what makes strategy-to-value traceability possible at all.
Outcome
An Outcome is the measurable change a piece of delivery is meant to produce, sitting between the Vision and the Epics that pursue it.
Outcomes are what portfolio governance funds and judges. Features are how you attempt an outcome, not a substitute for stating one.
Agile Release Train (ART)
An Agile Release Train is a long-lived group of teams in SAFe that plans and delivers together on a shared cadence.
Productivity Workbench supports ARTs for organisations running SAFe, alongside PI planning and portfolio alignment.
See Agile Release Train (ART) in context
Programme Increment (PI) Planning
PI Planning is the SAFe event where the teams on an Agile Release Train plan the next increment together, agreeing objectives and surfacing cross-team dependencies.
Preparation matters more than the event: candidate features need sizing and prioritising beforehand, or the room spends its time doing analysis instead of planning.
AI Product Studio
AI Product Studio
The AI Product Studio is the AI-native way of working: you state an intent, approve a plan, and a governed fleet of AI agents executes it, with every output passing a recorded human review before it ships.
It is an alternative to the Product Workspace rather than an upgrade to it. The sprint cycle is replaced by a continuous delivery loop.
See AI Product Studio in context
AI Delivery Lifecycle (AIDLC)
The AI Delivery Lifecycle, or AIDLC, is the delivery method the AI Product Studio runs on: eight phases carrying an idea from intent to release, each with a human decision gate.
It exists because AI moves where the effort goes. When execution is cheap, the scarce work is stating intent precisely and reviewing output properly, so the lifecycle is built around those rather than around estimation and iteration planning.
See AI Delivery Lifecycle (AIDLC) in context
Intent
An Intent is a natural-language statement of a problem to solve, turned into structured, agent-ready work.
It is the entry point for everything in the AI Product Studio. Intents replace the ticket as the unit of request, because an agent needs the reasoning behind a request, not just its title.
Plan Gate
The Plan Gate is the human checkpoint between an agent proposing a plan and the agent executing it.
It is the cheapest place to catch a misunderstanding. Correcting an agent’s plan costs a minute; reviewing the work it produced from a wrong plan costs the whole run.
Run Board
The Run Board is where agent runs against intents are tracked, from queued through executing to awaiting review.
It is the Studio’s equivalent of a sprint board, but the unit on it is a run rather than a story, and it moves on the timescale agents work at rather than on a daily stand-up.
Review feed
The Review feed (formerly Review Theatre) is where a person sees everything the Definition of Done's agents want from each pull request: rework they asked for, questions they need answered, evidence that went stale when the pull request moved, and items still waiting on a reviewer role.
Answering an agent is recorded on the checklist item as evidence, never as a decision; the item is then reassessed. The gate itself still needs both its agents and a person to approve it.
Work Package
A Work Package is a unit of agent work grouped for review and release, moving through a lifecycle of states rather than being edited freely.
Its status is field-locked server-side: it changes only through the lifecycle endpoints that enforce the transition rules, not by editing the record.
Artefact Passport
An Artefact Passport is the evidence record for one work package or one release in the AI Product Studio: every gate item, the decision on each side of it, the evidence behind each decision, every exception, the CI results at the assessed commit and which run produced the code.
It exists so that "what was checked before this moved?" has one answer per artefact rather than a search across pages. It is assembled from the evidence ledger each time it is opened, states whether that ledger is intact, and lists what it does not know rather than assuming it. When a release is approved the passport is also issued: frozen and signed, so that someone outside the Studio can check the file with a published public key alone.
See Artefact Passport in context
Evidence Ledger
The Evidence Ledger is an append-only record, one per work package and per release, of every assessment, decision, exception and approval made on it, written at the moment each happens.
Each entry carries a hash of the record it came from and of the entry before it, so a missing or altered entry shows. The application cannot edit or delete an entry. If the ledger cannot record something, the change is still saved and the gap is itself recorded and shown.
See Evidence Ledger in context
Gate Exception
A Gate Exception is a recorded decision that a checklist item need not be met for one work package: who granted it, under which authority, why, and until when.
Waiving an item and overriding the reviewer agents are both exceptions. While one is in force the item counts as waived; once it has expired, or a material change has voided it, the item is unmet again and the work package cannot take its next step until the exception is renewed or the item is met. On a Studio's own items an expiry is optional.
Organisation Hub
Organisation Hub
The Organisation Hub is a management plane for administering many Product Workspaces and AI Product Studios from one place.
It is not a way of working and not a third product. It exists so a larger business can model its real structure, attach the right Workspaces and Studios beneath it, cascade policy and report across the estate. Every paid plan is bought by an Organisation, so it comes with Business and Enterprise.
See Organisation Hub in context
Guardrail Policy
A Guardrail Policy is a set of Definition of Ready and Definition of Done items, release rules and separation-of-duties requirements that an Enterprise organisation sets once and every linked AI Product Studio enforces.
A Studio may tighten an organisation item but never remove or weaken it. Activating a policy freezes a numbered version; a work package keeps the version it was created under, and its Artefact Passport names it.
Related
The FAQ answers questions about the product, and the documentation covers how each page works.