A thorough walkthrough of how Plot keeps your product knowledge accurate, governed, and AI-ready — from the moment you import a document to the moment your AI answers a question correctly.
Every product ships faster than its documentation can follow. The gap between what a product does and what the knowledge base says it does widens with every release, every API change, every deprecated feature.
When an AI tool draws from that knowledge, it presents what it finds with full confidence — regardless of whether it is still accurate. Plot closes the gap between what is true and what is retrievable.
Every document in Plot is broken into named sections. Each section has an owner, an expiry date, and a defined purpose. This structure is what makes governance possible — and what makes your knowledge base trustworthy enough to feed an AI.
Import existing documentation from Confluence or upload a PDF and Plot proposes a section structure automatically. You review, adjust, and confirm. Each section captures a specific piece of your product's story — what the system does, what rules it enforces, what inputs it requires, what conditions cause success or failure.
The section is the unit of governance. Not the document. This means a single document can have sections with different owners, different review schedules, and different statuses — reflecting the reality that different parts of a product evolve at different rates.
When you create a document, Plot provisions a Maintainers group automatically. The person who creates it becomes the first leader. From there, access is managed at both the organisation level and the document level — giving teams the flexibility to govern knowledge the way their product is actually structured.
Every update to a section goes through a defined workflow before it reaches your AI pipeline. Nothing is published without review. The audit trail records every decision — who changed what, when, and why.
When a section is edited, it moves to an in-review state. The assigned approver receives a notification and reviews the change — with a word-level diff view showing exactly what shifted within the surrounding sentence, not just which line changed.
The approver can approve, request changes, or reject. Every action is logged in the audit trail. The record shows who made the decision, what the previous content was, and what the approved content is. This is not just useful for compliance — it is the mechanism that makes your knowledge trustworthy.
When connected to GitHub, Plot surfaces a prompt in the VSCode extension the moment a pull request is merged that may affect existing documentation. The developer sees which sections might need updating, opens them directly from the prompt, and submits for review — without leaving the IDE. Code change and knowledge update become linked events rather than separate workflows.
The moment a section is approved, it propagates to your AI pipeline — within 60 seconds. Your AI answers from knowledge that has been reviewed, approved, and is meant to be there.
Plot connects directly to your AI pipeline through one of two paths: your own infrastructure, or Plot's managed RAG layer. In either case, the moment a section moves to Active status, the updated content is delivered downstream automatically. Nothing sits waiting for a manual export or a scheduled job.
The 60-second propagation window is not aspirational — it is the measured latency from section approval to index update, validated under load. Your AI answers from the version of the section that was last approved, not from whatever happened to be in the knowledge base when the index was last refreshed.
When a section is deprecated or deleted, it is removed from the index in the same propagation cycle. Ghost rows — index entries that no longer correspond to active content — are not permitted. The index reflects the current state of approved knowledge, always.
Plot supports two integration models. Both result in the same thing — AI that answers from knowledge that has been reviewed and approved. The difference is who manages the infrastructure.
Plot manages the documentation lifecycle. When a section is approved, a signed webhook delivers the updated section text — along with metadata including section ID, document ID, owner, and approval timestamp — to your own vector database endpoint.
You handle embedding, indexing, and retrieval using your existing infrastructure. Plot handles governance. Your inference costs stay transparent and in your control.
Best for teams with existing AI infrastructure and a dedicated engineering function.
Plot manages the complete AI pipeline — embedding, indexing, semantic search, and response generation. You receive a query endpoint and an embeddable widget, ready to deploy.
Ask questions in natural language. Plot retrieves the most relevant sections, assembles them into a structured prompt context, sends them to the configured LLM, and returns a source-attributed response. No infrastructure to manage. No vector database to maintain.
10,000 queries per month are included in the Growth tier, with overage billed per thousand. Query results are filtered by the requesting user's access permissions — the right people see the right knowledge.
Best for teams without a dedicated AI engineering function who want to deploy accurate AI responses without managing the infrastructure themselves.
From the moment a section is created to the moment it is deprecated, its status is always defined, always visible, and always meaningful.
14 days, full access, no credit card. Set it up now — before the moment you need it arrives without warning.