> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trubuild.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Brain and Agent Sam
> An organization-wide knowledge graph over your past procurement: explore it, search it, and plan the next tender with Agent Sam. Admin-only add-on.
Brain is a knowledge graph built from every project, package, evaluation, vendor, and document in your organization. It connects what happened across your past tenders (who bid, who won, how they scored, and on what evidence) and distills it into insights you can act on. Agent Sam, your organization's procurement strategist, sits on top of Brain to help you plan the next tender from what you already know.
Brain is an advanced add-on that is not enabled by default. It must be turned on for your organization, and access is limited to organization owners and admins, because the graph aggregates data across all packages. Regular members do not see it. To request it for your organization, contact TruBuild support.
## Why this matters
Procurement knowledge usually lives in people. When a commercial manager who ran your last three fit-out tenders leaves, so does their sense of which vendors front-load preliminaries, which criteria actually separated bidders, and which scope gaps keep recurring on a given asset type. Brain keeps that memory in the organization: every evaluation you complete adds to a graph the whole leadership team can query, so the next tender starts from evidence instead of recollection.
## What Brain contains
Brain has two layers: the raw graph of your procurement data, and the insights it derives from that graph.
### The knowledge graph
| Node type | What it represents |
| ---------------------------------------- | ------------------------------------------------------------------------------------- |
| **Project**, **Package**, **Asset** | Your procurement structure, from project down to individual assets |
| **Vendor**, **Vendor Entity** | Bidders and the corporate entities behind them |
| **Tech Evaluation**, **Comm Evaluation** | Completed technical and commercial evaluations, with scores and round information |
| **Criterion**, **Criterion Concept** | Individual evaluation criteria and the recurring concepts they map to across packages |
| **Document** | Source documents, each openable from its node |
Nodes are linked by relationships such as which vendor bid on which package, which bid won, and which criteria an evaluation scored against, so you can trace any conclusion back through the graph to its evidence.
### Insights
On top of the graph, Brain surfaces three kinds of insight:
| Insight | What it is | Example |
| ----------- | --------------------------------------------------------------------------- | --------------------------------------------------------------------- |
| **Pattern** | A recurring observation across past projects, shown with a confidence level | A vendor consistently strong on HSE but weak on programme |
| **Gap** | A missing capability or risk identified across projects | No bidder in your pool covers a discipline your pipeline needs |
| **Lesson** | An actionable learning from past evaluations | A criterion that never differentiated bidders and could be reweighted |
Treat patterns and lessons as leads, not verdicts. Each insight lists the documents and evaluations that evidence it, so before a pattern influences a shortlist or an award recommendation, open its **Sources** and confirm the evidence supports it.
## Explore the graph
The **Graph** view is an interactive map of your procurement knowledge.
Use the search box to filter nodes by title, for example a vendor name or package. The graph opens centered on a top insight by default.
The depth slider expands the view from 1 to 3 degrees of connection. Depth 1 shows direct relationships; depth 3 reveals indirect links, such as two packages connected through a shared vendor.
Open the filters panel to toggle which node types and relationship types are visible, for example vendors only, or insights only. Toggle the legend to see the color key for node types.
Click any node to open the inspector pane. Double-click a node to recenter the graph on it. Drag the inspector's left edge to resize it.
### The inspector
The inspector shows everything Brain knows about the selected node:
* **A context card** at the top: documents get an **Open document** button with a download link, vendors show their award status and entity information, and evaluations show scores and round information.
* **Quick Facts**: key details such as type, status, confidence, and dates.
* **Connections**: the node's neighbors, grouped by relationship type.
* **Sources**: the documents and evaluations that evidence this node, so you can verify any insight against its underlying record.
## Browse pages
The **Pages** view organizes every node into a browsable library, useful when you want to read through what Brain holds rather than navigate visually. Nodes are grouped into four folders:
| Folder | What it holds |
| ------------------------- | ------------------------------------------------------------------ |
| **Insights** | Patterns, gaps, and lessons |
| **Vendors & Evaluations** | Vendor entities, vendors, and technical and commercial evaluations |
| **Packages & Criteria** | Projects, packages, assets, criterion concepts, and criteria |
| **Source Documents** | All documents in the graph |
Each folder is paginated, with a search box that searches organization-wide within that folder (enter at least 2 characters). Every entry shows its type, title, a short preview, and when it was last updated. Click an entry to open its inspector page.
Start a tender-strategy session in **Insights**. Reading the current patterns, gaps, and lessons before you draft evaluation criteria or a bidder list is the fastest way to apply what previous tenders taught you.
## Search
The **Search** view searches the whole graph at once. Type into the **Search vendors, packages, criteria, evaluations…** box; plain-English descriptions work as well as exact names. Use the **Filter** chips to limit results to specific node types.
Each result shows a type badge, the title and a preview with your search term highlighted, and the last updated date, and links straight to that node's detail page.
This is the quickest answer to questions like "have we evaluated this vendor before, and on what?" during prequalification or bid leveling.
## Agent Sam
Agent Sam is the organization-wide assistant on the **Agent Sam** tab. Unlike the per-package [AI chat](/platform/chat), which works within one tender, Agent Sam draws on your whole Brain, so it can reason across projects and past tenders. Below the chat, a **Brain at a glance** panel shows your top patterns and insight pages as clickable cards, and a sidebar lists your past conversations.
### What to ask
Four quick actions get you started, and each maps to a common planning task:
| Quick action | What you get |
| ------------------------------------------- | -------------------------------------------------------------------------------------------- |
| **Draft an RFP from recent projects** | A downloadable Word document drafted from how your recent tenders were structured |
| **Build a pre-qualification questionnaire** | A PQQ based on your past evaluation patterns |
| **Suggest evaluation criteria** | Proposed weighted criteria with descriptions, grounded in what differentiated bidders before |
| **What works and what doesn't?** | An analysis of past evaluation effectiveness and vendor performance |
Beyond these, ask anything that benefits from cross-project memory: comparing a vendor's performance across packages, spotting recurring qualifications and exclusions, or checking which criteria historically separated bidders.
### How a conversation runs
Agent Sam works through the graph as it answers, showing each step in plain language, such as "Searched your past projects" or "Followed the connections", so you can see where an answer comes from.
When Agent Sam needs direction, it presents a short form. Answer each question, attach supporting documents if useful, or click **Use your best judgment** to let it proceed without your input.
Documents Agent Sam generates, such as a draft RFP or PQQ, appear in the conversation with a download link.
Use the thumbs up or thumbs down on any response. Your feedback helps improve answer quality.
Agent Sam's answers are grounded in your Brain, so they are only as current as your data. A generated RFP or criteria set is a strong first draft, not an issued document: review it against the current project's scope, contract form, and approval requirements before it leaves your organization.
## Related
The per-package assistant for questions and work inside a single tender.
Search and reuse tender query answers across packages.