What is a Decision Topic?

A Decision Topic is anything leadership wants Decision AI to keep watching - a KPI, a customer segment, a contract renewal window, a board question, or a specific risk. Each Decision Topic runs continuously against your Knowledge Graph and produces a Decision Brief whenever something material changes. In the DataBlueprint product UI, Decision Topics appear simply as 'Topics' in the left navigation.

What is a Decision Brief?

A Decision Brief is a board-ready summary produced when a Decision Topic refreshes. It includes the question, the finding, the supporting evidence (with row-level traceability to source systems), and a recommended next move with projected impact.

What is an AI credit?

Credits power DataBlueprint's Decision AI activities - asking questions, refreshing Decision Topics, generating Decision Briefs, and running risk scans across your Knowledge Graph. Consumption depends on the complexity of the request.

What is a Knowledge Graph?

A live, unified map of every entity in your business - customers, contracts, products, costs, people - and the relationships between them, built automatically from every system you connect.

How is DataBlueprint different from a BI tool like Tableau or Power BI?

BI tools show tables and dashboards and require a data team to model and maintain them. DataBlueprint understands business relationships across all your systems, answers strategic questions in plain language, and surfaces recommended actions with traceable evidence. BI tells you what happened; Decision AI tells you why and what to do.

Is DataBlueprint read-only?

Yes, by default. DataBlueprint connects to all source systems in read-only mode. Agent write-back to systems like HR, CRM, and PSA is opt-in per system on the Business tier and requires explicit per-action approval.

Where does AI inference run?

All AI inference runs on a private LLM powered by AWS Bedrock, inside your AWS environment. Prompts, completions, and query results are never sent to OpenAI, Anthropic public endpoints, Google, or any external model provider. AWS Bedrock contractually guarantees customer inputs and outputs are never used to train shared models.

Why DataBlueprint Pairs a Knowledge Graph With a Private LLM on AWS Bedrock

The graph holds the meaning of your business. The model reasons over it. Hosting that model privately on AWS Bedrock is what makes the answers safe to act on.

A Knowledge Graph alone is powerful but inert. Someone — or something — still has to ask it the right question and interpret the answer. That is the role of the language model. The choice of which model and where it runs is what separates a serious Decision AI platform from a demo.

DataBlueprint pairs the Knowledge Graph with a private LLM powered by AWS Bedrock. Here is why that combination matters.

Your data never leaves your tenant

Public chat models send prompts to a vendor. Even with privacy settings, you are trusting a third party with the contents of every executive question. That is incompatible with how most CFOs and operations leaders actually work.

A private LLM on AWS Bedrock runs inside an AWS account in a region you control. The Knowledge Graph DataBlueprint maintains for you stays in that same boundary. Prompts, graph queries, and generated Decision Briefs do not flow out to a public model endpoint.

Read-only by construction

Just as DataBlueprint connects to your source systems read-only, the model has read-only access to the graph. It can reason. It cannot mutate. The risk surface is small and well understood.

Reasoning needs structure

A modern LLM is a brilliant reasoner over structured context. Hand it raw spreadsheets and the answers degrade quickly. Hand it a Knowledge Graph — a clean, typed, relational model of your business — and the same model produces traceable, defensible answers.

That is the design choice DataBlueprint made: invest heavily in the graph, then point a private LLM at it. The answers carry citations back to the source rows in your CRM, ERP, or warehouse. Nothing is conjured.

Why hallucination drops

Hallucination is what happens when a model has to invent. When it can read the actual entities and relationships in your graph, there is nothing to invent. It is reading, not guessing. That is why every Decision Brief in DataBlueprint shows its evidence — the graph paths walked and the source records referenced.

What this means for your team

Your CFO can ask a sensitive question about customer profitability without that prompt traveling to a public model. Your COO can ask about technician utilization without exposing names. The combination of a private model and a read-only graph is what makes Decision AI usable in regulated, security-conscious environments.

That is the bet DataBlueprint is built on: structure first, private reasoning second, dashboards never.