Start with your questions

Genie Acceleration Program

Genie adoption, led by your own champions

You already know what Genie can do. The Blueprint Genie Acceleration Program empowers effective adoption. Gain access to Genie accelerators and facilitated hands-on workshops that will grow your organization's Databricks Genie capabilities into a Center of Excellence that outlasts the engagement.

Start with your questions

Genie for the CFO

by Blueprint Technologies

A governed finance answering space over the Finance entity model, with trusted assets, measures, filters, and expressions spanning revenue, margin, the close, working capital, cash, budget variance, and forecast.

Write a message...

Where are we over or under budget, by account, this fiscal year?

How much net revenue have we recognized this fiscal year?

How is days sales outstanding (DSO) trending over the trailing 12 months?

How many months of cash runway does each entity have at the current burn rate?

What is our latest revenue forecast for this fiscal year?

The real blocker

Building a Genie agent is not adoption

Self-service analytics rarely stalls on the tool. Switching on an assistant is the easy part. Whether a business team actually uses it rests on three things that sit outside it, and all three have to hold.

Confidence

An answer nobody trusts is not an answer

Someone who cannot tell whether a number is right does what they have always done and asks a person to check it. Adoption stops there rather than at the interface. Confidence does not come from a better chat window; it comes from being able to see the logic behind an answer and knowing a person verified it.

Context

The system does not know your business yet

Your teams already know what they want to ask. What is missing sits in the data layer: undocumented columns, join paths nobody declared, and the business vocabulary that lives in people's heads rather than in the catalog. Published research on real enterprise warehouses bears this out.

Tidy sample databases ~81%
Real enterprise warehouses, given the answer's structure as a hint 30.1%
Real enterprise warehouses, out of the box 11.4%

Execution accuracy reported in BEAVER, an enterprise benchmark for text-to-SQL. Published research measuring the general problem of question answering over real enterprise schemas; not a measurement of any single product.

Continuous improvement

Quality is judged once and never again

Agents get stood up without a benchmark set, so quality is an impression formed on demonstration day. Nothing then notices when a definition changes, a table moves, or two agents begin answering the same executive question differently.

Outcomes

What you keep

Today

An effective Genie agent

Built by your team in your own workspace, reading your own Unity Catalog and enforcing the permissions your people already have. The vocabulary, the join paths, and the definitions agreed at kickoff are encoded in it rather than living in someone's head, and your people know how it works because they built it. The agent can be rebuilt from the question set you agreed; the question set cannot be rebuilt from the agent.

Next

A data plan

From the assessment running alongside the build: a readiness scorecard you keep and re-run yourself, plus a prioritized list of what stands between you and the next wave of questions, ordered by how many each fix unlocks. The agreed question set doubles as the benchmark, so quality is something you re-measure rather than remember.

Horizon

An AI on Databricks strategy

The horizon past what you just built. A facilitated session places the work you just did inside a wider AI operating model, and you keep the readiness engine behind it: seven dimensions and fourteen evidenced findings, scored from system tables against Databricks' own evaluation criteria. Its reach is enterprise-wide, so you run it against your own account on your own timeline.

Ongoing

A Center of Excellence

The point of the workshops. People inside your business who can assemble an agent, define a measure once so every surface returns the same number, and govern model access the way they already govern tables. The engagement ends; they do not. This is the difference between having a Genie agent and being able to build the next one.

How it runs

Accelerators, support, workshops, and strategy

A sequence that starts with your questions and ends with the people who can run this without us.

Before we start

The questionnaire

You complete a short questionnaire that captures the questions in play and your data landscape. It decides what the program covers before anyone quotes it.

FoundationsCenter of ExcellenceClose

Readiness assessment install supportyou install and run it, with us alongside

Install and run
  • Install the assessor
  • Run it against your catalogs
  • Baseline in hand before kickoff

Kickoffagree the question set and what a correct answer is

Kickoff Blueprint-led
  • Agree the question set
  • Define what a correct answer is
  • Prioritize the remediation backlog

Genie for C-Suite install supportinstall the pack, bind it, deploy the agent

Install and deploy
  • Install the pack, bind your catalogs
  • Materialize the trusted assets
  • Deploy the agent

Workshop: the Genie agenthow one is assembled, tuned, and judged

Walkthrough, demo, lab Blueprint-led
  • How an agent is assembled
  • Tune it and keep the change
  • Read a benchmark, judge an answer

Workshop: Unity Catalog semanticsdefine a measure once, govern what it means

Walkthrough, demo, lab Blueprint-led
  • Define a measure once
  • Comments and synonyms as meaning
  • Certified against draft

Workshop: Unity AI Gatewaygovern model access the way you govern tables

Walkthrough, demo, lab Blueprint-led
  • AI services as Unity Catalog securables
  • Rate limits, budgets, and tags
  • Service policies as guardrails

Readoutthe assessment re-runs against its own baseline

Readout Blueprint-led
  • Re-run against your own baseline
  • Review the agent you built
  • Agree next steps

AI on Databricks strategythe horizon past what you just built

Strategy session Blueprint-led
  • Score the account on seven dimensions
  • Place it in a wider operating model
  • Keep the engine and re-run it

Accelerators

What you start from

Blueprint builds accelerators: shipped software and starter content that take the slowest work out of Databricks Genie adoption.

Starter packs for four business functions

Each pack carries a semantic layer, an entity model, and a bank of starter questions with example logic, mapped onto your own tables at kickoff. Your team edits and verifies from there rather than starting with an empty space.

PackCategoriesStarter questions
Finance928
People922
Marketing1124
Supply Chain1022
Four live packs96

The Genie readiness assessor

Published on Databricks Marketplace, where enterprises install and run it themselves. In this engagement it runs against your own Unity Catalog, on your own compute, and scores your data, producing the readiness scorecard you keep as a dashboard and re-run yourself, plus a remediation backlog with an effort estimate on every item.

The AI on Databricks readiness engine

The program answers the questions you bring to it and tells you what your data needs next. The wider question, what your platform can support across the enterprise, is what this engine scores. We run it with you in a facilitated strategy session and hand it over; its reach is enterprise-wide, so you run it against your own account on your own timeline.

1Platform capability. Whether the workspace can actually run the AI feature you want, probed directly rather than assumed.
2Data posture. Scored across seven dimensions, with data quality broken into six sub-scores.

Readiness

What you bring

Four things the program needs from you. Everything else we bring.

A named business sponsor Attends kickoff, approves the question set

Someone who can say which questions matter and commit to what a correct answer looks like. The question set agreed at kickoff is what the agent is measured against at the readout.

A named technical builder Available throughout

Your people install, deploy, and build, with Blueprint Delivery advising, teaching, and reviewing throughout. Naming that builder is what makes the program work; if you do not have one, implementation is available as a separate engagement.

A workspace with Unity Catalog Enabled and populated

Plus a technical contact who can grant catalog access and make a SQL warehouse available. Everything runs in your workspace, on your compute, so no Blueprint credentials are needed.

The questions that matter Captured before we quote

The questionnaire captures the answers your teams are waiting on and where that data lives. It decides what the program covers, before anyone commits.

Next step

Start with your questions

Tell us the questions your team keeps asking. The questionnaire takes a few minutes, and it ends with a recommendation scoped to what you just described.

We handle what you submit as business contact information. See Privacy and Compliance.