Faucet
The plumbing’s already there.
Faucet turns questions asked in Claude, ChatGPT or Copilot into interactive HTML reports on your semantic models, shared and governed beside your Power BI reports.
Reports land here when someone asks for one
The future of business intelligence
The future is clear. We didn’t want to wait.
AI is how people are going to use BI
Not a feature next to the reports. The front door. Ask in the AI you already use, get an answer from the models you already trust. Far more people take part, and BI pays off at a multiple of what it does today.
No more limits on what a report can be
A report is a web page now, so it can be anything you can imagine. Not a menu of the visuals that came in the box. Never built a report? Say what you want and it’s yours. Build them for a living? You just got the whole web to build with.
Why we built it for ourselves
Working with clients every day, and adopting AI inside P3 at the same time, we could see where this was headed and what was missing. We wanted to pull that future into today, for ourselves and for our clients.
Reports
Build your best report yet. No coding required.
Anyone can build one. Experts go further.
Self-service BI only ever delivered self-service reading. Now you say what you want to see and you have it – your real numbers, your real permissions, in the meeting where the question came up.
And it takes the ceiling off the experts. Nothing has to be assembled out of stock charts any more, so the people who do this for a living are building pages that were not possible two years ago.
Answers from Rob
Describe it
In plain language, where the question came up.
Publish it
A real page, permanent link, versioned. A revision never overwrites the old one.
Share it
To named people or groups. A forwarded link grants nothing.
Build on it
Fork it, extend it, ask the next question. Lineage is recorded.
Chat with your data
In the AI your team already has open.
Your data plugged into the AI’s brain
Not one more chat box to remember. Faucet connects your data to the client your company already standardized on. Every question runs as the person who asked it, against your security, unchanged.
We tune the model for the AI first: names that mean something, descriptions that carry the business rule, relationships that are right. A semantic model built for people clicking through a report is not the same thing.
Answers from Rob
No AI client standardized yet?
That is the first conversation. It’s short.
Plenty have not picked, or picked and never rolled it out. We will tell you our preference and exactly why, rather than pretending to be neutral.
Which one, and why
An honest comparison against how your people work, not a feature grid.
Getting it adopted
Licensing, rollout, the security review, and the training that makes it stick.
Agents
Stop doing the rounds. Put an agent on it.
Watch
Thresholds, anomalies, the metric that quietly drifted for five weeks, checked against your governed models every morning. One watches margin by branch and pings the regional VP the morning a threshold breaks – and says nothing at all on the mornings nothing moved.
Report
Standing reports refreshed on schedule, so the meeting starts with everyone looking at the same thing.
Act
Route it, escalate it, open the ticket. One reads the exception queue you have eyeballed by hand for six years. Action becomes part of the analysis rather than a separate effort.
Returns watch · daily 06:00
Security and governance
Inherits (not replaces) your model’s security.
It runs as the person who asked
Open a live report and the queries re-run under that person’s own identity – your source applies their permissions and their row-level security before a row comes back. If they cannot see it in Power BI, they cannot see it here.
Every action leaves evidence
You are opening up who can build, not who can see. Governance ends up with more visibility than it has today – every report, who owns it, who it reaches, every query and who ran it – and more places to apply policy: who can publish, who can share outside the tenant, which models are open to questions at all.
Bring your security team to the first meeting. Their questions are the ones this was built to answer.
Governance – last 30 daysIllustrative
What we do around it
Faucet is the faucet. We are still the plumbers.
No semantic model yet?
Then that is where we start. It is the difference between an AI that answers and one that guesses confidently.
Models built for humans, not for AI
Naming, descriptions, measure hygiene, and the business guidance an agent reads before it writes a query. Why that difference breaks most AI rollouts.
Not a Microsoft shop?
Databricks, Snowflake, something older and more stubborn. We meet your data where it already lives.
The richest starting library you have seen
We build the first set with you, on your data, so day one looks like month six.
Agents that go and do the work
Standing agents that hold a piece of your operation and escalate when it moves. Worth discussing early.
Something else entirely
Strategy is not a separate invoice here, it is how we start everything.
Why this is possible now
The era of product compromise is ending.
You buy software that does some of what you need, then reshape the company around the rest. That trade made sense when building the difference was slow and expensive. AI ended that math. We build the experience your business actually needs, instead of integrating someone else’s product as far as it will let us.
No lock-in
Your data stays put, so you’re not locked in.
Nothing moves
Faucet reads your models where they live. No copy, no migration, no new warehouse.
Reports are web pages
Published reports are HTML, not a format only one vendor can open.
Security was never ours
Permissions and row-level security stay in your source. Switch Faucet off tomorrow and nothing about them has changed.
Claude Partner Network
P3 Adaptive is Certified in the Claude Partner Network.
Our people hold Claude certifications, and we’re building from here.
The Claude Partner Network is how Anthropic works with the firms that put Claude into production for customers. The Services Track recognizes partners for building a certified delivery team, delivering Claude use cases for clients, and proving the work through public customer references. Partners advance through Certified, Select, Preferred, and Global Premier tiers.
Want to try Faucet on your own data?
Bring a model and a question you have never been able to answer.