Reveal AI: Bring Your Own Model, Across Every Platform

Today we are shipping the biggest update to Reveal AI since we introduced it. Reveal AI now works across every platform we ship — .NET, Node, and Java at full feature parity — and it works with the AI model you choose, including your own.

What Is New in Reveal AI

• Full feature parity across .NET, Node, and Java. The same conversational analytics experience, wherever your application runs.
• Bring your own model — OpenAI, Google, and Anthropic are supported out of the box, and you can plug in your own provider, including local models.
• Smarter dashboard generation. A new multi-step approach produces more accurate dashboards from natural-language questions, with better handling of sorting and date ranges like a specific quarter or month.
• Support for longer-running and reasoning-heavy AI models, so complex questions do not get cut off mid-thought.
• Per-tenant AI data refresh — for SaaS teams running many customers from one deployment, AI data can refresh for one tenant without disrupting the others.

Bring Your Own Model

This is the change we have heard the loudest request for from customers.

Some of you have negotiated enterprise agreements with a specific AI provider. Some of you serve customers in regulated industries where data residency or model provenance is non-negotiable. Some of you want to run an open-source model locally to control costs or to keep data on your own infrastructure. Some of you simply do not want your analytics SDK picking your AI vendor for you.

With this release, Reveal AI supports OpenAI, Google, and Anthropic out of the box. You can also plug in any other model you want — including local models. The configuration is direct and documented. You choose the provider, the model, and the parameters that match your customers’ requirements, not ours.

If you have a customer who insists on a specific AI provider before they sign — that conversation just got easier.

AI Now Matches the Rest of Reveal — Wherever You Build

Reveal AI used to be .NET-first. That is no longer the case. With this release, our Node and Java SDKs now ship the full conversational AI experience at feature parity with the ASP.NET SDK.

If your product runs on Node or Java and you have been waiting for AI to come to your stack, the wait is over. Same conversational analytics, same dashboard generation, same configuration model — in the runtime your product already uses.

Smarter Dashboard Generation

A natural-language question is the easy part. Turning it into the right dashboard is the hard part — and historically, the part where conversational analytics tools have made the most embarrassing mistakes.

In this release we changed how Reveal AI builds dashboards from natural-language questions.

The new multi-step approach reasons through the question in stages — what is being asked, what data is relevant, what visualization fits, what filters apply, what time range applies — and then produces the dashboard. Sorting and date-range handling, in particular, are significantly more accurate. “Show me revenue for Q2 by region” now does what a human would expect.

In practice, this is the kind of change you do not notice when it is working. You only notice it when conversational AI used to fail in front of your customer, and now does not.

Built for SaaS

If you embed Reveal into a multi-tenant SaaS product, two things in this release were built for you specifically.

First, AI data refresh is now per-tenant. You can refresh the AI’s view of one customer’s data without disrupting the others. No more global-refresh blast radius.

Second, longer-running and reasoning-model support means the more complex questions your power users ask do not time out partway through.

What This Means for You

Existing Reveal AI customers: your integration keeps working. Take a look at the provider configuration — if you have been wanting to switch to a different AI model, this is the moment.

Existing Reveal customers not yet using AI: Reveal AI is now available on whichever platform you are on — including Node and Java. If you have been holding off, the eval just got easier.

Evaluating Reveal for the first time: you are evaluating an embedded analytics SDK that will not pick your AI vendor for you. Bring the model you trust and let your team focus on the rest of the build.

Also in This Release

Beyond Reveal AI, this release adds three new data connectors and pushes hard on accessibility:

• New connectors: SQLite, DuckDB (with MotherDuck support), and Cube.dev. More of the data sources our customers asked for.
• Accessibility and keyboard navigation: keyboard accessibility is on by default, screen-reader support is improved across the experience, and keyboard navigation in viewing mode and the export dialogs is significantly better.
• Stability: since Reveal 2.0 in May, we have closed more than 80 issues in the core SDK, including over 35 bug fixes. The full list is in the release notes — and if you reported one of them, thank you.

Getting Started

Existing customer: the Reveal AI configuration guide is updated. Switch providers, plug in your own model, or stand up multi-tenant AI refresh — all in your existing project.

New to Reveal: Request a Demo or request access to our free trial.

On Node or Java: Reveal AI is now generally available on your stack.