What's new in Toucan: June 2026 product update

Seven updates are shipping now, the largest of our recent releases. They span the AI agent, the chart library, and the platform controls that keep AI analytics governed as you scale it across customers.

Here is what changed, and why each one matters if you are embedding Toucan in your product.

AI agent

Choose your LLM provider and model

Org admins can now choose which LLM provider and model power their AI experience. Pick Anthropic, Mistral, or others, set the model your organization runs on, and switch anytime or fall back to the default.

For an ISV, model choice is not a detail. It affects answer quality, cost per query, and sometimes which providers your own customers will accept in their stack. Now that decision is yours, not ours. The agent is mainly tested with Claude Haiku and Devstral, so changing the model can affect both response quality and credit consumption.

Click an element, edit it directly

In AI mode, you can now target any chart or element directly on the dashboard, and the conversation locks onto it. The panel shows exactly which element you are editing, with the direct click-to-edit feel of a modern AI builder.

We also removed the split between a dashboard conversation and a separate chart conversation. One panel for the whole dashboard. Cleaner, faster, and more AI-native.

Platform

Certified Answers

Certified Answers lets admins analyze the recurring questions your end users ask, group them into patterns, and map those patterns to pre-built charts. Metrics are coming next.

This is how you move from ad hoc AI responses to consistent, governed answers at scale. When fifty users ask the same question five different ways, you define the answer once, certify it, and everyone gets the same chart. Lower AI cost, higher trust, and a continuous read on what your users actually want.

Charts

Three new chart types

Three new visualizations: line charts with area, bubble charts, and stacked area charts. More ways to show what the data actually means, not just more chart icons in a menu.

Down to the minute

Time aggregations now go down to the hour and the minute. Built for real-time use cases: manufacturing, operations, anything where seconds matter.

If your customers run operational dashboards, like a factory watching machine output, minute-level granularity is the difference between catching a 14:17 spike and missing it entirely.

Heatmaps with color gradients

Heatmap cells now color by metric value on a continuous min to max gradient. Patterns that were invisible in a grid of numbers become obvious at a glance, like which team and which hour your support tickets actually cluster in.

Dashboard

Search your Library

You can now search your Library by dashboard name, directly from the Library page. It is client-side and instant, with no page reload and nothing persisted in the URL.

Once a customer has more than a handful of dashboards, search stops being a nice-to-have. This keeps the Library usable at the scale your best accounts will reach.

Ship analytics your customers actually use

All seven updates are live. If you already embed Toucan, they are in your environment now.

If you are still weighing build versus buy for in-product analytics, here is the short version: Toucan is AI-native embedded analytics for SaaS products, multi-tenant by design, with your first dashboard shipping in days rather than quarters.

Try for free Book a demo