Skip to main content

Releases

View 2024 releases

Changelog

Subscribe to updatesRead more from Hex
Latest release:

🗓️ Introducing Agent Tasks​

Schedule the Hex Agent to answer your recurring questions automatically and deliver results straight to Slack or email.

Most insights aren't one-offs. The same questions come back every week: before the Monday standup, before the all-hands, before every sales call. Agent Tasks handles them for you.

Some ways teams are using it:

  • Get a summary of what shifted in your revenue dashboard every Monday morning before standup without having to log in
  • Consolidate activation and usage trends into a single weekly update, then reply in the Slack thread to dig into a segment
  • Run a check on your key funnel metrics every morning with suggested reasons behind any sudden change

Want more prompt templates and worked examples? Check out the Tasks cookbook →

Getting started: Start from Threads to draft a solid prompt, and you'll see what the output looks like before committing to a schedule.

Be specific about format (charts, bullets, brief summary), and test the run before turning on the schedule.

🌐 Web search for the Hex Agent​

Hex already knows what's happening inside your business, now the Hex agent can fill in the gaps with what's happening outside.

The Hex agent can search the web, powered by Parallel.

Ask the agent an open-ended question and it'll search the web and synthesize an answer alongside your data. Or paste a URL and the agent will extract what's relevant directly into your analysis.

Web search is on by default for Professional, Team, and Enterprise plans. Admins can turn it off in AI & Agent settings.

🎛️ Choose your model​

Now you can pick which model runs the Hex agent. Easy task? Use a cheaper model. Beefier, open-ended task? Reach for the best reasoning. Switch in personal settings, or mid-conversation in Threads.

choose your model

Speaking of...

🧠 Fable 5 in Hex​

Disclaimer: on June 12th, Anthropic disabled Fable 5 in response to the US government’s export control directive. Because of this, Fable 5 is currently unavailable in Hex.

Fable is Anthropic's newest frontier model and the first to meaningfully improve at complex analytical reasoning. So much so, we needed to build new evals. It's now available in the model picker.

To use it, an admin must enable model data retention in AI & Agents -> Settings.

Anthropic retains conversation data for a limited time for safety monitoring; we recommend reviewing with your security and legal teams first.

🗄️ Hex is now in Codex​

At Hex, we want you to be able to analyze data from wherever you do your job. And we’re working hard on building our leading analytics agents and our unique approach to context into the places our customers are working.

Starting today, you can bring Hex directly into Codex to run analyses, pull in existing work, and spin up new projects.

💪 What you can do​

  • Kick off Hex analysis from Codex - Need to understand churn by segment? Dig into pipeline performance? Codex can invoke Hex to pull the data, run analysis, and return results without switching tabs.
  • Pull Hex Threads into Codex - If you've already done analytical work in Hex, you can pull in your prior Hex work to incorporate data into whatever you build.
  • Control Hex programmatically using the CLI - For more specific, structured tasks - such as creating projects, adding cells, or configuring analysis - you can leverage the CLI connection in Codex for a more precise interface into Hex.

To use Hex in Codex, head to Plugins and install the Data Analytics plugin. Then connect to Hex and you're ready to go!

🔌 Hex is now an MCP client​

Your data tells you what happened. The why lives somewhere else — in strategy docs, meeting notes, Slack threads, customer conversations.

Today, the Hex agent can read all of it.

Connect any tool that exposes an MCP server to the Hex agent.

As part of our public beta, we’re launching with two named connectors: Notion & Linear. Other MCP servers can also connect to the Hex Agent, and we have more named connectors in the works.

That means when your team asks a question in a Thread, the agent isn't just looking at your data — it's cross-referencing the Notion doc from the product launch, the Linear ticket that closed last week, the strategy doc your team updated before the weekly sync.

👷 How it works​

  1. Go to Context Studio → Apps → Add MCP server
  2. Add Notion, Linear, or any tool that exposes an MCP server
  3. Once enabled, users connect their personal accounts via the + menu in any thread

Connecting more apps is only half of it — the agent works through semantic models and definitions your team has already validated, so the answers are consistent and can be trusted.

Available today in public beta for Team and Enterprise plans.

🛌 Signed embedding for Generative apps​

You can now use the Hex Embed API to securely embed Generative Hex apps in your web application with pass-through authentication and row-level security.

This brings a whole new level of customization to embedded Hex apps. You can match the visual style of your brand or web app with the help of the Hex agent!

A great way to jump-start this work is to copy and paste your own css or a screenshot of your web app into the prompt bar and the agent will do the rest.

A side-benefit of this is that app interactions are way snappier than classic Hex apps — filters, drill-downs, and re-renders feel near-instant, which makes a real difference for embedded experiences.

Learn more in our docs about signed embedding with Hex.

🗂️ Connect repos as agent context​

Your data doesn't exist in a vacuum. Your dbt models, transformation logic, and app instrumentation hold a lot of context that makes your warehouse data actually legible. Now the agent can use it.

Connect one or many repos to your Hex workspace and the agent will reference them when answering questions — understanding how a table was built, what upstream logic shaped it, or how a product event was implemented. Admins can connect repos via API; once uploaded, they appear in the admin panel with a last-updated timestamp, and the agent surfaces its retrieval steps directly in Threads.