Creatio News

Creatio AI Studio, October 8, 2026: Dictation in AI Twin, Draft-First Agent Changes and Metrics in AI Credits

The Creatio AI Studio update of October 8, 2026 adds dictation in AI Twin and in the test chat, agent changes saved as drafts by default, per-environment tool status for MCP integrations, and run metrics in AI credits. Here is what changed and what it means for companies.

📅 November 3, 2026 ⏱ 5 min read ✍ Optingo editorial
Creatio AI Studio, October 8, 2026: Dictation in AI Twin, Draft-First Agent Changes and Metrics in AI Credits

Creatio AI Studio is the web interface of the Creatio agentic platform for building, deploying and governing AI agents. We wrote about it in the context of Creatio 10X and on our Creatio AI Studio page. In October 2026, Creatio published three AI Studio updates: on October 8, 15 and 22.

Below is the first one. The full official notes are in the October 8, 2026 release notes. The notes do not mention feature flags, licensing requirements or beta labels for any of the changes.

Dictation in AI Twin

In the message box of the AI Twin chat (the assistant that helps you build and change agents), you can now speak instead of typing. Click Dictate, record your message and choose Stop and insert text. The transcribed text lands in the message box, where you can edit it. Discard recording lets you start over. Nothing is sent until you send it yourself.

Dictation also works in an agent’s test chat, where you check how an agent responds before it reaches users. The cost of speech-to-text is recorded in usage, in AI credits.

What it gives you: longer task descriptions for an agent (instructions, test scenarios) are easier to dictate than to type, and a person still approves the text before it is sent.

Agent changes saved as a draft

When you ask AI Twin to create, change or fix an agent, the changes are saved as a draft by default. If you did not ask for deployment, AI Twin offers it as the next step. It publishes and deploys only when you ask, to the current environment or one you name.

What it gives you: a safer default. An agent that people use does not change during a conversation with the assistant until someone deliberately deploys it.

Skills

  • Size limit for skill instructions. The SKILL.md file can be up to 157,856 characters. If a skill is longer, move the details into files in the references/ folder.
  • Skills in realtime voice conversations. Agents can use their attached skills in realtime voice conversations as well.

Testing agents

Besides dictation in the test chat, the time limit for multi-turn cases changed: each user turn gets 120 seconds, so a three-turn case can run for up to 360 seconds.

MCP integrations: tool status per environment

An MCP integration can have a different URL in each environment (for example test and production). The Tools tab now shows which tools are ready in which environment. Two statuses appear:

  • “Not yet discovered — run Discover Tools for this environment.”,
  • “Not reachable in this environment.”

What it gives you: differences between environments are visible right away, so you don’t accidentally deploy an agent that has no access to its tools in production.

Observability and usage

  • The Model calls tile shows the total number of model calls, including those initiated by the platform (such as memory updates). When there are any, a line below the total splits calls between the agent and the platform.
  • The session duration tile is now called Active duration.
  • Run and session metrics show usage in AI credits: in the tiles, the model call details and the session list.

What it gives you: it is easier to tell the cost of the agent’s own work from the cost of the platform’s supporting operations, and to see it in one currency.

What to check after the update

  1. Long skills. If one is close to the 157,856-character limit, move supporting material to references/.
  2. MCP integrations across environments. Run Discover Tools in every environment where the integration is going to be used.
  3. Budgets. Look at the metrics in AI credits before you set limits for production agents.

Want to see how this fits your company?

As a certified Creatio Level IV partner, we help design AI agents with access control, human approval and predictable cost. If you are considering your first agent, or want to put the existing ones in order, get in touch.

Other Creatio AI Studio updates

See also the other Creatio AI Studio updates from October 2026:

Want to see Creatio live?

Book a free 30-minute demo tailored to your industry.