Most AI research tools follow a predictable, frustrating pattern. You ask a complex question. The system disappears. Several minutes later, it returns with a polished answer and a list of links.
You cannot see what it prioritised, what it ignored, or whether it actually followed the brief you intended. You are forced to trust a finished output without seeing the work that produced it.
Our latest release is a rebuilt Deep Research experience designed to solve this lack of visibility. It works inside your existing Threads, alongside the models you already use, and gives you total control over the investigation. This is research you can direct, not just receive.
How do you choose the right Deep Research depth in Euryka?
Not every task requires a full-scale investigation. Some questions need a quick, focused look, while others require an exhaustive deep dive. In the Euryka composer, you can now toggle Deep Research on for a specific message and select your tier:
Research Standard: A focused investigation designed for well-defined questions.
Research Max: Broader source coverage and a more thorough, comprehensive report.
Deep Research only runs for the messages where you enable it. Your Thread stays on your selected model for everything else, keeping your workflow efficient and cost-effective.
How can you plan a Deep Research investigation before it runs?
For complex briefs, you can now enable collaborative planning. Instead of searching immediately, Deep Research proposes an investigation plan for your review.
This allows you to sharpen the direction, add specific questions the report must answer, or remove research paths that will not be helpful. This is particularly valuable for teams. A strategist can set the direction, a colleague can refine the questions, and the entire team can agree on what "good coverage" looks like before the system starts gathering material.
How can you monitor an AI research investigation in real time?
As the investigation begins, a new Artifacts panel brings the entire process into view. You can follow the report as it develops in real-time.
The panel streams the sources being used, the reasoning steps being taken, and any visuals being created. There is no more waiting for a final document to understand where the research is heading. If you see the system drifting off-track, you know immediately.
Why does building AI research inside your Thread improve confidence in results?
The real value of AI research is not just how fast it happens. It is the confidence you have in the result.
By making research part of the Thread where your questions and decisions already live, Euryka turns a black-box output into a collaborative process. You gain confidence that the investigation covered the right ground and that your team had a chance to shape the brief.
Building Block by Block
Each release brings us closer to a unified creative ecosystem. Recent updates paved the way for this transparent research experience.
Auto-routing and model tiers ensured that your research always uses the right level of power. Before that, model-agnostic infrastructure gave you the freedom to switch engines as winners changed. And Euryka Flows provided the visual canvas to turn your research into repeatable, on-brand campaigns.
