Skip to main content

Agentic chat

Introduction to agentic chat in Sana Agents

R
Written by Rasmus Lindberg
Updated this week

Introducing our next architectural paradigm

Sana Agents has evolved with a completely redesigned architecture that transforms how you interact with enterprise knowledge and complete tasks. This next generation platform goes beyond simple search and chat to deliver agents that can understand complex requests, reason through multiple steps, and help you complete your work.

How the new architecture works

The new Sana Agents architecture includes new, powerful capabilities:

  • Planning: The first step the agent does is to create an outline for all steps needed to complete the user's request. During the job, it outlines for the user what it's doing, and its progress. As it completes tasks, you can follow along how it arrived at answers and solutions

  • Multi-step reasoning: The system can perform multiple passes to update its plan, better understand context, diving deeper into relevant information before providing answers, or use new tools

  • Tool integration: Sana can now use specialized tools to complete tasks across your connected systems, from sending API requests to writing and executing code for data science

  • Document comprehension: Sana can now fetch and understand full documents rather than just snippets, providing more comprehensive insights, and varies less on configurations like chunk sizes

Standard and Pro modes

Sana Agents offers two modes to suit different needs:

Standard mode:

  • Faster responses for everyday questions

  • Balances speed with intelligent reasoning

  • Perfect for quick information retrieval and straightforward tasks

Pro mode:

  • Activates deeper reasoning capabilities

  • Plans and executes multiple searches across your knowledge sources

  • Can process larger amounts of information for complex tasks

  • Ideal for in-depth research and multi-step workflows

  • Can decide to take a fast pass when deeper plans are not needed

What this means for you

With this architectural advancement, Sana Agents becomes more than just a search tool—it becomes a true productivity partner that can:

  • Handle complex workflows that previously required manual effort across multiple systems

  • Provide more accurate and comprehensive answers grounded in your organization's knowledge

  • Adapt to your specific needs and work patterns

  • Complete tasks autonomously while keeping you informed of progress Sana AI

Frequently asked questions

How is this different from the previous version of Sana?
The new architecture represents a fundamental redesign that enables Sana to reason through complex tasks in multiple steps, rather than just retrieving information. It can now understand context better, use specialized tools, and handle more sophisticated workflows.

Will I notice a difference in how I interact with Sana?
The interface remains intuitive and easy to use, but you'll notice more comprehensive answers, better understanding of complex questions, and the ability to complete multi-step tasks. The system will also be more transparent about its reasoning process.

When should I use Pro mode?
Use Pro mode when you need deeper research capabilities, are working with complex questions that require synthesizing information from multiple sources, or when you need Sana to perform multi-step tasks.

How does Sana protect my data with these new capabilities?
We remain committed to the highest standards of data security, as detailed in our documentation. The agentic chat follows the same principles as before.

Can Sana now integrate with more of my tools and systems?
Yes, the new architecture makes it easier to add new integrations and tools. With this launch, we're introducing a suite of new connectors, with many more to come.

Did this answer your question?