Ask your workspace anything

Ask a question and get an answer from workspace resources without selecting files manually.

Created by Chris Moen • Version 140 • 24 steps

Find answers without the manual hunt

This workflow transforms how you interact with your internal knowledge base. Instead of spending time digging through folders or trying to remember which document contains a specific detail, you can simply ask a question. The system searches through your workspace resources, connects related pieces of information, and provides a clear answer based on your actual data. It removes the friction of manual file selection, allowing you to focus on using information rather than finding it.

How the automation works

The process begins by normalizing your question to ensure clarity. Using OpenAI, the flow generates multiple search query variants to cover different ways the information might be phrased. It then performs several parallel searches across your workspace, including a deep dive into structured transcripts and linked text resources.

Once the system gathers the relevant hits, it merges and hydrates the data to provide context. The Large Language Model (LLM) then extracts specific findings from both the full transcripts and general workspace evidence. Finally, the flow synthesises these findings into a concise markdown response, citing the sources it used so you can verify the details.

Key features and benefits

By automating the research phase of your work, you ensure that no internal resource is overlooked. This flow is particularly useful for product managers, researchers, and team leads who need to synthesise large amounts of documentation quickly.

  • Multi-angle searching: The flow creates several search variations to catch relevant results that a single query might miss.
  • Transcript analysis: It specifically looks for speaker data and structured paragraphs within meeting recordings.
  • Automatic synthesis: You get a formatted answer instead of a list of links, saving you the time of reading through every document.
  • Contextual accuracy: By using your specific workspace resources, the AI provides answers that are relevant to your company's unique projects and terminology.

Reliable internal research

This template uses OpenAI and direct API integrations to bridge the gap between your raw data and actionable insights. It handles the heavy lifting of data retrieval, allowing you to treat your entire workspace as a single, searchable brain. Whether you're onboarding a new team member or preparing for a stakeholder meeting, this workflow provides the evidence you need in seconds.

Steps

  1. Normalize the question (function)
  2. Plan search queries (llm)
  3. Parse search plan (function)
  4. Search workspace for the original question (search)
  5. Search workspace for query variant 1 (search)
  6. Search workspace for query variant 2 (search)
  7. Search workspace for query variant 3 (search)
  8. Merge hydrated search hits (function)
  9. Prepare focused transcript read (function)
  10. Find structured transcript speakers (search)
  11. Find structured transcript paragraphs (search)
  12. Find structured transcript JSON (search)
  13. Prepare structured transcript read (function)
  14. Load linked text resource (http)
  15. Hydrate linked text resource (function)
  16. Extract findings from full transcript (llm)
  17. Extract findings from workspace evidence (llm)
  18. Parse findings (function)
  19. Load focused transcript (function)
  20. Load transcript speaker data (function)
  21. Answer from extracted findings (llm)
  22. Answer from hydrated workspace evidence (llm)
  23. Parse answer (function)
  24. Render markdown output (function)

FAQ

What does the Ask your workspace anything flow do?

This flow searches your entire workspace to answer questions based on your stored resources. You don't need to manually select files because the workflow automatically finds and analyses relevant documents, transcripts, and data for you.

How does the workflow find the right information?

The flow uses a multi-step process to normalise your question and create several search variations. It then gathers information from workspace searches, linked text resources, and structured transcripts to provide a comprehensive response.

Which services and integrations are required?

This automation connects with OpenAI to plan search queries and extract findings from your data. It also uses API integrations and Hypertext Transfer Protocol (HTTP) requests to load and hydrate your workspace resources.

Can I customise the output of this workflow?

You can customise how the workflow interprets your data by adjusting the search query planning and evidence extraction steps. The final output is rendered in markdown, which makes the answers easy to read and share with your team.