> ## Documentation Index
> Fetch the complete documentation index at: https://docs.thinknoesis.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Knowledge Synthesis: Merge Branches into Insights

> Select two or more branches in Noesis to generate a unified summary. Surface contradictions, reinforce knowledge, and jump back to any source node.

Branching lets you explore multiple directions simultaneously — but eventually you need to bring those threads back together. **Knowledge Synthesis** is Noesis's built-in tool for doing exactly that: select two or more conversation nodes, trigger a synthesis, and the AI generates a unified summary that merges the key insights from each branch, highlights where they agree, and surfaces any contradictions worth resolving. The result is a new node in your conversation graph — a consolidated piece of understanding that links back to every source that shaped it.

## What Is Synthesis?

Synthesis is the process of taking the outputs of two or more distinct conversation branches and merging them into a single, coherent summary. Rather than manually reading through each branch and writing up your own notes, you let the AI do the heavy lifting: it reads all selected branches in full, identifies the common threads and the tensions, and produces a structured synthesis document.

The synthesis node becomes a first-class citizen of your [Concept Map](/features/concept-map) — it appears as a distinct node type with edges connecting back to each of its source branches. You can navigate back to any source at any time.

## The Synthesis Workflow

<Steps>
  <Step title="Open the Concept Map">
    Click the **Map icon** in the toolbar to open the Concept Map canvas. You need at least two nodes (conversation branches or individual message nodes) to run a synthesis.
  </Step>

  <Step title="Select your source nodes">
    Click the first node you want to include. Then hold **Shift** and click additional nodes to add them to your selection. Selected nodes are highlighted with a distinct border. You can select as many nodes as needed — two is the minimum.
  </Step>

  <Step title="Trigger the synthesis">
    With your nodes selected, click the **Synthesize** button that appears in the floating action toolbar above the canvas. Noesis sends the full contents of all selected branches to the AI for synthesis.
  </Step>

  <Step title="Review the synthesized output">
    The synthesis result appears as a new node on the canvas and opens in the chat panel. It contains a unified summary merging the key insights from each source branch, with explicit callouts for areas of agreement and contradiction.
  </Step>

  <Step title="Navigate the source switcher">
    Below the synthesis output, use the **source switcher** to paginate through each contributing branch. Click any source reference to perform a **focal shift** — jumping directly back to that original branch in the chat view so you can re-read the full context behind a particular insight.
  </Step>
</Steps>

## Synthesized Source Switcher

The source switcher is a compact navigator at the bottom of every synthesis node. It lists each contributing branch as a numbered source. Clicking a source does two things simultaneously: it highlights the relevant contributing content within the synthesis, and it cues up that branch in the chat panel so you can drill back into the original conversation. This makes it easy to fact-check the synthesis or explore a specific thread in more depth without losing your place.

## Focal Shift

**Focal shift** is the ability to jump from a synthesis node back to any of its source branches in a single click. If the AI's synthesis references a conclusion from Branch 3, clicking that reference in the source switcher instantly navigates your chat view to Branch 3 at the exact point where that conclusion was reached. This keeps your original research fully accessible even as your concept map grows more complex.

## Resolving Contradictions

When you synthesize branches that explored competing hypotheses or used different AI models, you will often find that the outputs disagree. The synthesis AI is specifically prompted to surface these contradictions rather than paper over them. Look for sections labeled **"Tension"** or **"Conflicting outputs"** in the synthesis result — these are your most valuable signals, pointing to genuine ambiguity worth investigating further.

You can then branch from the synthesis node itself to run a targeted follow-up that resolves one specific contradiction, creating a new layer of structured inquiry.

<Note>
  Knowledge Synthesis is available on **all Noesis plans**. There is no limit on the number of synthesis operations you can run, and you can synthesize any nodes in your conversation graph regardless of how old they are.
</Note>

<Tip>
  One of the most powerful synthesis patterns is to ask the **same research question on multiple AI models** — say, Claude Opus 4.8, GPT 5.5, and Gemini 3.1 Pro — each on its own branch, then synthesize all three branches. Because each model has different strengths and training biases, the synthesis often produces richer, more balanced insights than any single model would alone. This is especially useful for mapping out competing schools of thought or stress-testing a hypothesis against different analytical perspectives.
</Tip>