Step 1
Frame the research question
The workflow captures the question, expected output and boundaries that should shape the research task.
What you can see
A clear task gives later review a defined purpose.
Product tour 02
Let AI help with the research without giving it the final word
This tour retrieves approved material, asks a selected model to organise what it finds and then pauses for a subject-matter expert to review the work before anything continues.
Research question
Evidence + AI
Expert review
Interactive walkthrough
Five steps from a research question to a reviewed result
The workflow helps the reviewer; it does not replace professional judgement or regulatory responsibility.
Step 2
Retrieve approved evidence
Pūnaha searches the selected knowledge collection for material relevant to the question.
What you can see
The organisation chooses which sources belong in the approved collection.
Step 3
Prepare a structured analysis
The configured model receives the task and retrieved context, then prepares the requested structure.
What you can see
The output is assistance for review, not automatic authority.
Step 4
Pause for subject-matter review
A person examines the evidence and proposed analysis, then approves or rejects the workflow continuation.
What you can see
The decision becomes part of the workflow state.
Step 5
Continue with the approved result
The accepted material can be merged into the final output or passed to another configured step.
What you can see
Execution history preserves the path taken through the workflow.
Controls in the flow
Make every hand-off easy to see
Approved sources
The workflow searches the selected knowledge collection rather than treating any available content as equal.
Defined model deployment
The AI step uses the supported deployment selected for this task and environment.
Human decision
The workflow cannot cross the approval step until the configured reviewer acts.
Versioned process
The published workflow version and execution history show which process was used for the work.
Bring a research process where the review really matters
Show us the evidence sources, AI assistance and reviewer decisions you want to understand.