AI Investigator
Find what’s actually slowing delivery down.
Nave’s AI Investigator reads your computed flow metrics, identifies the stage and work items behind the problem, shows you the evidence, and tells you where to look and what to fix.
The problem
Your dashboard tells you something changed. Now what?
- Cycle time is high.
- WIP is increasing.
- Flow efficiency has dropped.
- A process step is taking longer than normal.
The metric tells you there may be a problem. But someone still has to open the charts, inspect the workflow, find the relevant items, compare the evidence and determine what’s driving it.
“When I notice that Cycle Time has increased, for example, I don’t immediately conclude that the team has become less productive. I will investigate,” she says. “Where did this increase happen? At which stage? Which items are driving this result?”
Joelma Castro Agile Master and Squad LeaderThat’s the investigation Nave does for you.
One investigation
One real investigation, four steps.
From the first sign of a problem to a specific place to intervene.
Problem
Something is wrong.
“18 items are stuck in Ready for SIT”
Investigation
Here is the evidence.
“Arrival rate into Ready for SIT is 0.62/day but throughput out is only 0.43/day”
Finding
The constraint is downstream SIT capacity.
“SIT processes at 0.43/day and cannot absorb what Code Review produces”
Action
Here is where to intervene.
“Investigate what limits SIT throughput to 0.43/day”
What it does
It doesn’t just summarize the dashboard.
Generic AI
“Cycle time has increased. Consider reducing WIP.”
Nave
“18 items are stuck in Ready for SIT”
“Arrival rate into Ready for SIT is 0.62/day but throughput out is only 0.43/day”
“Investigate what limits SIT throughput to 0.43/day”
Reads the metrics
The Investigator starts from the flow metrics Nave has already calculated.
Finds where the problem is
It identifies the process stage associated with the issue.
Finds the work behind it
It identifies the individual work items contributing to the problem.
Shows its evidence
Every conclusion is supported by the underlying delivery data so you can see why Nave reached it.
Gives you somewhere specific to act
Instead of telling you to “improve cycle time,” it points to the constraint or work that deserves attention.
“I encourage teams to pay closer attention to the details hidden within their charts. These details might not seem significant at first, but they can actually offer valuable insights that can positively impact business outcomes.”
Zoran Vujkov Agile Coach, United CloudEvidence-based findings
It doesn’t force a finding.
Celebrate insights identify what is working and suggest next steps to protect it. Open See why for the analysis and numbered evidence.
Every insight is labeled Urgent, Monitor, or Celebrate.
- Urgent
- Monitor
- Celebrate
The built-in instructions tell the Investigator not to investigate to fill a quota. It can return no findings, shown as an empty panel.
Evidence first
The numbers are computed, not generated.
The AI only reads them, so swap the model and the numbers don’t change.
- Your delivery data
- Nave computes the metricsCycle timeWIPThroughputFlow efficiency
- AI Investigator reads the computed evidence
- Finding + evidence + suggested action
- AI does not calculate your flow metrics.
- AI does not decide what counts as a statistical signal.
- AI investigates the evidence Nave has already computed.
Using it
Open the Investigator where you already review delivery.
- Where do I launch it?
- Open the Insights tab in the right-hand panel of a team dashboard, beside the Process Improvement dashboard and flow charts.
- What can it investigate?
- It investigates the computed flow metrics, stage statistics, work items, item ages and blockers for one team dashboard at a time.
- What date range/data does it use?
- It uses the dashboard’s selected date range, which is the last three months by default. You can also set it to follow the chart filters.
- Can I run it again?
- There is no run button. A new investigation starts automatically when the dashboard, date range, working-hours setting or followed filters change.
Need the findings outside Nave?
Fetch Investigator findings through Nave’s API or MCP server and bring them into your own tools and compatible AI assistants.
Explore API & MCPHuman oversight
Nave investigates. You decide.
What Nave can tell you
- what changed
- where the problem appears
- which items contribute
- what the evidence says
- where to investigate or intervene
What you still decide
- whether intervention is appropriate
- what organizational context matters
- what change to make
Improvement loop
Investigation is one part of improvement.
Detect
Nave’s calculations identify a meaningful signal.
Investigate
AI Investigator finds the evidence behind it.
Intervene
You decide what to change.
Measure
Use the next period of delivery data to see what happened.
Data handling
How Nave handles data sent to AI.
The model receives computed metrics, stage statistics and the items in each stage with their IDs pseudonymized by one-way hashing. Item names, URLs and descriptions are never sent to the AI provider.
Nave does not train or fine-tune AI models on customer data. AI providers are used through their APIs for inference only.
Nave uses Anthropic and OpenAI as AI providers. Both are listed on Nave’s published sub-processor list, alongside Google Cloud Platform for hosting.
AI is optional and can be switched off for an account. Individual users can also switch it off.
Every AI request and response is stored for traceability, and the AI cache is kept for 30 days. Stored AI results are deleted when the account is deleted.
The output is advisory. People review the evidence and decide whether to act.
Questions
Questions about the AI Investigator
No. Nave calculates your flow metrics before the AI receives anything. The Investigator reads those computed results and investigates the evidence behind them.
No. Nave determines statistical signals through deterministic calculations. The AI investigates the evidence Nave has already computed.
It uses computed flow metrics, stage statistics, blockers and the items in each stage with their age. It uses the dashboard’s selected date range and can follow its chart filters. Work item IDs are pseudonymized before they are sent to the model.
Yes. Open “See why” to view the analysis and numbered evidence behind an insight. Work item names link to the corresponding item in the source tool.
The Investigator analyzes one team dashboard at a time. It reads that dashboard’s computed metrics, stage statistics, work items and blockers, but does not receive Process Improvement limits or statuses.
No. Nave does not train or fine-tune AI models on customer data. Anthropic does not train on data sent through its commercial API.
Nave uses Anthropic and OpenAI through their APIs. We do not identify a specific model or version here.