Flow AI Assistant (closed beta)
All Flow AI features, including the Flow AI Assistant, are built under Appfire’s AI initiatives and strictly adhere to our enterprise-grade policies and security standards.
Visit Appfire’s EULA > paragraph 17 to learn more.
The Flow AI Assistant feature is currently in a closed beta.
The Flow AI Assistant (or Flow AI chat for short) is your strategic partner who can assist you with translating raw data into a clear narrative of team success. Whether you are prepping for a board meeting or a 1:1, the Flow AI can help you maximize the impact of the Flow app.
Access and permissions
The Flow AI Assistant is disabled by default. It can be enabled by a Flow admin in Settings > Report Settings > Configurations > AI Features. Until enabled, no AI features can be used.
The AI Assistant requires the AI Insights permission.
What does Flow AI Assistant do?
When you interact with the Flow AI Assistant, it retrieves metrics data from the same Flow APIs that are used in Flow reports, including:
Git metrics (commit data, PR statistics)
Ticket data (sprint metrics, delivery performance)
Team and user metadata (names, team memberships)
The AI Assistant respects your existing Flow permissions. It cannot access data you don't already have permission to view.
What happens to your data?
Element | How it works |
|---|---|
AI Model | Requests are processed by Claude (Anthropic) via AWS Bedrock. |
Prompt processing | Your prompts and Flow data are sent to the AI model for response generation. |
Conversation history | Prompts are encrypted and temporarily stored in AWS S3 to enable multi-turn conversations. |
We do not use your data to:
Train models: Your prompts and responses are never used to train our AI models.
Share your data with model providers: Your inputs and outputs are never shared with third-party model providers (for example, Anthropic or OpenAI).
What does Flow store?
Flow AI stores the following:
Conversation sessions: Your conversations are encrypted and stored temporarily in S3 for session continuity.
Analytics/telemetry: Your aggregated usage metrics (with PII redacted) are stored for product improvement.
Flow AI Assistant infrastructure
The AI Assistant runs on Flow's existing AWS infrastructure:
AWS Bedrock: Managed AI service.
AWS S3: Session storage (encrypted at rest).
Flow API: Data retrieval through existing authenticated endpoints.
Flow maintains SOC 2 Type II and ISO 27001 certifications. The AI Assistant inherits these controls.
MCP (Model Context Protocol) access
For users connecting using MCP clients (for example, Cursor, Claude Desktop):
Authentication uses Flow API keys.
All security controls remain in effect.
Visit the MCP configuration page for more.
Engage with the Flow AI assistant
The AI Assistant is located in the right-upper corner, under the Chat button.
Click it to open the chat window.
A chat window displays.
You can begin your conversation using one of the four pre-defined chat prompts:
Help me prepare for 1:1
How did the last Sprint go?
Let’s plan a retrospective meeting
How can I get the most out of Flow?
Or, ask your own question in the text field at the bottom of the chat screen.
Try starting with:
"How is my team’s workload looking this month?"
"What was our average Time to merge over the last 30 days?"
"Can you show me our DORA metrics for Q1 compared to Q4?"
What data can the Flow AI Assistant provide?
The AI Assistant can assist you with the following:
Metrics and insights
Data is most powerful when it tells a story. Flow AI helps you dive into the specific behaviors that drive your delivery:
Code metrics: Gain visibility into the daily pulse of development. Flow AI analyzes Coding Days to visualize focus time, uses Impact to measure the technical weight of changes, and tracks Efficiency to balance new code against churn.
Pull Request metrics: Identify exactly where the review process stalls. By monitoring Time to First Comment and PR Iteration Time, Flow AI ensures your team is not waiting on feedback, while Thoroughly Reviewed PRs help maintain high-quality standards.
Collaboration metrics: Measure the "connective tissue" of your team. Flow AI tracks Reaction Time and Responsiveness to see how effectively engineers support one another and maintain momentum.
Ticket metrics: Bridge the gap between code and delivery. Flow AI analyzes Cycle Time and Queue Time to identify workflow bottlenecks, and uses Backflow Rate or Jitter to spot work being sent back or stalled due to shifting requirements.
DORA metrics: Elevate engineering maturity by tracking the gold standard of DevOps. Flow AI pulls Deployment Frequency and Lead Time for Changes to measure speed, alongside Change Failure Rate and MTTR to ensure reliable scaling.
Teams and people
Building software is a human endeavor. Flow AI helps you stay connected to the people behind the code:
Contributor deep-dives: Flow AI quickly retrieves specific stats for any group or individual, making it easy to prepare for performance reviews or team health checks.
Trend identification: Flow AI spots long-term shifts in behavior, such as gradual burnout risks or sudden surges in collaborative activity, allowing you to offer support where it’s needed most.
Planning and investment
Align your engineering effort with your business goals:
Sprint analysis: Flow AI reviews your execution at the end of each cycle with detailed ticket breakdowns, helping you understand what was committed versus what was actually delivered.
Investment profiles: Get a clear picture of resource allocation. Flow AI categorizes work by Investment layer so you can see exactly how much time is spent on "New Features" versus "Technical Debt" or "Maintenance."
Documentation and report expertise
Thinks of the Flow AI as your personal platform consultant who can explain the "why" behind any metric or report. If you’re unsure how a specific metric like Legacy Refactor is calculated or which report best visualizes Cycle Time, Flow AI provides instant clarity and context.