Appfire AI

Appfire AI is your AI-powered assistant in Agile Poker’s Interactive sessions. It helps teams estimate work by identifying potential risks, challenges, and considerations, providing context-based insights, and suggesting estimations based on available data.

How to enable Appfire AI

At the global level

Appfire AI is enabled by default and is currently in beta. Admins can enable or disable it for all instances through global app settings.

1. Navigate to Manage apps.

2. Find Agile Poker in the apps list.

3. Click Configure.

4. Under Features, toggle Appfire AI on or off.

   Open

   

At the session level

The game administrator can turn Appfire AI on or off when setting up a new game. To enable it, configure the game basics, go to the Manage participants step, and switch on Appfire AI.

Open

How it works

Appfire AI button appears in the participant list of an Interactive session and operates in two modes:

• Manual mode (default): The moderator must click the Ask AI button to request AI-generated insights.

• Auto-trigger mode: During game setup, the moderator can enable Auto-trigger request per issue. In this mode, AI automatically generates suggestions when an issue is selected.

How to use Appfire AI in an Interactive session

1. Select an issue in the session.

2. Click the Ask AI button (if in manual mode) or let AI generate insights automatically (auto-trigger mode).

   Open

   

   Manual mode

3. Once AI completes the analysis, click Show hint to view the AI’s response, which includes a summary, risks, recommendations, and similar issues.

   Open

   

4. If the response isn’t helpful, click the reload button to generate another answer.

   Open

   

5. Provide feedback using the thumbs up or thumbs down buttons to help improve AI quality.

   Open

   

6. Once your team has had time to discuss and enter their estimates, click Show estimates. At this stage, the AI also reveals its numeric estimation alongside team input.

   • The AI’s estimation appears as a number below the AI participant avatar. It’s based on available context, such as the issue summary and description. If there’s little or no input (for example, an empty description), it can suggest 0.

     Open

     

7. Wrap up the process by discussing insights and submitting final estimates as a team.

Data security

We take data security seriously and are committed to transparency in how Appfire AI processes information.

• We leverage technology from Microsoft’s Azure OpenAI Service to provide you with this service. The data sent to the Azure OpenAI service includes the text from issue summaries and descriptions.

• Neither Appfire nor Microsoft stores or uses your data to train AI models. See Appfire’s AI consent form for more details.

• All data is encrypted during transmission. Each customer’s data is kept separate in our production environment. Data from different customers is never mixed or processed together.

• We respect the user’s permissions. The large language model (LLM) that generates AI responses only uses information the requesting user already has access to. It also considers the session's visibility scope when generating estimates.

We hope that you find this feature helpful to your workflow! We appreciate your feedback and suggestions on how to improve it.