AI Panel & Quick Actions

The AI panel combines chat, model selection, attachments, and high-speed planning actions.

Model selector behavior

The chat input shows available models based on current setup:

• TensorPM model (logged-in users): Lite Model or Pro Model
• BYOK models when API keys are valid
• Local AI model when local server is enabled and configured

If no model is available, the selector shows AI and a setup hint.

Chat usage patterns

High-value chat prompts are usually:

• context gap detection
• options + trade-off generation
• transformation into concrete deliverables
• risk and dependency surfacing

Quick Actions

Use the fast-forward button (Toggle Quick Actions) in the chat input header.

Current quick actions include:

• Create items
• Create categories
• Re-evaluate selected items
• Split selected items
• Evaluate project status
• Check (data quality / context)
• Strategize (strategic analysis)
• Analyze (goal coverage)
• Plan (execution planning)

Generation ranges in Quick Actions

Current panel limits:

• item generation count: 1 to 20
• category generation count: 1 to 10

These limits are intentionally small to keep outputs reviewable.

Selection-dependent actions

Re-evaluate and Split require selected Action Items.

Typical setup:

1. Open Actions.
2. Switch to List.
3. Turn on Edit.
4. Select one or more items.
5. Run the Quick Action.

Custom instructions

Quick Actions support custom instructions for repeatable output style and planning constraints.

File attachments

The chat input supports file attachments and drag-and-drop files for context-aware requests.

Send behavior when models are missing

If no model is available, send actions are blocked until one of these is true:

• valid BYOK key is configured
• TensorPM provider is available via login/subscription state
• local AI is enabled and has an active model

Next steps

• Provider and account setup: Account & AI Modes
• Priority workflow: Guidance
• Better base context: Project Creation Wizard