Philosophy
Teamtastic is not a always-on chatbot in the corner. The database and your team's actions are the source of truth. AI drafts and explains in context, you review and confirm before anything is saved or sent. That keeps automation helpful without silent changes to tasks or messages.
Where AI appears
- Task creation: draft a description from title, assignee, and due date.
- Wiki: generate a starting outline from a short prompt.
- Handover notes: when reassigning a task, draft context from recent activity.
- Insights: plain-English explanation of why a task is at-risk (managers).
- PTO: summarize pending leave requests for managers.
- Suggest reply: draft DM or comment replies you edit before sending.
- Digests: optional one-line narrative on morning brief or daily digest.
- Ask: natural-language Q&A over tasks, wiki, and activity (workspace feature flag).
What AI does not do
- Auto-assign tasks without your confirmation (suggestions may appear; you choose).
- Post messages or change statuses without an explicit click.
- Replace your Wiki for legal, HR, or security policy, human review required.
- Browse the public internet inside your workspace.
Ask (Q&A)
When enabled, Ask answers questions like "What blocked tasks does Engineering have?" using your org's data. It requires backend search infrastructure (embeddings) and an org feature flag. If Ask is missing, your Director may not have enabled it. Use the command palette or dedicated entry point where shown.
Suggest assignee
Some workspaces show assignee suggestions based on workload and team rules, not an LLM guessing randomly. You always pick the final assignee.
Usage limits and privacy
AI calls use a per-user daily token budget. Heavy use in one day may hit the cap; try again tomorrow or ask your Director about plan limits. Usage is logged for admins. Prompts may be sent to the model provider under your org's data processing terms, see Privacy Policy for details.
Getting better results
- Write specific task titles before asking for a description draft.
- For wiki drafts, state audience and goal ("onboarding checklist for engineers").
- Edit every draft, treat output as a first pass.
- For at-risk explanations, open the task itself if the summary seems stale.
If AI is unavailable
You may see an error when quota is exceeded, the feature flag is off, or the service is temporarily down. Core product (tasks, PTO, messages) still works, AI is additive.
Questions about pricing or plans? See the FAQ.