AI features

Every AI surface in Taskachu — what it does, where it lives, what it costs against your monthly quota. The whole point is that AI is woven through the build loop, not bolted on as a chatbot sidebar.

Overview

All AI features share three properties:

• Workspace-scoped — the agent only ever sees the workspace you’re in. There’s no cross-workspace memory or leakage.
• Quota-gated — every call counts against your monthly AI quota (100 calls / month Free, much higher on Pro). The counter resets on your billing cycle.
• Async-first for heavy calls — generation features that take more than a few seconds run on a BullMQ queue. You get a job id and the result streams back via Socket.IO. The UI never blocks.

Card decomposition

Turn a single high-level card into a list of subtasks. Open a card, click the Ask AI drawer, and tap Generate subtasks. The agent reads the card’s title + description, optionally asks one clarifying question, and writes the subtasks back into the card’s checklist.

• Where — every card’s AI drawer (right side).
• Quota — 1 call.
• Async — yes (BullMQ job, ~5-15s).

Board prioritisation

Ranks every card on a board by impact + effort, returns the top picks with rationale, and (optionally) applies the suggested priorities to the cards themselves. The ranked list opens in the Prioritization side panel.

• Where — board toolbar → Prioritize.
• Quota — 1 call per board.
• Async — yes.

Description drafting

Have a card with just a title? Open the AI drawer, click Write a description. The agent has a short back-and-forth with you (1-3 messages) to extract the intent, then writes a Markdown description suitable for handing off.

• Where — card AI drawer.
• Quota — 1 call per accepted draft.

Ask the card (per-card chat)

A scoped chat that knows just this one card’s context — title, description, subtasks, comments, labels. Useful for “summarise where we got to” or “turn the last three comments into an actionable next step.” History is persisted on the card so it’s there next time.

• Where — card AI drawer → free-form input.
• Quota — 1 call per turn.

Ask Taskachu (workspace RAG)

Workspace-scoped retrieval-augmented assistant. Indexes every document, card title, and description; answers questions like “what did we decide about pricing last quarter?” with citations back to source cards/docs.

• Where — sidebar → Ask.
• Quota — 1 call per query.
• Indexing — automatic on doc create/update; manual Sync to AI button on cards.

Daily standup

Per-board standup with a structured check-in (yesterday / today / blockers) and an AI-generated team plan. Tracks check-ins from every member, then synthesises a board-wide focus for the day.

• Where — board toolbar → Standup.
• Quota — 1 call to generate the plan, 0 for check-ins.
• Bonus — auto-posts the plan to Slack if you’ve connected a Slack workspace.

Weekly review

End-of-week retrospective. Reads the week’s standup check-ins, the cards moved to Done, and the cards that stalled; writes a Markdown retro with stats (planned / completed / completion rate / blocker count) and recommendations for next week.

• Where — board toolbar → Weekly review.
• Quota — 1 call per generated review.
• Async — yes.

Brainstorm spaces

Free-form chat surface that’s good at “help me think about X” conversations. When you land on something worth keeping, the brainstorm can be synthesised into a structured board (columns + cards) or a document, with one click.

• Where — workspace → Brainstorm.
• Quota — 1 call per message + 1 for synthesis.

Agent export

Generates a structured handoff packet for an external coding agent (Claude Code, Cursor, Codex). Outputs three files — AGENTS.md, MEMORY.md, COPILOT.md — that you drop into a repo so the agent arrives briefed on the project goal, decisions, and relevant cards.

• Where — board toolbar → Agent export, or per-card from the AI drawer.
• Quota — 1-3 calls depending on which files are generated.
• Note — for live workspace access during coding, use the MCP integration instead.

AI quota & limits

Every AI call across every surface draws from the same per-workspace monthly counter:

When you hit the cap, every AI feature returns a clear modal with a one-click upgrade link. Quota is workspace-shared — a 5-person Pro workspace doesn’t get 5× the cap.

What lives outside Taskachu

For deeper coding workflows, drive Taskachu from your AI tool directly over MCP — the agent reads / writes cards and docs as a first-class citizen. See the MCP integration guide.

Looking for the actual product? Try it free: app.taskachu.com.