TskVlo’s built-in AI assistant drafts tasks from a prompt, answers questions about your board, and summarizes long comment threads. It runs on your own AI provider — you connect any OpenAI-compatible API, from OpenAI to a model running on your own hardware.
Step 1: Connect an AI Provider (Admin)
A workspace admin needs to configure the AI endpoint once. Until then, AI features stay hidden for everyone.
- Open Workspace settings → AI / LLM configuration
- Fill in the three fields:
- API base URL — e.g.
https://api.openai.com/v1 - API key — your provider’s API key
- Model — e.g.
gpt-4o-mini
- API base URL — e.g.
- Click Test connection — TskVlo sends a minimal test request and reports success or the exact error
- Click Save LLM settings
Your API key is stored securely and never shown again in plaintext.
Provider Examples
| Provider | Base URL | Notes |
|---|---|---|
| OpenAI | https://api.openai.com/v1 | e.g. model gpt-4o-mini |
| Mistral | https://api.mistral.ai/v1 | EU-based, popular for GDPR-sensitive teams |
| Groq | https://api.groq.com/openai/v1 | Fast inference for open models |
| LM Studio / Ollama / vLLM | your own server URL | Open models on your own infrastructure |
Self-managed endpoints: TskVlo’s servers call your endpoint directly, so an LM Studio or Ollama instance must be reachable from the internet (for example behind a reverse proxy or tunnel) —
localhoston your laptop won’t work.
Image support: attaching images in chat requires a vision-capable model (e.g. GPT-4o, Pixtral). Text features work with any chat model.
Step 2: Chat with the Assistant
Click AI Assistant in the sidebar. Each chat is private to you, and you can keep several — they’re auto-titled from your first message and can be renamed or deleted anytime.
Scope the chat to a board to give the assistant context: it then sees the board’s groups and tasks, including field values like status, dates, and assignees. Ask things like:
What's still open in this sprint, and who has the most tasks assigned?We're planning the Q3 website relaunch. Draft the tasks: content audit,
new design, CMS migration, SEO check, and launch comms.Step 3: Review and Create Draft Tasks
When you ask for new tasks, the assistant replies with draft tasks — titles, descriptions, and suggested groups (it can also propose new groups). Nothing is created yet:
- Review the drafts in the chat
- Select the target board and group
- Click Create tasks
The tasks appear on the board instantly. The assistant never writes to your board without this confirmation step.
You can also attach an image — a whiteboard photo from sprint planning, a wireframe — and ask the assistant to turn it into tasks.
Summarize a Comment Thread
On any task with comments, open the Comments tab and click Summarize with AI. You get a structured digest:
- Summary — a short overview of the discussion
- Decisions — what was agreed
- Open questions — what’s still unresolved
- Action items — agreed next steps
The summary is written in the same language as the thread. Click Post as comment to anchor it in the discussion, Regenerate for a fresh take, or Dismiss.
Troubleshooting
- “LLM not configured” — an admin needs to complete Step 1 under Workspace settings → AI / LLM configuration.
- Connection test fails — check that the base URL ends at the API root (e.g.
.../v1, without/chat/completions) and that the key is valid for that endpoint. - Drafts but no “Create tasks” button active — select a board and group first.