Productive's AI features are designed to meet you where you are. You might start by asking the AI Assistant a quick question, then notice you're typing the same kind of request every week. Eventually, you realize a whole workflow could just run on its own.
This article walks you through that progression: from your first conversation with the Assistant, to saving instructions as Skills, to setting up Agents that work without you.
๐ All AI features require Productive AI to be enabled by an admin in Settings > Productive AI.
Some features, including Skills, Agents, and Connectors, are available on Professional and Ultimate subscription plans only.
Start With the AI Assistant
The AI Assistant is the fastest way to get something done. Open it from the icon in the upper-left navigation, type your request in plain language, and it responds.
You can ask it to:
Summarize an open task or project
Draft a client update based on recent activity
Create or update tasks, assign them, set due dates
Search across your workspace using plain language
Generate reports by describing what data you want to see
Summarize your worklog by project, hours, or time period
The Assistant can do anything in Productive that you can: creating, updating, and managing work across the platform. And it already knows what you're looking at, so if you have a project open, you don't need to re-explain the context. You just ask.
Example: You're staffing a new project and type: "Who on the design team has capacity next week, and are there any tasks on the Acme project that still need assigning?"
The Assistant checks the Resource Planner and the project at the same time and gives you an answer: no tab-switching, no manual cross-referencing.
๐ If the first response isn't quite right, follow up or ask it to try again. The Assistant is built for iteration.
๐ For tips on getting better results, see Productive AI: Prompting and Instructions.
Save Repeated Instructions as Skills
Once you've been using the Assistant for a while, you'll probably notice a pattern: you're writing the same instructions over and over.
"Always use formal language."
"Check these three things before approving."
"Format the output as a bullet list."
That's what Skills are for. A Skill is a saved instruction set you write once and reuse whenever you need it. Instead of repeating yourself, you reference the Skill in a conversation and the Assistant follows those instructions automatically.
Example: You create a Skill called Weekly Project Health Check with instructions: "Look at the linked project. List all tasks due this week with no assignee. Check if any services on the budget are more than 10% over estimate. List any team members booked on this project with over 80% utilization this week. Summarize in bullet points."
Every Monday, you open the project, reference the Skill, and get a structured briefing covering tasks, budget, and resourcing in one pass. No rewriting the prompt, no jumping between sections.
๐ Skills are private by default. You can share them with specific users, teams, or Agents.
๐ What Are Skills?
Delegate Recurring Work to Agents
When a task isn't just repeated but truly routine, it's a good candidate for an Agent. If it follows the same logic every time and doesn't need your input to start, an Agent can handle it.
Agents are autonomous AI workers that run inside Productive on a schedule or trigger. Unlike the Assistant, which responds when you ask it something, an Agent acts on its own. You configure it once, assign it Skills, permissions, and connectors, and it handles the work.
Agents don't just report. They can create tasks, post updates, and take action across Productive on your behalf. You could use an Agent to:
Check for overdue tasks every morning and send a summary
Flag active projects with unbooked capacity before the gap becomes a problem
Flag deals that haven't had activity in weeks
Send overdue invoice notifications to a Slack billing channel on a set schedule
Example: You assign your Weekly Project Health Check Skill to an Agent and set it to run every Friday at 4pm. It checks all active projects, flags any with overdue tasks or budgets over estimate, and posts a summary to your team's Slack channel. You walk into Monday already knowing where the problems are.
๐ Agents are available on the Ultimate plan. Only Admins and Managers can create and configure them by default.
๐ AI Agents: Overview
Connect to External Tools With Connectors
You'll reach for Connectors when the data you need lives outside Productive.
Connectors link the Assistant to outside services: things like your calendar, a communication tool, or an external project tracker. Once connected, the Assistant can read data from those tools or take actions through them, directly in your conversation.
Example: You connect your calendar and type: "Track my time from calendar events for today." The Assistant finds your meetings, matches them to the right services based on your history, and logs the time entries. One request covers the whole day.
Connectors can also be assigned to Agents, so an Agent can pull in or act on external data as part of its workflow.
๐ An admin needs to enable a Connector for your organization first. Each user then connects their own account in Settings > Connectors.
๐ Connectors
How It Fits Together
These features work alongside each other. A natural progression looks like this:
Use the Assistant for day-to-day tasks: writing, searching, updating records
Build Skills when you find yourself writing the same instructions repeatedly
Set up an Agent when a recurring workflow no longer needs you to kick it off
Add Connectors when the work spans tools outside Productive
You don't need to use all of them at once. Start where it's useful and build from there.
