Productivity Cycles and AI Agents

Tana’s out-of-beta release is paired with a practical system for turning focus into measurable cycles—and then feeding those records to an AI “productivity coach” that gives feedback based on what actually happened. The core idea is to structure work into repeatable time blocks (30 minutes), group three blocks into a 90-minute “brain cycle,” and log energy, motivation (“Elan”), and whether the goal was completed. That data then becomes the input for live searches and an AI agent that reviews recent cycles and suggests concrete improvements.

The build starts with a smallest unit called a “Nano cycle,” implemented as a super tag in Tana. Each Nano cycle note automatically includes fields for date, energy (configured as a dropdown using battery emojis), Elan (dropdown using fire emojis), and goal completion (checkmark vs X). To avoid re-creating the same structure every time, the fields are added to the super tag template, and default values can be set so a new cycle starts with consistent prompts.

Next comes the workflow inside each Nano cycle. A planning template prompts the user to define the goal, decide how to start, and anticipate likely distractions—plus a counter-strategy for each. After the work block, a review template asks what was accomplished, what distracted attention, and what to do better next time. Tana commands automate the mechanics: buttons appear to insert the plan and work sections, clone the review section, and auto-fill the date/time for the cycle using a relative date string (e.g., “now + 30m”). Conditions control when buttons show up—plan/start only when the date field is empty, and review only when it’s filled—so the cycle behaves like a guided checklist.

To reduce busy work during review, the system uses a live search that automatically aggregates completed tasks. When items are checked off anywhere in the plan/work sections, a “grandparents/descendants with refs” search for done items surfaces them in the accomplishments area. This means the user doesn’t have to manually copy results into the review.

The 90-minute layer (“micro cycle”) groups three Nano cycles together with short breaks. Its plan step adds an extra element: consequences of not doing the task, framed as more motivating than purely positive visualization. A live search then pulls together what was completed across the three Nano cycles, including energy/Elan changes and distraction notes.

Finally, an AI agent is wired into the workspace. A “cycle review session” tag collects micro cycles from the last day via a live search, then passes that context into a Tana agent configured with a system prompt tuned for an expert productivity coach. The agent can answer questions like what to improve, and it responds using the logged patterns—such as noticing energy and Elan dropping in a specific micro cycle and recommending more robust distraction management. The result is a feedback loop: structured logging → automated aggregation → AI coaching grounded in the user’s own recent work history.