More than 10 years ago I went to a lecture by Bayram Annakov, founder of App in the Air and Onsa.Ai, about Mihaly Csikszentmihalyi’s book Flow – the state of full immersion where effort and time fall away. The Inflow app he was building at the time let you do short mood and energy check-ins through the day, and the lecture was really about what those check-ins surface over time: the things you assume give you energy often quietly drain it, and the ones you’d never credit are what carry you. I found it interesting enough to start tracking my own mood and energy and see what I’d notice.

I tried a few setups over the years. First the Inflow app itself – simple and fast, but too thin. Then a spreadsheet, which let me track in real detail but cost too much time. Then Tick-tick, my task manager at the time. Each had its trade-offs, and the habit evolved alongside whatever I was using.

In 2020 I learned about the second brain idea – an external system that holds your notes and references so your mind doesn’t have to – and moved to Roam Research. Roam popularised the graph-based approach to notes, where every note is a node and links between notes are first-class, so the knowledge base grows as a web rather than a folder tree. After a couple of years, frustrated with Roam’s slowing development but not quite satisfied with alternatives, I finally stumbled onto Tana – another graph-based knowledge base, but wrapped in a much better product. The check-in habit didn’t suddenly become effortless; I’d been doing it for years already. But Tana’s mobile app made it noticeably easier to keep up, which mattered more than I expected.

The habit, manual edition

My daily check-in node in Tana used to look like this:

• I track when I started and ended the day.

• I rate mood and energy on a 10-point scale.

• I write a short diary entry, where I reflect on the previous day and outline what to focus on today.

• I tick off a few habits (read, study, code, meditate) as simple checkboxes.

On Sunday evening, I would open the weekly view, scroll through the daily notes, and write a weekly outcome – analysing what gave me energy and what drained it. Then I’d set the priorities for the next week, and the cycle would repeat.

There’s a quarterly goals layer above this too – topic for another article.

The habit worked, and I genuinely liked the quality of reflection it produced. The only friction was getting myself to sit down and actually write the diary, especially on Sunday evenings when the weekly review took the better part of an hour.

Flipping the habit: Voice and AI

Two things turned it from a chore into a flow:

1. Voice recording in the Tana mobile app removed the friction of finding the daily node and editing each field.

2. The Tana MCP server connected to my Claude Code made my entire Tana workspace reachable from from CLI – I can search nodes, read them (including audio transcription), and write structured content back.

The routine now is two actions: a 2-minute morning recording and a 30-minute Sunday session that’s mostly reflection.

Most mornings I record a short audio check-in in Tana – mood, energy, what I want to focus on, a reflection on yesterday – usually as I leave the house and start my commute, talking while walking. That’s it for weekdays.

On Sunday I open Claude Code and run the skills listed below. The first skill transcribes the week of recordings and populates the daily pages with diary entries, mood, and energy values. The second reads those days and writes the weekly outcome – what gave me energy, what brought me down, and a draft of priorities for the next week. Then I review what it wrote, edit, and reflect on it.

The review step is the important one. The AI doesn’t replace the reflection – it just removes the part where I’m copying numbers into fields and re-reading my own bullet points to remember what I felt on Tuesday. And because Claude Code keeps memory across sessions in this project, it often spots patterns across several weeks, or shows how the week is feeding into my quarterly goals.

Surprisingly, the whole initial setup took about an hour. It came together naturally, and I started reaping the benefits immediately.

The skills

I built two Claude Code skills to streamline the routine. It took a while of iterating on the prompts – running them by hand, tweaking what came back, running them again – before I trusted either enough to save it as a skill. The first populates daily diary pages from voice recordings; the second generates the weekly summary the same way I used to write it myself.

Both skills live on GitHub if you want to look at them or adapt them to your own setup: https://github.com/reDim89/redim-skills.git

Why I wrote this

What strikes me is how much the good habits used to fail mostly on friction: the small daily cost of doing the thing, and how much easier it became.

James Pennebaker’s four decades of expressive writing research show that consistent reflective writing meaningfully improves mood, stress, and self-knowledge (overview paper, open PDF). The friction is what kept many people from sticking with it.

Voice input, mobile capture, and an LLM that can read your notes and write back into them change that math. The habits worth keeping just got cheaper to keep – which feels like an opening worth taking, toward a life that’s a bit more reflected on.

If you have your own diary setup, I’d be curious how you’ve shaped it – tools, cadence, what’s worked. Leave a comment.