I completed a tiny experiment on using my Day cluster notes in Obsidian: Plan, Log, Journal, Review, Actions, and Links. These are the results from my 5 workday trials (Jan 12–16, 2026) with a brief summary, what went well (Plus), what didn't (Minus), and what can I do (Next). I described the pact for this experiment in my initial post that began this series:
Highlights
Before this experiment, I spent over 60 minutes of manual work to do a diabetes review every morning for the previous day, and now it's an automated workflow that takes under 10 seconds!
This is done automatically for me when I create my Day cluster notes in Obsidian for each day: Plan, Log, Health, Journal, Links, Actions, and Review. This includes a Python review service that generates a Claude AI analysis of Plan vs Log for my final Review of the previous day, as well as generating my Diabetes Review via a Python health service.
It uses a Dexcom API that analyzes and produces a report with summary, statistics, tables, and graphs from my glucose readings captured by my continuous glucose monitor (CGM) every 5 minutes (288/day).
I used my Day cluster notes each day as part of my workflows, and refined the scripts, services, and templates based on my actual usage and experiences. These iterations provided significant progress and momentum that helped to incrementally improve my daily practices throughout the week of this experiment.
I now have a working production system based on my Knowledge System Architecture (KSA) that leverages a Neo4j knowledge graph, Python services, Claude AI (LLM), Dexcom API, and Obsidian notes.
➕ Plus
- Automatically creates a Day cluster of linked Obsidian notes with built-in navigation wayfinders for Plan, Log, Health, Journal, Links, Actions, and Review.
- Generates Day support notes for View, Base, Board, Canvas, and Drawing as supporting visual tools.
- Generates Obsidian note for AI Analysis from Plan vs Log, as well as Diabetes Review via Dexcom API from CGM glucose data.
- Perform automatic Git commit for Obsidian vault with all changes since yesterday.
- Automatically open a Day cluster workspace using QuickAdd macro commands.
- Integrating and phasing out previous Templater scripts into enhanced Python service templates.
- Created initial templates for Week and Year clusters.
- I have published this entire experiment while learning in public on my Ghost website and the Circle communities (LYT and Ness Labs).
➖ Minus
- It took a while working out the details for the production Dexcom API access, but I was still able to add initial diabetes analysis with basic summary and statistics.
- Since I spent more time on the Dexcom API, I wasn't able to integrate the Glooko website for my Omnipod 5 insulin pump and meal carb data (yet).
- Without automated workflows, it took too much time to get anything done, and often it didn't happen or other stuff kept getting deferred.
- There's only so much time each day, and it's important to focus my intentions on the primary effort and key actions in time block sessions.
- I didn't spend as much time in Review/Preview at the end of each day, but my automation now simplifies that workflow quite a bit with the automatic Plus-Minus-Next analysis of Plan vs Log as well as Insights.
➡️ Next
- Integrate insulin pump and meal carb data from Glooko website to combine with my Dexcom glucose as well as HealthKit for food, activity, sleep, and mindfulness data.
- Integrate previous R diabetes management work into my Python services and Obsidian notes with enhanced Diabetes Review summary, tables, statistics, and graphs for Day, Week, Month, Quarter, and Year.
- Integrate previous Swift diabetes management work with HealthKit into my Python Services and Obsidian notes, along with additional meal plan, recipes, calendar, and photo services for my meals.
- Extend Day cluster work into a similar approach for Week, Month, Quarter, and Year clusters.
- Shift my focus from Day to Week cluster workflow practices.
Summary
During this transition after the experiment, I have the opportunity to consider several alternative paths forward using a steering sheet: Persist, Pause, or Pivot.
I will Persist with this experiment as I continue with using my Day Cluster notes. This has proved to be extremely valuable in so many ways, even though it has been something of an effort. Both the external and internet signals have all be mostly positive and encourage me to continue and extend this effort.
My initial focus will complete the development of my enhanced templates, extend my work initially into Week and Year clusters, and significantly enhance my Diabetes Review effort.
I will not publish my trials each day, but I will post my progress on my website and communities as appropriate as I begin to post much more publicly about my work.
