The more time I spend with LLMs, the more use cases I'm discovering. I find that people who are not quite there yet with AI are often not limited by their techniques, tooling, or skills, but by their imagination.
The professional and organizational side of AI is my career and where I spend most of my energy, but much of that work isn't free to share here; here are 32 random mostly personal things I have done over the past ~3 years that I either currently use, or was able to remember or find artifacts for. I hope at least one item on this list helps inspire you to dig in further.
Some of these will get dedicated posts in the future. Note that none of these are individual Claude/ChatGPT conversations - if they are on this list, they turned into agents, repeatable projects, skills, or workflows. Also, you are safe to assume if there is something involving personal data, I'm using local or privacy respecting models.
1. Restored an old Trek bike
Found a 7100 on the side of the road that needed some love. Stripped it down, painted the frame, sourced parts, diagnosed issues, followed repair sequences. Was amazing for just finding the names of parts to be able to ask youtube the right questions. Most enjoyable bike I ride now.
2. A personal health coach built on my actual lab data
operates from my Function Health (referral link) results plus years of past labs data, combined with my specific health challenges and goals. Nutrition, exercise, lifestyle. WARNING Always use extreme caution with taking health advice from AI! Caveat aside, last year during a sick visit a doctor wrongly diagnosed a single massively elevated biomarker as a chronic condition and prescribed me some nonsense lifetime medication for it. I used AI for a second opinion and it flagged it as an acute flair up that perfectly matched my symptoms and experience. I held off on filling the prescription and was back to a completely optimal baseline 6 weeks later and multiple follow-up labs showed that Dr. might need to add some AI to his tool set!
3. A council of reviewers
For things that matter (like this blog), I do not use AI to write. Writing is a form of thinking, and I'm not interested in outsourcing that to AI. But I do use AI in the writing process. When I'm done with a draft for any proposal or communication, I (currently) have 11 sub-agents that read and provide feedback from different perspectives - a contrarian-clarifier for precision, an organizational-anthropologist who asks "will people actually adopt this?", or a governance-security-cynic who assumes everything will go wrong.
4. A shared travel planner my wife and I used in London
Made our travel experience the most enjoyable yet. Fed it past likes and dislikes, interests, locations, etc. and multiple times a day we would pop it open to ask what we should do next or how to adapt a plan. Will never travel without one of these again.
5. An automatic physical inbox processor
Scan papers with an old ScanSnap -> sent to a local inbox -> OCR'd by a local LLM, renamed, organized, and anything needing followup lands in my calendar or Things automatically.
6. Deep research for two different car purchases
The most equipped we have been, knew the exactly models and trims we wanted before we started our searches. Acted as a negotiation and process partner along the way allowing us to have no buyer's regret and identify better extended manufacturer's warranty options outside of the dealers.
7. A reading assistant with my active reading list
Asks Socratic questions chapter by chapter as I work through books. Greatly helping build out my PKM. AI as Co-Intelligence!
8. Daily email, Slack, and Google Calendar assistance
The boring stuff. Triage, drafting, scheduling. Boring is undersold though, I hated doing it before, and it wasted time.
9. Helped my wife start a business in under a month
Edenland Farm - LLC formation, EIN, banking, insurance, permits, privacy policies, terms & conditions, copy critique, inventory management, website accepting payments, wholesale partnerships, automated shipping label printing. Zero to operating business in literally under a month, with much better organization than any past ventures we've done.
10. A framework for strategic initiatives
A complete reusable framework built from time-tested mental models and patterns for building and aligning entirely new systems of work. Covers the full cycle from strategic vision to execution planning, measurement, and even language and communication guidelines for staying consistent and sticky. It defines what to build at each stage, what artifacts to produce, and how the components stay aligned as the initiative scales.
11. Designed a self-hosted newsletter reader
Interacts with a dedicated IMAP mailbox using infinite scroll, marks as read as I skim past. Replaced inbox-zero pressure for non-essential mail with social-media-style casual consumption. It's technically still doom-scrolling, but at least it's only newsletters that (I think) I signed up for.
12. A complete Personal Master Context document
Probably the first big breakdown for my AI usage. Iteratively built personal and work system prompts covering personality profiles (DISC, StrengthsFinder, MBTI, Enneagram, SIMA), active ventures, values, goals, blind spots, preferred tools, writing styles, etc. Massive quality improvements when you onboard your AI.
13. Built a proper inventory of my entire homelab
All hardware details, every ip address, IoT device, VLAN, switch, AP, deployed apps and versions, firewall rules, the works. A visual network map. Surfaced issues I'd missed for years. One of simple fixes it identified literally doubled our internet speeds.
14. Setting up a Living Trust
Legal document prep, research, decision frameworks. Obviously DYOR and finalize with a proper lawyer!
15. A GTD coach operating directly on my Things 3
Tuned on GTD principles, my specific areas of life and working style, operating directly inside my actual task manager (Things 3) with my actual tasks. Weekly reviews actually get done now.
16. An assistant that knows our farm's soil
Built from deep research using exact latitude/longitude, NRCS soil survey data, and local climate records to provide an always on expert for our gardening, lawn care, and landscaping.
17. Published my first iOS app to the App Store
18. Meeting coach that reviews my transcripts
Post-meeting evaluator. This one is kind of mean, and I don't always want to use it. (e.g. "Heavy use of 'kind of,' 'um,' 'you know' — 40+ instances.")
19. Data dashboards
Creative, interactive monthly survey visualizations. Way faster than building out manually in Google sheets, and actually enjoyable to review.
20. A content queue system
Built and harnessed a collection of agent skills managing the onslaught of incoming YouTube, blogs, Substacks, and PDFs. Extracts text and transcripts, summarizes, then analyzes against my role and responsibilities to grade what I should read deeply vs. skim vs. archive.
21. A Home Assistant assistant
I've been using home assistant for well over a decade, but it's always half broken because I have a job and a life and don't want to learn all of the YAML it uses. The system is 100x better right now, has multiple local AI assistants with custom voices and humorous personalities my kids love. What Alexa/Siri should have been.
22. Migrated my 10+ year personal yearly review and goal-setting process to an interactive assistant
Every year in December I spend days reviewing my year and planning out the next year. I iteratively developed system around my goals and values. It is great but time consuming. This year I turned it into an interactive system, fed it evidence-based goal-setting research, and had a completely different experience. 10/10 would recommend.
23. Deep research for my optimal local LLM setup
Hardware specs, model comparisons, benchmarks for my specific use cases. Learning an absolute ton, but also up and running significantly faster than if I was stuck reading documentation for each technology.
24. Countless servers debugged, root causes finally found
Interactively working through massive log files, tracing issues across deployments. Problems we have solved in the past by reboots and crossing fingers.
25. An interactive strategy mentor
React app with Claude API integration that runs live Socratic coaching dialogues based on Seth Godin's strategy framework.
26. Workflow Apprentice
An AI onboarding assistant that learns workflows from conversations and builds assistants. An AI that builds AIs.
27. A complete rapid web app deployment pipeline
Repeatable workflow from idea to production: GitHub → CI/CD → Docker → GHCR → Helm → K8s → Cloudflare, with staging environments and semantic versioning.
28. A complete Obsidian vault assistant
Tuned to my entire workflow: tasks, time-boxing, PARA methodology, publishing lifecycles. Probably the one I use most frequently. Hint: It's just claude code and markdown!
29. Family resilience planning
Assistant for overall preparedness across food, water, energy, heat, specific to our family's needs and location.
30. Vehicle maintenance and fuel tracking
An agent that logs gas fill-ups from natural language, tracks odometer readings across three vehicles, and cross-references service schedules so I get reminders when service is due.
31. Homeschooling and curriculum assistants
Several, built for my teacher wife who homeschools our younger kids. Lesson planning, curriculum mapping, progress tracking. She says she has had her best year yet.
32. An AI-assisted bee school journey
Research, note-taking, and planning while learning beekeeping with our local Beekeepers Association.