Daily notes
One idea a day. Plain English.
Short notes on AI systems, automation, and building with Claude Code. Each one is a single idea you can try the same day.
September 2026
- 17Set aside time to try to break your key automations
- 16Ask if a tool removes work or adds work
- 15How to know you're ready to help someone else
- 14Count how many things route through you
- 13Read your follow-up messages the way a stranger would
- 12Keep an idea log so you never run dry
- 11Bring a real example, not a vague idea, to a session
- 10Give every workflow step a plain label
- 09Buy back an hour before you buy back a role
- 08Revisit the rollout and adjust after the first stretch
- 07Teach your AI tool your own editing habits
- 06How the local versus cloud mix changes as you grow
- 05Make your alerts say what broke, not just that something did
- 04Simple wins over powerful for a small team
- 03Ask why a tool works, not just how to use it
- 02Delegate the outcome, not just the task
- 01Build one onboarding path, then let exceptions be exceptions
August 2026
- 31Turn one long answer into a short series
- 30Pick a first project you'll actually finish
- 29Keep steps in the order a person would do them
- 28The weekly check that keeps automation honest
- 27Rotate who champions the tool each quarter
- 26Some videos are faster to cut by hand
- 25Privacy applies to your photos and voice too
- 24Slow automations down to avoid silent overload
- 23The whiteboard sometimes beats the project tool
- 22Build a little every day instead of one big push
- 21Sort your list into urgent and important
- 20A follow-up reminder needs a reason, not just a nudge
- 19Use a simple test before you post a clip
- 18Review what changed before you trust an update
- 17Judgment calls stay with a person not a workflow
- 16Slow down before automating an unclear process
- 15Ask each person what they'd want off their plate
- 14Build one editing workflow you reuse every time
- 13Local AI is great at boring internal work
- 12Ask what happens if a step fails halfway through
- 11Naming things the same way across tools saves headaches
- 10Finish the ugly version before you make it pretty
- 09Set hours for approvals instead of instant replies
- 08Send the update clients want, not the one you want to write
- 07Match clip length to where it is going
- 06Ask why it picked this approach over another
- 05Not every task deserves automation
- 04The signal a task is ready to automate
- 03Set a rule for a human check before anything ships
- 02AI subtitles open your video to new viewers
- 01The pause before you hit enter
July 2026
- 31Keep a short incident log when something breaks
- 30Don't buy the tool built for a bigger business
- 29Pick a project you'll actually use again
- 28Split messages that need a reply from the rest
- 27Watch for the onboarding step where people drop off
- 26Build a small bank of background footage
- 25Write the plan first, get buy-in, then build
- 24Write the workflow down before you build it
- 23Write the steps down before you automate them
- 22Track time saved, not tasks automated
- 21Sort your footage before AI ever touches it
- 20Local AI still needs occasional upkeep
- 19Do not stack too many jobs on one automation
- 18A password manager is the one tool everyone needs
- 17Turn a repetitive task into your first project
- 16Ask if the business runs without you a while
- 15A welcome packet should answer questions early
- 14Turn comments into your next post
- 13Keep tools separate so one failure stays small
- 12A checklist is a workflow waiting to happen
- 11The narrow automation that buys back an afternoon
- 10Make it easy to opt out of a bad fit
- 09Let AI catch the continuity slips you missed
- 08Where an idea lives before it is ready
- 07Attach a review date to every automation you build
- 06Review your stack after your busiest season
- 05Chase your own what if questions
- 04Trade the status meeting for a short update
- 03Keep a short list of messages only you get to send
- 02Split your recording days from your editing days
- 01Ask what happens with missing or wrong input
June 2026
- 30Pick one source of truth for your trigger
- 29Contractors handle judgment, tools handle repeats
- 28Test a new tool on the lowest stakes work first
- 27A short checklist for reviewing any AI cut
- 26Trust in a local model builds slowly
- 25Watch for outside tools changing under your feet
- 24Build vs buy is a real question, answer it on purpose
- 23Explaining your project out loud finds the gaps
- 22Hand off context, not just instructions
- 21Approval steps should take one click, not a meeting
- 20Keep clips short enough to finish
- 19Build in short sessions, not one marathon
- 18Batch small events instead of reacting to each one
- 17Automate booking before anything else
- 16Build a short answer sheet instead of a memory test
- 15Your job shifts from cutting to directing
- 14A short weekly check on where things went
- 13Build automations in small pieces you can test alone
- 12One source of truth beats five half updated ones
- 11Use a tight deadline to force yourself to finish
- 10Trace repeat interruptions back to their root
- 09Payment setup belongs in onboarding, not after the fact
- 08Turn a work session into content, no extra filming
- 07Let a real error guide the next step
- 06The difference between an action and a condition
- 05The task list test for what to automate first
- 04Skip the one size fits all mandate
- 03Turn one long video into many short clips with AI
- 02Some businesses need to lean local by default
- 01Test in a sandbox before touching real customer data
May 2026
- 31Too many notifications means too many tools
- 30Rebuild the same small project with a new tool
- 29Test what happens if you go quiet for a day
- 28Write your onboarding steps down before you automate them
- 27Plan your caption before you record
- 26Test with a spreadsheet before a live database
- 25Start from a template instead of a blank page
- 24A spreadsheet plus one automation beats a hire
- 23Match the tool to the person, not the other way
- 22Keeping one style across every AI-assisted edit
- 21A quiet first pass before the real polish
- 20Tell your team exactly what the automation covers
- 19Pick tools with a plain export option
- 18When one-on-one help is worth asking for
- 17Give each project one point of contact
- 16A missed call deserves a same-day follow-up
- 15Sort finished clips from leftovers as you go
- 14Don't build what you can't explain to your team
- 13Every step needs a clear input and output
- 12Do not hire someone to babysit the tools
- 11Give people a grace period to get comfortable
- 10Where AI editing runs out of feel
- 09How I explain local AI to someone new to it
- 08Write down the edge cases your automation cannot handle
- 07Watch for tools that quietly renew
- 06What a group session teaches that solo work can't
- 05Close the day instead of leaving things open
- 04Ask what went wrong right after onboarding, not months later
- 03One story, several formats
- 02Ask for a few approaches before picking one
- 01Chain small workflows instead of one giant one
April 2026
- 30Hire for relationships, automate the repetition
- 29Always explain why before you hand out a tool
- 28Letting AI sync music to your cuts
- 27Local AI still needs a settings check
- 26Keep a human eye on anything touching money
- 25A shared doc beats a new app for small teams
- 24Stop waiting until you know enough to start
- 23Block time to think like you block a meeting
- 22Every automated reply needs a clear way to reach a person
- 21Keep a running list of clip-worthy moments
- 20Keep a graveyard folder for ideas that failed
- 19Don't automate what you don't understand yet
- 18A reminder system replaces a lot of follow-up
- 17Reward people for finding what AI got wrong
- 16Matching b-roll to your script, the AI way
- 15Keep two AI tools open, not just one
- 14Watch for inputs quietly changing shape over time
- 13Don't let one person own a tool alone
- 12Ship something small every week
- 11Announce what you are dropping this month
- 10A shared client hub beats a hundred check-in emails
- 09Try more than one hook on the same idea
- 08Give feedback in plain terms, not code terms
- 07Put approval before money or public words leave
- 06Sort receipts and invoices without a person
- 05Split drafting work from deciding work
- 04Ask AI to find the hook buried in your footage
- 03The promise I make about client information
- 02Give yourself a way to pause an automation fast
- 01Ask who fixes it when it breaks
March 2026
- 31Keep a short log of your mistakes and fixes
- 30Put decision rights for your team in writing
- 29Automate the easy step first, not the sensitive one
- 28Review your footage while it is still fresh
- 27Build tools around what you already do by hand
- 26Know what done looks like before you build
- 25Watch where the week actually goes
- 24Spread a win from one desk to the team
- 23AI reframing helps you cut once, post everywhere
- 22Set one simple rule for local versus cloud
- 21Do the task by hand before you automate it
- 20The hidden cost of switching tools later
- 19Teaching what you just learned locks it in
- 18Ask a second person before big calls
- 17The close-out message matters as much as the welcome one
- 16Turn today's recording into today's written post
- 15Treat every finished claim as a question
- 14Filters keep a trigger from firing on noise
- 13Automate what you would delegate anyway
- 12Keep a shared list of prompts that worked
- 11Pacing is still a human call, not a machine one
- 10Where local AI hits its limits
- 09Back up your automation settings, not just your data
- 08Choose a tool by the job, not the features
- 07Signs your first project is already too big
- 06Write down steps so decisions repeat without you
- 05Test your onboarding as if you were the new client
- 04Grab extra shots while the camera is already on
- 03Ask what it needs from you before it starts
- 02Give a trigger a clear owner
- 01Shadow the process before you automate it
February 2026
- 28Write real instructions, don't assume training stuck
- 27Batch editing saves the most time on repeat formats
- 26Use a local model to clean text before cloud AI
- 25Retire automations that no longer fit the business
- 24Stop buying tools out of habit
- 23Understanding a tool beats memorizing its steps
- 22Rank your calendar by who could cover it
- 21Renewal reminders should start earlier than feels natural
- 20Decide what you will not post
- 19Don't skip the setup step it explains
- 18Test a new workflow on yourself first
- 17Seasonal spikes do not need a new hire
- 16Let the slowest person set the pace
- 15Editing by transcript changes how you cut video
- 14The cost math behind running AI on your own machine
- 13Track how often things go wrong, not just when
- 12Pick tools a new hire can learn fast
- 11Copying a project on purpose is a real way to learn
- 10Say no by default to new asks in the moment
- 09Tag incoming messages by what they actually need
- 08Turn the questions people ask into content
- 07Test the worst case before the normal one
- 06If this then that is the whole idea
- 05Automate reports before anyone builds them by hand
- 04Say the quiet part about job fears out loud
- 03Let AI flag the boring parts, you decide the cut
- 02AI that still works when the internet does not
- 01Keep a manual backup path for every automated step
January 2026
- 31Kill a tool before you add a new one
- 30Keep a running list of everything you've built
- 29Wait a full day before checking delegated work
- 28A follow-up sequence needs an exit door
- 27Grab your best line and make it its own clip
- 26Keep decisions in your own notes, not just chat
- 25Keep each workflow doing one job
- 24Turn your own checklist into a workflow
- 23Roll out one tool at a time
- 22Cleaning up noisy audio is an easy AI job
- 21A private place to think out loud with AI
- 20Alert on the near-misses, not just total failures
- 19Write down what runs your business
- 18Why pairing with someone ahead of you helps
- 17Build one dashboard and check it weekly
- 16Set response-time expectations on day one
- 15Your best old post deserves a second run
- 14Small tools beat big platforms early on
- 13Time-based triggers versus event-based triggers
- 12Automate the calendar back and forth
- 11Give every tool a single clear owner
- 10Auto chapters make long videos easier to watch
- 09Private does not mean always right
- 08Write a plain description anyone could read
- 07The two tool rule for new processes
- 06Build for yourself before you build for anyone else
- 05Give away good work, not just boring parts
- 04Chase incomplete forms before they go cold
- 03Keep a small set of formats you reuse
- 02Let it draft the checklist, you approve it
- 01Approval does not mean review everything forever
December 2025
- 31Data entry is the easiest first automation win
- 30Hold a weekly show and tell for wins
- 29Speeding through boring stretches with AI first
- 28Let local AI handle the rough draft only
- 27Sort automations into nice-to-have and must-not-fail
- 26Match the tool to the size of your team
- 25Error messages are trying to teach you something
- 24Let small mistakes go without stepping in
- 23Decide in advance what needs a human sign-off
- 22Write captions that work with the sound off
- 21Write down exactly how you'll know it worked
- 20One trigger doing a job two ways is one too many
- 19The inbox system that replaces an assistant
- 18Automate the boring parts before the judgment calls
- 17Don't let AI pick your thumbnail alone
- 16Write down what never leaves your machine
- 15Build an alert for when nothing happens at all
- 14Make sure you can get your data back out
- 13Learning or just researching? Know the difference
- 12Batch your decisions instead of reacting
- 11One place for client files beats a scattered inbox
- 10Turn one call into a handful of short clips
- 09Ask what could break this before you trust it
- 08The best first automation touches one app
- 07Start automation inside the business first
- 06Post a short list of what stays off limits
- 05AI color correction gets you close, not perfect
- 04Why local AI sometimes feels slower
- 03Do not let an automation run untouched forever
- 02Build your stack around one hub tool
- 01Break your first project into small pieces
November 2025
- 30Close the week before you open the next
- 29A quiet client is a signal, not good news
- 28Cut the throat-clearing before you post
- 27Save every version so you can go back
- 26A workflow is a recipe with steps in order
- 25One steady automation beats five shaky ones
- 24Give people permission to catch AI's mistakes
- 23AI syncing turns multicam into a quick job
- 22Test a local model before you trust it
- 21Keep a short changelog of what you changed and why
- 20Test a tool for a month before you commit
- 19Describe it in plain English before you build
- 18Name an owner for every repeating task
- 17Automate the update, not the apology
- 16Batch record a stretch of content in one sitting
- 15Build the ugly version first
- 14Automate your busiest day not your biggest dream
- 13Someone has to own the automation
- 12Keep client work off the table at first
- 11A rough AI cut beats a blank timeline
- 10The easiest way to try a local model
- 09Put automation check-ins on your actual calendar
- 08Fewer tools beat more tools
- 07How to ask a good question when you're stuck
- 06Cap how many decisions a meeting makes
- 05Give new clients a simple map of what happens next
- 04Name your footage so future you can find it
- 03Name things the way your business already talks
- 02A workflow needs a name a stranger could read
- 01Test the automation on yourself first
October 2025
- 31Train on real work, not generic demos
- 30AI can find your best lines in raw footage
- 29The computer you already own is probably enough
- 28Decide in advance when a human should step in
- 27Free tools still cost you time
- 26Pick a boring first project on purpose
- 25Write the steps before you hand off work
- 24Reminders work better before the deadline than after
- 23Build a repurposing checklist and use it
- 22Keep a running list of what it can't do yet
- 21Short workflows are easier to fix later
- 20Compare tool cost to hire cost in hours
- 19Write down the win before you scale it
- 18Let AI clear out the ums before you touch a clip
- 17Local models are not as sharp, and that is fine
- 16Feed your automation messy input on purpose
- 15Before buying a tool, ask what it replaces
- 14The tutorial hopping trap
- 13Trade an open door for office hours
- 12Never let a bot answer an angry message
- 11One idea per clip, always
- 10Ask it to explain what it built, in plain words
- 09Map the manual process before you automate it
- 08The follow-up email nobody has time to send
- 07Run a tiny pilot before the full rollout
- 06Cutting silence is the safest AI edit to trust
- 05Draft private, polish in the open
- 04Write down what normal looks like first
- 03The one tool, one owner rule
- 02What to bring to your next help session
- 01Set a threshold for decisions that need you
September 2025
- 30A welcome flow needs a human check before it sends
- 29Keep a swipe file of hooks that worked
- 28Test with your messiest real data first
- 27Put a human approval before anything goes out
- 26Fix the process before you automate it
- 25Let the team pick the first tool to try
- 24AI b-roll picks still need your eye
- 23When a cloud AI tool is the right call
- 22Every automation needs one named owner
- 21Pick tools that talk to each other
- 20Group learning versus one-on-one help
- 19Keep a short decision log
- 18Confirm every submission the moment it comes in
- 17Let your transcript write your captions
- 16Write your done checklist before you start
- 15Automate the task you dread most first
- 14Judgment tasks stay human, repeats do not
- 13Let your most curious person go first
- 12Captions are the fastest win in AI editing
- 11The line I never let cross into cloud tools
- 10Keep a run log for every automation you build
- 09Run a subscription audit twice a year
- 08How to pick your first AI project
- 07Make the weekly review a real appointment
- 06Put your intake in one form, not five emails
- 05Write your hook before you hit record
- 04Give a tool one job, not five
- 03What counts as a step in a workflow
- 02Automate one task before you write a job post
- 01Start with one small task, not the whole job
August 2025
- 31Let AI make the first cut, you make the last one
- 30What running AI locally actually means
- 29Give every automation its own failure alert
- 28When a spreadsheet is still the right tool
- 27Why building beats watching tutorials
- 26Sort your tasks by who else could do them
- 25The first message a new client gets sets the tone
- 24Treat every recording as raw material
- 23Describe the tool before you build it
- 22Start with the trigger not the tool
- 21Time audit before you hire anyone