Field notes.

What I have learned building AI into other people's operations: how the work gets chosen, built, reviewed, and owned.

SeriesField notes
Issued
Sheets42

Note index

Newest first. Date issued · subject · reading time.

  1. How should a small or mid-sized company plan AI automation for 2027?Plan AI automation for 2027 one workflow at a time: pick it, measure a baseline, run a bounded pilot with stop criteria, and decide each quarter.9 min
  2. How to budget for AI automation in 2027Budget AI automation for 2027 one workflow at a time: one-time and recurring costs, review time, a reserve, what to measure, and when to stay manual.9 min
  3. How to make a SaaS product agent-ready in 2027A 2027 planning checklist for agent-ready SaaS: an MCP server with job-shaped tools, OAuth scopes, discovery files and tests, and when a generator is enough.9 min
  4. Your customers already told you what to build. You just can't find it.A plain-file feedback inbox that keeps every customer quote verbatim, groups quotes into issues, and ranks them by how many people asked. Free Claude skill.3 min
  5. Scope the review step before you price the automationTurn one recurring workflow into an automate, assist, discovery, or stop recommendation with review, acceptance, and illustrative economics.3 min
  6. Build a voice guide from the writing you already haveTurn authorized writing samples into channel-specific voice rules and a shareable drafting prompt.3 min
  7. A prospect brief should tell you when to stopTurn public business evidence into cited account hypotheses, contrary explanations, rejection rules, and unsent outreach drafts.3 min
  8. Your useful pages may be missing from your own navigationCompare a site's page inventory with its actual navigation paths and prepare relevant repairs for disconnected content.3 min
  9. Stop reading signup totals. Read the names.An instant alert for every new account, a morning list of who got stuck and where, and a drop-off review that walks your real funnel. Free Claude skill.3 min
  10. The list your AI can't finish for youAgents can edit code and check results, but shouldn't pay, sign in as you or message customers. A dated owner-action list draws that line. Free Claude skill.3 min
  11. If AI agents work for you, someone has to write the daily reportA daily brief built from evidence, the commits and requests typed into AI sessions, showing what moved, what's stuck and what was promised. Free Claude skill.3 min
  12. Audit your homepage like a skeptical buyer wouldA claims review that grades your copy against the 22 Immutable Laws of Marketing and checks every number and logo against your evidence. Free Claude skill.3 min
  13. Your product already has a moment people would share. It's just hidden.Find the one screen a customer would screenshot and share, then clear the gates, crops and generic previews that keep it from spreading. Free Claude skill.3 min
  14. Before you spend a dollar on ads, click your own adTen ad concepts with real hooks, exact targeting, and a click-by-click check that every ad's path works through to the payment step. Free Claude skill.3 min
  15. Build the marketing kit around one dated truth sheetCreate coordinated one-pagers, press facts, and content calendars from dated product truth, approved evidence, and brand assets.3 min
  16. Give every article a claim ledger before you give it a hookProduce a useful sourced article through distinct drafting, factual editing and style-review passes.3 min
  17. Turn completed work into a pitch an editor can useFind a public-interest angle in documented work and create outlet-specific articles, pitches and submission notes.3 min
  18. A correct total can hide an incorrect AI answerReview AI output against original source evidence with explicit omissions, false positives, source conflicts, and acceptance boundaries.3 min
  19. Agentic AI vs workflow automation: which does your business need?Choose between a fixed workflow and a model-directed AI agent using practical criteria for variable work, permissions, retries, human review, and acceptance.5 min
  20. Local AI vs cloud AI for business: choose by the workflowCompare local AI and cloud AI against data boundaries, review effort, reliability, and operating ownership before choosing a business pilot.5 min
  21. Private AI for business: a data checklist before the pilotA practical private AI checklist covering document permissions, processing locations, logs, retention, integrations, and pilot acceptance evidence.6 min
  22. Open source AI for business: check the rights and the operating planHow to assess open source and open-weight AI for business: model licenses, component inventory, workflow tests, maintenance, and a usable handoff.5 min
  23. Before you buy an AI agent, ask to see the testsFive pieces of evidence to request before buying an AI agent, plus a dated Soxoa Studio test that failed and changed what we built.5 min
  24. Muse for small business: three workflows worth testing firstA practical guide to testing Meta Muse for small business: campaign prep, customer follow-ups and monthly review, with clear ownership and measured results.4 min
  25. Instinct for small-business owners: where a personal AI assistant fitsA practical look at Instinct for business owners: personal assistant use cases, current access and pricing limits, and what to clarify before a team pilot.4 min
  26. OpenAI dots for small business: give an agent one job it can ownWhat OpenAI dots can do, who can access them, and how a small business can test one ongoing workflow with clear instructions, review, and a stop point.4 min
  27. What AI consulting costs in Sacramento depends on the workA practical guide to the factors behind an AI consulting proposal in Sacramento: workflow scope, data, integrations, human review, operating costs, and travel.6 min
  28. Automation assessment worksheet: choose a workflow worth testingUse a blank worksheet and a completed invoice intake example to scope an automation pilot, measure review time, and define when to stop.4 min
  29. What actually happens in a one-day live buildThe hour-by-hour shape of a Live Build Workshop: choosing the workflow, red-lining it, building on real examples, breaking it on hard cases, naming an owner.5 min read
  30. How to pick the first workflow (and the three you should not start with)Use five screening tests and a worked scorecard to compare automation candidates by frequency, variability, exception rate, and ownership.7 min read
  31. Leave it manual: the most useful note on the drawingDecide when to keep a workflow manual. Weigh volume, review time and the cost of errors, then record when to revisit the decision.5 min read
  32. The first ninety days of a fractional Chief AI OfficerWhat an outside AI lead does in three months: inventory and decisions, one build through a review gate, then a policy people follow and a first scorecard.5 min read
  33. The stop decision: a build sprint is allowed to end with nothing installedEvaluate on real examples before installing, agree the bar before you look, and see why a written recommendation not to build is a deliverable worth paying for.5 min read
  34. Your AI readiness is mostly an ownership questionUse the five dimensions of AI readiness and a one-workflow ownership and access inventory to decide what your business can test safely.6 min read
  35. AI Automation for Sacramento Contractors: Bids, Change Orders, and Field PaperworkWhere AI automation works for contractors: bid drafting, change orders, submittals and RFIs, field paperwork, progress billing, and what should stay human.14 min read
  36. CPA Firm Automation: What Sacramento Accountants Automate Before Tax SeasonWhat Sacramento CPA firms automate before tax season: document intake, organizer chasing, engagement letters, prep checks, status updates and what stays human.12 min read
  37. The Real Monthly Cost of Running AI (And How to Keep It Sane)Build a monthly operating budget for AI automation. Track usage, staff review and monitoring, and plan who can pause the workflow and cover it manually.8 min read
  38. How to Audit Your Business for AI Opportunities: The Four-Phase Method I UseBefore you buy an AI tool, run the audit: inventory every recurring task, score the candidates, fix the top three properly, then verify the hours came back.8 min read
  39. AI Agents, Explained for Business Owners (Minus the Hype)What “agent” actually means, the work agents hold up on today, where they still fail, and the three guardrails I would not deploy one without.9 min read
  40. Speed to Lead: The Cheapest AI Win for Sacramento Service BusinessesService businesses lose inquiries in the gap between when a lead arrives and when a human is free. Here is what an AI follow-up system does about it.7 min read
  41. AI for Sacramento Law Firms: Intake, Drafting, and Getting Hours BackWhere AI fits around legal work in a small firm: intake, document assembly, billing narratives. And the parts that stay with an attorney.8 min read
  42. AI for Medical Practices in Sacramento: What’s Actually Worth the InvestmentWhere AI earns its place in a medical practice: prior authorization, intake, eligibility, care gap outreach. And what to leave alone for now.9 min read

Disagree with one of these?

Bring the workflow you think proves me wrong. Thirty minutes, with me, and an honest answer about whether there is anything worth building.