Case Studies
What “practical AI” looks like.
Three anonymized small-business engagements — one of each tier. Real outcomes, real cost math, and (where it’s the right answer) clients we told to skip the spend.
Home services
Build tier+9 pts
closed-quote rate over 60 days
Quote follow-ups that actually go out — for a 6-tech HVAC shop
Owner was sitting on roughly 40 unsent quote follow-ups a week because nobody had time. We wired Claude to their existing CRM email so every quote that hadn't been responded to in 48 hours got a personalized, plain-English follow-up — owner reviewed and clicked send. Closed-quote rate moved from 18% to 27% over the next 60 days.
- follow-ups now sent vs. ~5 before
- ~40/wk
- from scoping call to live
- 9 days
Legal services
Plan tier~7 hrs/wk
saved on intake review
Intake call summaries the partner can actually read — small law firm
Managing partner was spending 90+ minutes a day reading intake notes from the paralegal. Plan tier laid out the exact tool stack (existing call recording → Whisper → Claude → email summary) and the prompts to use. Firm implemented it themselves with the plan. Partner now skims a 4-bullet summary in 90 seconds.
- plan delivered
- 8 days
- firm built it themselves from the plan
- $0 setup
E-commerce
Audit tier~$22k/yr
saved by skipping the wrong vendors
AI Stack Audit for a $2M D2C brand — what to skip and what to ship
Owner was being pitched four AI vendors quoting $1,800/mo combined. Audit ranked all of them against actual workflow impact, recommended skipping three, and flagged one in-house Claude workflow worth ~$28k/year in saved labor. Owner chose to run the in-house option themselves; came back six months later for an Implementation Plan on the next workflow.
- audit delivered
- 5 days
- vendor pitches recommended to skip
- 3 of 4