Non-profits and small to medium-sized businesses · Austin, TX
Get the hours back.
Keep the people.
AI should take the repetitive parts of the job, not the job.
One call, then four phases, one at a time.
30 min
Enough to tell you whether I can help. One discovery call, no deck, no paid discovery phase.
0
Reseller agreements, referral fees or software commissions. Tools get recommended on merit or not at all.
60 days
Start to finish, discovery call to handoff. Something running in production by week four. Then I leave.
If the Fit Check finds nothing worth building, you don't pay for it.
Who it's for
Small teams where one person holds it together.
Non-profits
Grant reporting, donor communications, program data, board policy. Most teams already use AI ad hoc. Almost none have a repeatable workflow or a written policy the board has seen.
Small and medium-sized businesses
Quoting, scheduling, customer follow-up, back-office reporting. Ten to two hundred people, one office manager holding it together. The owner has tried ChatGPT. Nobody else has.
Not a fit: Organizations that want a chatbot on the website and nothing else, or a 40-page roadmap with no build.
Easy wins
Where the hours usually hide.
Low risk, high repetition, and a person still in the loop.
Non-profits
First draft of the grant report narrative, assembled from last cycle's boilerplate and this cycle's numbers
A person edits every draft before it goes out
Re-keying intake forms, applications, and sign-up sheets into the CRM
No judgment involved; spot-check weekly
Donor thank-you letters and acknowledgments, personalized from the gift record
Human sign-off on anything donor-facing
Board packet assembly from the reports you already produce
Someone still decides what goes in
Small and medium-sized businesses
Quotes and estimates drafted from a template plus the job details
The owner approves every number
Customer follow-up after a quote, a job, or a missed call
A person reviews before it sends
Scheduling and confirmations, with the back-and-forth handled for you
Anyone can override the calendar
Weekly back-office reporting pulled from the systems you already use
The numbers are checked, not trusted
How it works
One discovery call. Four phases.
A workflow is one repeatable task that eats staff time. You choose which ones to fix; starting with one is fine. Each phase is fixed price, quoted after the discovery call. No hourly billing, no retainer.
- 00Free
Discovery call
30 minutes
You describe how work gets done. I say whether AI is worth your time this year.
More, and what you get ↓Less ↑
You describe how work actually gets done. I tell you whether AI is worth your time this year. Sometimes the answer is no.
- 01Priced
Fit Check
Two weeks
Every repeatable task I can find in two weeks, ranked by hours saved and by risk.
More, and what you get ↓Less ↑
I shadow the work and list every repeatable task that eats staff time, as many as two weeks allows, ranked by hours saved and by risk. You get the whole list, with the ones worth building first marked. If it finds nothing worth building, you don't pay for it.
You get
- Every workflow found, ranked by hours saved and by risk
- The workflows I looked at and rejected, with reasons
- A go / no-go on whether this is worth your year
- 02Included
AI Governance
Drafted in the Fit Check, proposed before the build, signed before handoff
A model proposed to your board before anything is built. Signed before handoff.
More, and what you get ↓Less ↑
If your board has no AI governance, I propose a model for it; if it has one, I work inside it. It starts in the Fit Check with the first decision, what never goes into a tool, and ends with a policy your board signs: acceptable use, data classification, human sign-off with named owners, an approved-tool list, and a do-not-build list.
You get
- Acceptable-use policy and data classification, in plain language
- Human sign-off list, approved-tool list, and a do-not-build list
- A named owner for every rule, and a revision pass after board feedback
- 03Priced
Workflows Fixed
Four weeks
The ones you select, built in the tools you have, with a working demo every week.
More, and what you get ↓Less ↑
You choose which workflows from the list. I build them alongside your staff, in the tools you already pay for wherever possible. You see a working demo every week, and results are measured against the Fit Check baseline in hours and dollars. Starting with one is fine.
You get
- The workflows you selected, built and running against real work
- Before and after, measured in hours and dollars
- Documentation the next hire can follow
- 04Included
Handoff to Your Team
Two weeks, then a 30-day check-in
Training, a named owner per workflow, documentation, and a 30-day check-in.
More, and what you get ↓Less ↑
Training on the workflows we built, not AI in general. An owner named for every workflow and every rule. Documentation the next hire can follow. Thirty days later I check in to see what stuck.
You get
- Hands-on training on the workflows we built
- A named owner for each workflow
- A check-in 30 days after handoff
What a Fit Check hands youThe one-page list, in the shape you’ll get it.
This is an illustrative example for an invented organization, so you can see the shape before you pay for the real one. Every number is labelled with where it came from, because a finance officer assumes the worst case if you don’t say.
| Workflow | Hours per week | Risk | Verdict |
|---|---|---|---|
| Grant report narrative, first draft | 6 (their estimate, unmeasured) | Low. Reviewed by a person before it leaves. | Build first |
| Re-keying intake forms into the CRM | 4 (counted, one week) | Low. No judgment involved. | Build |
| Donor thank-you letters | 3 (their estimate, unmeasured) | Medium. Donor-facing; needs sign-off. | Build, with a human gate |
| Board packet assembly | 5 (their estimate, unmeasured) | Low. | Do it yourself — I’ll show you how |
| Case notes summarization | 8 (their estimate, unmeasured) | High. Beneficiary data. | Not yet. Policy first. |
Rows 1 to 3 are the ones I would build first; you choose how many. Row 4 is yours to do. Row 5 waits until the policy exists, and I’d rather say that now than after.
What’s asked of you
- A named decision-maker who can say yes
- Access to the documents the work runs on — no client or donor data
- Half an hour a week
- Two or three staff free for the handoff
Your data, as contract terms
- During the build, anything of yours that touches AI goes through a commercial account whose terms prohibit using it to train a model, with 30-day retention. Nothing connects to a system of record.
- What I hand you runs on an AI account you own, on your bill. I keep no access after handoff.
- No client, donor or beneficiary data is requested. My working copies are deleted after handoff, with written confirmation.
What I'm building right now
Two workflows in progress, described as they are.
A room-booking system for a non-profit job-training center
Three rooms, a public calendar, a staff approval queue, recurring bookings, and a guidelines agreement every requester accepts. It replaces a form-and-email loop a small staff was holding together by hand, and every rule they need to change is editable by them.
A CRM and outreach engine for a digital-raffle fundraising company
Lead intake, AI-drafted follow-ups a person approves before anything sends, and lead-sheet scanning. Every AI call is logged with its cost, there is a daily spend cap, and failures alarm rather than fail quietly. The human gate is built in, not promised.
The best tip I give away
Stop writing prompts. Make the model write them.
You know your problem. You don't need to know how to phrase it for a language model — that part is the model's job, and it is measurably better at it than you are. Describe the job, make it write the prompt, then run it.
Not folklore: model-written instructions matched or beat human annotators on 19 of 24 tasks (Zhou et al., 2022), and self-optimized prompts beat human-designed ones by up to 50% on harder reasoning benchmarks (Google DeepMind, 2023).
Copy this prompt
Write me a comprehensive prompt to put into an LLM that will address this need: [describe the need in plain language, including who the output is for, what format you want, and anything that would make the result wrong]. Before you write it, ask me any questions you need answered to write a better one.
Why WiderGood
Recommendations come from use, not slides.
Fifteen years running operations and analytics in healthcare, including quality analytics for more than 2 million patients across roughly 2,600 sites, and a first-ever data warehouse and 150 governed metrics for a multi-clinic group. I have watched the consultant's deck arrive, get admired, and get filed. My job was the part that came after: building what it described, or didn't, and making it run with the teams on the ground.
I also build and ship my own AI products, so recommendations come from use, not slides.
More about the practice →- The smallest useful thing first
- Nothing you cannot maintain
- The skeptical case, written down
- Plain language, always
Questions
What people ask before they book.
- Do we need an AI strategy first?
- No. Strategy comes from what worked in the first workflows you fixed, not the other way round.
- Which tools do you use?
- Whatever you already pay for, where it can do the job. I take no commissions, so the recommendation is the recommendation.
- Our data is sensitive.
- That is what the policy is for: least-privilege access, human sign-off on anything customer- or donor-facing, and no sensitive data in consumer tools. During the build, anything of yours that touches AI goes through a commercial account whose terms prohibit training on it, with 30-day retention. What I hand you runs on an AI account you own. My working copies are deleted after handoff with written confirmation. Those are contract terms, not assurances.
- Will this replace staff?
- No. It replaces the parts of the job nobody wanted. PwC's 2025 Global AI Jobs Barometer, built on close to a billion job ads, found jobs still growing even in the roles most exposed to automation.
- What do we commit to?
- One phase at a time. Each phase has a milestone you sign off on before the next one starts, and either of us can stop at any milestone. Nobody signs up for sixty days on day one. During the build you see a working demo every week.
- What does it cost?
- Fixed price per phase, quoted after the discovery call. No hourly billing, no retainer. The Fit Check carries a guarantee: if it finds nothing worth building, you don't pay for it. A typical first engagement is the Fit Check plus one workflow: about six weeks.
- What happens after you leave?
- Each workflow has a named owner on your side and documentation the next hire can follow. Thirty days after handoff I check in to see what stuck and what didn't. If something broke, that call is where it gets fixed.