I grew up in Lafayette and spent a lot of my childhood in Walnut Creek, from summer camps to the bike trails at Heather Farms. My parents live here now.
I still run Prozan AI from Lafayette, helping businesses across Walnut Creek put AI to work. For its offices, that might mean reading referrals and new-patient forms into your records, drafting routine letters and renewal reminders for you to approve, or answering the phone when the front desk can’t. I build those systems into the tools you already use, and you review the work before anything goes out.
More offices than shops
About one in five businesses here is a professional office and about one in nine is in finance or insurance, both well above the county’s share. About one in eight is a medical, dental or other health practice. Just over half of businesses with employees still have fewer than five.
Inside those offices, the same few people answer the phones, read the paperwork and chase the follow-ups. That’s where I’d look first.
Where I’d start
Intake at a medical or dental office. Referrals, insurance cards and new-patient forms arrive as faxes, PDFs and phone photos, and someone retypes each one. AI reads them into your system, flags what’s missing and drafts the call back, so staff spend the morning on patients instead of data entry.
The document pile at a law or accounting firm. A new matter or a new tax client brings a folder of statements, leases and letters. AI pulls the dates, names and figures into a summary your team checks, then drafts the routine letters from it.
Renewals at an insurance agency. Renewal dates, missing signatures, the client who still hasn’t sent last year’s paperwork. AI keeps the list current, drafts each reminder in your words and puts it in front of you to approve.
What I’ve built, and how it works
I studied data science at Berkeley and worked as a data scientist at Meta and DoorDash. For Private Fleet Net Zero, a freight marketplace, I built software that took over a 13-step check someone did by hand every morning. It runs a test deal from both sides, cross-checks 99 values in 8 places and emails the CEO before the workday, so conflicting details get caught before a customer acts on them. Within its first week, it identified two bugs missed during manual testing. The case study has the details.
The process is the same for every office. We start with a free 20-minute call to see whether there’s a worthwhile project; an assessment ranks which job to automate first; the implementation integrates with your practice or document software; your staff get trained on it; and I keep it running after launch. I’d rather do one properly than three halfway. Book a free call.
Questions people also ask
We’re a medical practice. Can AI work with patient information?
Yes, once the rules are set. It runs only on accounts whose terms cover health records, with the business associate agreement HIPAA requires, and it reads only the folders we agree on. Your staff checks anything clinical before it’s saved or sent. Any step that can’t meet those rules stays manual.
We have forty people on two floors. Do you work with firms that size?
Yes. In a firm that size one team usually carries most of the repeated work, so I’d start there: build it for them, count the hours it saves, then train the rest of the office on it.
If you run a business in Walnut Creek and any of this sounds like your week, book a free 20-minute call.