I build AI products that turn data into decisions.
I'm John Tan — a repeat founder, CPO, and AI operator based in Napa. Across Auggie, PVAI, and CorkVision, I build systems that codify expert workflows, surface customer signal, and help teams make better decisions at the point of action.
18× MARATHON FINISHER · 4× IRONMAN FINISHER · 70.3 WORLD CHAMPIONSHIP FINISHER
Expert workflows → decision systems.
Every venture I build starts the same way: spend time with the people doing the work, map the decision they're actually trying to make, then build the shortest possible path from fragmented signal to a clear next action.
Not AI for AI's sake. Real moats — built from workflow context, customer signal, and the trust of the people closest to the problem.
One thesis. Four proof points.
Auggie
Customer intelligence that doesn't disappear after the research report.
AI-powered consumer insights and virtual focus group platform for marketers. Gives brand teams a persistent, queryable customer intelligence layer organized around Brand Workspaces — synthetic personas, real customer signal, and live interactive focus group sessions in one place.
Co-Founder & CPO — product strategy, platform architecture, end-to-end product development. Launched publicly April 2026.
- Virtual focus groups
- Customer Signal Engine
- Persona enrichment
- Customer Voice Map
- Insight traceability
Explore Auggie →
PVAI Consulting
Turning enterprise workflows into scalable AI systems.
An AI automation consulting practice. PVAI helps enterprise teams identify, design, prototype, and deploy AI automation — built around structured discovery, automation-opportunity mapping, RACI clarity, gap analysis, interactive prototypes, and bottom-up ROI models. Engagements span $140M+ in annual marketing spend, paid-search automation, and research-methodology productization.
Co-Founder & CPO
Selected engagement outcomes
- Campaign brief turnaround30–45 min→Under 5 min
- Monthly research analyst hours666 hrs→91 hrs
- Insight coverage across studies40%→100%
Anonymized; figures are from production deployments.
- Enterprise AI automation
- Workflow discovery
- ROI modeling
- Interactive prototyping
- Data · Tools · People
Explore PVAI →
CorkVision
An AI copilot for winery decisions.
Helps wineries sell more, operate smarter, and decide better — without replacing the platforms they already use. Corky turns fragmented winery data into action plans, not another dashboard. Built from 150+ winery conversations, a harvest at Larkmead, and beta traction with 18+ wineries since August 2025.
Founder
- Winery AI copilot
- DTC operations
- Club · Inventory · Marketing · Hospitality
- Commerce7 · WineDirect · Mailchimp · ShipCompliant
- Action plans, not dashboards
Visit CorkVision →
Afterglass
Remember every bottle worth remembering.
Remember every bottle worth remembering — a consumer wine companion built around WSET-style tasting grids. Where CorkVision serves the winery, Afterglass serves the wine lover.
- Wine memory
- Tasting grids
- Consumer wine
- WSET
Discover Afterglass →
AI needs workflow context, not just a chat box.
A chat box answers questions. It doesn't price a release, allocate an inventory pool, or prioritize a research question. The teams winning with AI embed it inside that real decision — and earn the trust of the person whose name is on the outcome. That's where the durable advantage sits: in how the work actually gets done, not in whichever model is under the hood.
Why these problems, why now, why me.
Enterprise infrastructure, scaled and exited.
Co-founded Alphaserve Technologies, a global IT services firm acquired in 2020. Operational rigor, global teams, vendor trust.
Decision systems, not dashboards.
Auggie, PVAI, and CorkVision share a pattern: codify the expert workflow, then surface the next decision — with the receipts.
150+ winery conversations. A harvest at Larkmead.
I don't build for industries I haven't sat inside. Two years embedded with owners, GMs, and DTC leads. Napa is the proof — and the testbed.
- 01Vertical AI: why workflow context beats model size
- 02AI adoption inside traditional industries (wine, hospitality, ops)
- 03From research report to persistent customer intelligence
Building AI that earns its place.
A loosely connected set of essays from inside the build. The model is rarely the point — what's defensible, what's trustworthy, what actually changes how someone works lives somewhere else.
June 2026 · Field Notes
Why the Moat Is the Workflow, Not the Model
Frontier models are a commodity input. What compounds — and what a competitor can't simply buy — is the workflow underneath: the accumulated context, the proprietary data, the traceable trust. A model is a capability. A workflow is a position.
June 2026 · Field Notes
From Research Reports to Persistent Customer Intelligence
A report answers the questions you thought to ask the day you commissioned it. A persistent intelligence layer answers the questions you have today, against everything it has learned about your customers up to this moment — and it's a little smarter than it was last week.
June 2026 · Field Notes
What Wineries Actually Need from AI
After 150+ conversations with winemakers, DTC managers, and owners, the pattern is clear: wineries don't need another dashboard or chatbot. They need help inside the tools they already live in — Commerce7, Mailchimp, ShipCompliant — that respects compliance and earns trust before it asks for any.
"A journey of 140.6 miles begins with a single step."
Before all of it — the companies, the AI, the Napa years — there was endurance racing. The humor, the discomfort tolerance, and the willingness to sign up for hard things before feeling ready. The archive lives on.
18x Marathoner Finisher · 4× IRONMAN Finisher · 70.3 World Championship Finisher
Read the archive →