§ 05 · Ask Kevin · AI assistant
Ask Kevin
AI answers based on my published work · You can check the sources
Ask Kevin · AI
Ask about the work Kevin has done, how an engagement starts, or what it would cost. Answers come from his published case studies, and every source is one click away.
Recorded example · 2026-09-10One question · 3 sources, word for word · No hosted model
Specimen · Recorded run · 2026-09-10 · Corpus b588a6f3 · No generative model · Local embedding + verbatim extraction
“We run Salesforce, three legacy databases, and our salespeople qualify RFQs by hand. What could Kevin build for us?”
No generative model · Local embedding + verbatim extraction
- FAQ · What does Kevin Fincel deploy?Top sim 0.663
What does Kevin Fincel deploy? Kevin builds AI agents and MCP servers, retrieval systems, and integrations with ERP, CRM, analytics and legacy software. The work includes testing answers, access controls and human approval for sensitive changes.
- Case 06 · Agent-written changes to live ad accounts, safely · ProblemTop sim 0.627
The team manages about 18 industrial-sector ad accounts. Agents can help with audits and campaign builds, but API changes affect live client spend. I needed a review process that made each change traceable and required a person to approve it before execution.
- Case 04 · A live ERP, an API-first storefront · ApproachTop sim 0.618
I separated the API, sync worker and Next.js storefront. The worker mirrors NetSuite into Postgres; the storefront reads through the API and never calls the ERP during a page request. A native part-number field was filled on 99.4% of sellable items. That gave us the grouping key for roughly 28,100 products with condition variants, without adding a client field. The sync fetches before writing, compares content hashes and refuses to reconcile a suspiciously small response. The database allows one runner at a time. CI checks OpenAPI drift, privileged routes fail closed, and a nine-tool read-only MCP server uses the same service layer as REST.
This example was recorded at build time using the same model as the browser. It uses source passages word for word; no hosted model wrote the answer.
You can question any sentence and send it to me for a reply.