AI infrastructure and role-based use
Start with business goals, existing systems, and role-specific tasks. Combine flagship and economical models, then configure accounts, permissions, and workspaces so employees can use them at a controlled cost.
AI Product Manager · Enterprise AI Deployment (FDE)
I adapt AI tools to each company so every role can use advanced models affordably. Through consulting and hands-on coaching, I turn workflows into Skills and systems into MCP tools, with shared data for people and agents. People keep core judgment, confirmation, and correction.
When an AI initiative stalls between business, product, technology, and organization, I connect the boundaries.
Start with business goals, existing systems, and role-specific tasks. Combine flagship and economical models, then configure accounts, permissions, and workspaces so employees can use them at a controlled cost.
Through consulting and hands-on coaching, turn internal workflows into Skills and systems into MCP tools with matching Swagger/OpenAPI documentation. Consolidate siloed data into a core database shared by people and agents.
Build a usable demo close to the business to confirm requirements, then engineer for the user scale or work with the engineering team. Use role-based training and hands-on coaching to hand over acceptance, operations, and iteration.
Experience across these organizations and business contexts
My advantage is not one tool. It is putting business language, product structure, and implementation into one delivery chain.
From November 2025 to April 2026, connected PRDs, meeting notes, Feishu knowledge, and database queries in one agent workflow during an ERP implementation.
From October 2021 to August 2024, built process, CRM, and operating systems from zero to support decisions rather than simply digitizing paper forms.
40+ business workflows
From October 2021 to August 202410000+ workflow runs each year
CRM with 872 client records
Since July 2026, I have built an employee AI workspace for an SME, bringing models, business tools, and Skills together. The Windows desktop client is in a small, controlled pilot.
Start with business goals, existing tools, and role-specific tasks. Choose suitable models and platforms, then use a working demo to confirm requirements and boundaries.
Through consulting and hands-on coaching, turn workflows into Skills and systems into MCP tools with API documentation. Consolidate siloed data into a core database shared by people and agents.
Let AI handle information flow while people retain core judgment, confirmation, and correction. Hand role-based guidance, acceptance, and operations to the team.
I maintain another, loop, and OpenCode Annotate for session management, product workflows, and review feedback. another builds on an MIT-licensed fork; OpenCode Annotate reuses upstream components. My work covers adaptation, integration, and delivery.
View all public projectsFor unclear direction, repeated solution churn, or an independent technical and product judgment.
For moving from a scoped plan to a working system and operational handover.
For aligning methods, establishing standards, and completing one real scenario together.
Pricing follows project scope, collaboration depth, and delivery mode. I do not sell generic model subscriptions.