Joshua Yenson

AI Operations Builder · Ho Chi Minh City, Vietnam

I find the leverage point in a messy workflow — and direct AI to build the fix.

I'm Joshua James Yenson. Not an engineer by trade — an operator who spent 5+ years running people and workflows in a fast-paced, multi-campus environment. When something was costing time or leads, I didn't wait for IT — I directed Claude and Claude Code to fix it myself. Now I build that same kind of fix for small businesses: missed calls, slow follow-ups, and manual admin, turned into tools that actually get used.

01
AI tools shipped & in daily use
5+

AI tools shipped & in daily use

02
team members managed day to day
20+

team members managed day to day

03
qualified leads sourced by an AI pipeline I built
184

qualified leads sourced by an AI pipeline I built

This is what changes

If a workflow is quietly costing you leads or time, I can probably fix it.

Missed calls, unanswered website enquiries, manual admin — the same pattern I fix for my own team at Vinalearn, built as a product for yours.

01

Before

  • A call comes in while you're on a job. It goes to nothing.
  • A website visitor asks a question after hours. No one answers.
  • By the time you follow up, they've already called someone else.
02

After

  • The caller gets an instant text back — the lead stays warm.
  • The website visitor gets an answer immediately, day or night.
  • You follow up when you're free, with the details already captured.
Start with the simplest fix

That's the whole shift.

Claude/Claude Code/Claude API/Anthropic SDK/n8n/Agent Reach/React/Next.js/SQLite/Prompt Design/Claude/Claude Code/Claude API/Anthropic SDK/n8n/Agent Reach/React/Next.js/SQLite/Prompt Design/

How I build

Same loop, every project

Five very different tools, one repeatable process — this is what actually happens between 'this is annoying' and 'the team uses it every day.'

STEP · 01

Spot the leverage point

I sit inside the workflow first — as the manager running it, not an outside consultant. The bottleneck has to be real and specific, not a hypothetical one.

STEP · 02

Scope the fix precisely

What surface changes, what data feeds it, what "good" looks like. A vague brief produces a vague tool — this is the step people skip.

STEP · 03

Direct AI to build it

I'm not an engineer — Claude and Claude Code do the implementation. My job is precise direction and reviewing every output against how it'll actually be used.

STEP · 04

Ship it, then iterate

It goes in front of the team that has to use it. I watch what breaks, diagnose root cause over quick fixes, and iterate until adoption is real, not rolled out.

The short version

Most people wait for a handoff.
I open the tools and build the fix myself.

Capabilities

Two skill sets, one loop

The reason these tools get used instead of shelved is that I've sat on both sides of the problem: I know the workflow because I run it, and I know how to direct AI to fix it.

AI & Automation

01
  • Daily, hands-on use of Claude & Claude Code
  • Workflow automation (n8n, project-based)
  • Prompt design & AI orchestration
  • Scoping & directing AI-built tools
  • Testing & iterating on AI output against real usage

Operations & Leadership

02
  • Teacher recruitment & onboarding, full cycle
  • Structured observation & evaluation frameworks
  • KPI systems & performance management
  • Change management & genuine adoption, not just rollout
  • Formal reporting & documentation

Next step

Have a workflow that's costing you leads or time?

I'm always interested in operational problems that a well-scoped AI tool could actually solve.