AyZar career development, from a week to 5 minutes
AyZar Outreach helps people plan their careers, but the process behind it was slow and manual. I brought them in as MyPost's first retainer client, then managed the build of the AI workflow that now prepares each student.
- Role
- AI Product & Project Manager, MyPost
- When
- 2026
- Client
- AyZar Outreach, MyPost's first retainer client
- Stack
- TypeScript · Claude API (the AI agent) · Supabase
Student information
Staff drop in what they know about a student, in one place.
Brief + research
The AI reads the student, proposes career paths, and researches each one.
Roadmap
It turns the chosen path into concrete steps the student can follow.
Complete package
It assembles an employer-ready package the student can send out.
- 5 minutesfor a full student packagedown from nearly a week of manual prep
- 2 weeksto process 50–60 studentsdown from 4–6 weeks of work
The problem
AyZar had plenty of resources for students. What it didn't have was a reliable way to get the right one to the right person at the right time.
Preparing for a single student meeting took the team nearly a week. Across a cohort of 50–60 students, that meant 4–6 weeks before everyone had a plan. The question: how do we let AI do this much faster, and keep the work just as good?
What we built
One central AI agent that understands the student, built as a few narrow components that each do one job, instead of one big system. It runs the four steps shown above, from student intake to the finished package.
The approach, and why it's portable
A good AI is one you barely notice. The same four steps carry over to recruitment, employee training and upskilling. I later presented them as a talk, AI in the Workforce: Building Practical Solutions.
Pick one bottleneck
Not the whole pipeline: the single most painful step. Here, getting ready to meet a student.
Audit the data first
Garbage in, garbage out. The student information had to be clean before any AI touched it.
Layers, not one tool
Narrow pieces that each do one job well: one to gather and interpret, one to act, one to report.
Measure what changes
Time per student and students per cycle: the numbers that say whether it's worth scaling.
My role
- Met AyZar's outreach team at an industry event, pitched the AI work, and won the meeting.
- Mapped their career-development process and picked the bottleneck worth automating.
- Designed the workflow and managed its build, from student intake to the finished package.
- Measured the before and after in time per student and per cohort.