Skip to content
Vidhu Builds
All work

Case study 02 / 04

Cialfo

Built a notifications system, a multi-agent AI copilot for counselors, and the usability fixes in between.

Role
Product
When
Sep 2025–present
Company
SaaS for international schools, used by 300K+ students and 3,000+ counselors.
SaaS for schools Sep 2025–present

+33%

Results

What changed

  • +33%

    monthly active users, from a notifications system

  • +30%

    sessions per user per month, from the same notifications system

  • −23%

    dead clicks, from usability work

  • −25%

    counselor workload and time to value, from a multi-agent AI copilot

  • −30%

    related support tickets, from document-submission workflow discovery

  • −35%

    quick-backs, from the same usability work

My role

Product, since September 2025, on a platform used by 300K+ students and 3,000+ counselors at international schools. I work on engagement, AI for counselors and the workflows around them.

A notifications system

The problem Students and counselors missed updates that needed their action, so they came back less often than the work required.

What we did Designed and shipped a notifications system that brings people back to the task waiting for them.

The result +33% monthly active users and +30% sessions per user per month.

A multi-agent AI copilot for counselors

The problem Counselors spend a large share of their week on repetitive work around each student.

What we did Built a copilot made of several AI agents, each handling one part of that routine work.

The result −25% counselor workload and time to value.

Document submissions

The problem Document submission generated a steady stream of support tickets.

What we did Ran discovery on the workflow to find where people got stuck, then fixed those steps.

The result −30% related support tickets.

Usability

The problem People clicked on things that did nothing, and left pages moments after opening them.

What we did Found the patterns behind it and fixed them, screen by screen.

The result −23% dead clicks and −35% quick-backs.

What I learned

Engagement problems are often notification problems. People don’t come back to a product that never tells them there’s something to do.

An AI copilot is judged by the hours it gives back, not by how impressive it looks in a demo.

Got something worth building?

Book a 30-minute call, or tell me about it in a few lines. I’ll reply with questions, or a time to talk.

Book a call (opens Calendly in a new tab)