Jeremy Neely Text Jeremy

AI & Agents · Production

AI Hours Tracker

AI interviews staff and writes up each AI time-saving win for leadership.

Built for: The AME Group's engineers, Secure AI team and leadership

Starting a new entry, answering the AI interview and watching the record fill in (mock data, offline).
The tracker dashboard with hours-saved totals and full entries (sample data).
The tracker dashboard with hours-saved totals and full entries (sample data).

The problem

The AME Group logs every AI-built app and workflow in a SharePoint list to show leadership the value. The default form was tedious, so entries came in thin and inconsistent, and nobody could see total hours saved at a glance.

What I built

I built a SharePoint web part that replaces the list view with hours-saved totals and full, readable entries. Claude Haiku, through the Hatz AI gateway, interviews the author and drafts the description and business case.

How it works

  1. Pick Interview, One page or Wizard mode.
  2. The AI asks a few short questions about what was built.
  3. It drafts the description and business case as a live record fills in.
  4. The author edits the entry and saves it to SharePoint.
  5. One button drafts a monthly executive summary of all entries.
How AI Hours Tracker is wiredThe AI key lives in an Azure Function proxy, never in the browser. Claude Haiku runs through the Hatz gateway.
Comes inInside AI Hours TrackerGoes out Author's answers Interview, One page, Wizard SharePoint list existing entries AI interview a few short questions Draft copy description, business case Author edits live record fills in Saved entry to the SharePoint list Hours-saved totals dashboard view Executive summary monthly, one button

Built with

  • TypeScript
  • React 17
  • SharePoint Framework
  • Azure Functions
  • Hatz AI gateway
  • Claude Haiku 4.5
Text Jeremy about this Next: Meeting Summary Builder All 38 tools Texting opens your Messages app with a note already written. He replies himself.

More screens

Interview mode with the description and business case drafted in the live record (sample data).
Interview mode with the description and business case drafted in the live record (sample data).
One page mode turning a short brief into a full record (sample data).
One page mode turning a short brief into a full record (sample data).
The AI asking follow-up questions with quick-pick answers (sample data).
The AI asking follow-up questions with quick-pick answers (sample data).
A drafted monthly executive summary of all entries (sample data).
A drafted monthly executive summary of all entries (sample data).
Wizard mode, step one of four, asking what was built.
Wizard mode, step one of four, asking what was built.
Starting a new entry, answering the AI interview and watching the record fill in (mock data, offline), a frame from the demo clip (sample data).
Starting a new entry, answering the AI interview and watching the record fill in (mock data, offline), a frame from the demo clip (sample data).
Starting a new entry, answering the AI interview and watching the record fill in (mock data, offline), a frame from the demo clip (sample data).
Starting a new entry, answering the AI interview and watching the record fill in (mock data, offline), a frame from the demo clip (sample data).

More of Jeremy's work

All 38 tools, by category