Workshop recording

How to Automate Project Time Tracking With Claude and ChatGPT

A project time tracking workshop for agencies, accounting firms, and professional services firms

Tuesday, November 3, 2026, 9:00 to 10:00am PT (12pm ET / 5pm Lisbon)

Last updated:

You can automate project time tracking by letting Rize capture your work in the background and letting Claude or ChatGPT tag, review and report on it through the Rize MCP. No timers, no timesheets. The workshop runs that loop in four steps: capture the data, organize it into clients and projects, review it with an agent and a quick human check, and improve how you work from what it shows. Rize records the app, window title and URL of your active window about every 15 seconds, never the contents, so your AI gets granular work data without screenshots. Routines run the review on a schedule and their output is available through the MCP, so an agent checks the tagging before you do.

Get the setup, the prompts for each step and the Friday review routine in the AI Time Tracking Playbook →
How to Automate Project Time Tracking With Claude and ChatGPT (44 min). Watch on YouTube

The step-by-step playbook

The AI Time Tracking Playbook

Everything from the session in one place: the two minute setup, the prompts for each step, the Friday review routine and a prompt cheat sheet. Free, no sign-up.

  1. 1Set up Rize and connect it to Claude or ChatGPT in two minutes
  2. 2Capture: what Rize, connectors and MCP feed your AI, with prompts to try
  3. 3Organize: pick what to track, then let the AI set up and tag your time
  4. 4Review: the weekly review prompt and the Friday review routine
  5. 5Improve: the monthly spot, decide, change, check loop and its prompts

Key takeaways: how to automate project time tracking with Claude and ChatGPT

  • Time is your inventory. If you run an agency or a firm, it sets capacity, margins and pricing, and you can't improve what you don't measure.
  • Manual timers produce late, inaccurate data, and decisions built on bad data are worse than decisions built on none. Automatic capture fixes the data before you build anything on it.
  • Capture first. Rize records your active window about every 15 seconds and groups it into time entries. Connectors add your calendar, Slack and meeting transcripts, and MCP plugs all of it into Claude or ChatGPT.
  • Start simple. Decide the question you want answered, track only what answers it, and let the AI set up clients and projects from a description of how you work. Over-engineering the setup is the most common mistake.
  • Review daily for the first week. The system is learning how you work, so ask the agent why it tagged something, correct it, add a rule, and turn on the Tag review routine so an agent checks the tagging before you do.
  • Insight is not the goal. Set a monthly review that recommends changes, make them, and check next month whether they worked. That loop is how a team gets faster on purpose.

Questions from the live Q&A

Where is the playbook?
At rize.io, under Workshops, then Automate project time tracking. It has the setup, the prompts for each step, the Friday review routine and a prompt cheat sheet.

The four steps of automatic time tracking

Each step from the workshop in one line, with a prompt from the playbook to try in Claude or ChatGPT.

StepWhat you doPrompt to try
CaptureRize records your active window. Connectors add calendar, Slack and meetings. MCP feeds it all to your AI.What did I work on today?
OrganizePick the question you want answered, then let the AI set up clients and projects and tag your time.Anything you couldn't match to a client? Check my calendar to figure it out.
ReviewAn agent flags gaps, wrong clients and entries with no project. You approve the fixes.Review my week and flag anything that looks off.
ImproveSpot what took too long, pick the fix, make it, and check next month that it worked.What repetitive work took the most hours? Rank it by hours per month.

Where the time data comes from

What each source adds, and how to connect it.

SourceWhat it addsHow it connects
Rize desktop appApp name, window title and URL of your active window, about every 15 seconds. Never the contents.Install it on Mac, Windows or Linux. Tracking starts right away.
VisionScreen captures for the apps and websites you choose, where a window title is not enough.Turn it on per app or website. Keep the image or just a summary.
ConnectorsCalendar events, Slack messages, meeting transcripts and CRM clients.Add them in Claude or ChatGPT, or in Rize under Settings, then Integrations.
MCPThe open standard that lets Claude or ChatGPT read and change your Rize data.Add Rize as a connector in Claude or as an app in ChatGPT, then sign in.

Why automatic time tracking matters: the benchmarks

BenchmarkWhat it measuresSource
66.4%Billable utilization at professional-services firms in 2025, the lowest on record, against a 75% target.SPI Research, 2026 Professional Services Maturity Benchmark
9.9%Firm profitability (EBITDA), well below the 13.8% five-year average.SPI Research, 2026 Professional Services Maturity Benchmark
47%Accuracy of timesheets filled in weekly. Daily logging only reaches 66%.Accelo, Time IS Money study of 500+ professional-services workers
52%Of projects experience scope creep.PMI, Pulse of the Profession (2018)

The AI Time Tracking Playbook turns these numbers into a step-by-step process. Open the playbook →

What the workshop covers

  1. 1Capture: why your AI is only as good as its data, and how Rize, connectors and MCP feed it
  2. 2Organize: decide what to track, then let AI set up and tag your time
  3. 3Review: agents checking agents, with Routines and human review
  4. 4Improve: turn time data into changes, and check that they worked

What you leave with

  • A prompt library for project time tracking in Claude and ChatGPT
  • A weekly project update routine
Read the AI Time Tracking Playbook →
Macgill Davis

Hosted by Macgill Davis, CEO and Cofounder of Rize. For agencies, accounting firms, and professional services firms.

Automatic time tracking FAQ

Yes. The full session is at the top of this page, and the AI Time Tracking Playbook is free too.

In the AI Time Tracking Playbook: the setup, the prompts to capture, organize, review and improve your time data, the Friday review routine, and a prompt cheat sheet.

No. Macgill connects Rize to Claude and ChatGPT live, so you can watch, or follow along in your own workspace if you already have one.

The app name, window title and URL of your active window, about every 15 seconds. It never captures the contents of the window unless you turn on Vision for specific apps or websites, and you can tell it never to track an app or website at all.

Correct it and add a rule so it sticks. Review your entries daily for the first week, since the system is learning how you work, and ask the agent why it tagged the time that way. The Tag review routine checks the tagging for you on a schedule.

Yes. Every time entry is available through the Rize MCP and API, so Claude or ChatGPT can hand them to a custom invoicing flow. Rize also has built-in invoicing that generates and sends PDF invoices.

Yes. Install the Rize Chrome extension in each profile and give it a custom name in the window title. Rize sees that name in every tab and tags the time to the right client or business.

Workspace admins see every team, team admins see their team, and members see only their own data. Rates and profitability need the finance permission. App, website and Vision data never leave the individual; only the time entries built from them are shared.

Automate project time tracking with Claude and ChatGPT.

Automate project time tracking for your own team. No timers, no timesheets.

Start a Rize team trial

Watch the workshop recordingRegister for the next workshop