Setup in 2 minutes
- Create a Rize account at rize.io and install the desktop app. Tracking starts right away.
- Connect Rize to your AI.
- Claude: Settings, then Connectors, then add Rize and sign in.
- ChatGPT: Settings, then Apps and Connectors, then add Rize and sign in.
- Check the connection. Ask: “What clients and projects do I have in Rize?”
- Just work. You don't start or stop timers. Rize pulls in your time and the AI tags it.
Need the connection details? Read the Rize MCP setup guide →
Capture
Your AI is only as good as its data. With no data, Claude or ChatGPT gives you a generic answer. With your data, the same model gives you a precise one.
Where the data comes from:
- Rize tracks your work automatically in the background. No timers.
- Connectors add context: your calendar, Slack, meeting recorder, project manager and email.
- MCP is the standard plug that connects these tools to your AI, like USB-C for AI. It's an open standard, so this approach works with or without Rize.
Prompts to try:
What did I work on today?
Summarize my week by client and project.
Which meetings this week were for which clients?
Organize
Start with the question you want answered, then track the minimum that answers it. For most teams that means clients and projects. Add tasks, labels and contracts later, when you need them.
| Your question | What to track |
|---|---|
| Are we billing for all the work? | Clients and projects |
| Which clients make money? | Clients, projects and contracts |
| Who has capacity? | Team members and projects |
| What should we automate? | Tasks or labels for repeated work |
Design your setup. Paste this into Claude or ChatGPT:
I run a [12-person agency] with clients like [X, Y, Z]. Our goal is [billing accurately / knowing which clients make money]. Suggest a simple setup of clients, projects and labels for tracking our time.
Set it up and tag your time:
Look at my calendar and last week's work, and suggest the clients and projects I should set up.
What did I work on this morning, by client?
Anything you couldn't match to a client? Check my calendar to figure it out.
That [2pm] call was [Client A], not [Client B]. Fix it, and from now on tag anything with [clientdomain.com] to [Client A].
How many hours have gone into [Project] so far?
Already keep clients in a CRM like Attio or HubSpot? Ask your AI to pull them into Rize instead of typing them in.
Review
AI does the first pass. You stay in the loop. A review agent goes over your week looking for gaps, the wrong client, and entries with no project. Then you approve the fixes.
Review my tracked time for this week. Flag anything that looks off: gaps in my day, time tagged to the wrong client, entries with no project, and meetings that don't match the client on my calendar. For each one, suggest the fix and wait for my approval before changing anything.
The Friday review routine
Put the review on a schedule with Rize Routines, so it comes to you. Ask your AI:
Every Friday at 4pm, review my week, flag anything that looks off, and send me a project update with hours by client and project.
Or review in Rize: open Review Time Entries on your calendar and accept or reject each suggestion.
Routines can pull in context from your calendar, Slack or meeting recorder, so the review checks your time against what actually happened.
Improve
Measuring is the start. The value is in what you change, and in checking that it worked. Run this loop every month:
- Spot: find what took more time than it should.
- Decide: pick the fix, backed by the numbers.
- Change: make the change, or automate it.
- Check: did it work? Review again.
Prompts for each step:
What took more time than it should have this month?
What repetitive work took the most hours? Rank it by hours per month.
Give me two options to fix [Client], with the numbers.
Draft the weekly status report for [Client] from my tracked time.
Every Monday, send me any client over 80% of their hours.
Compare [Client] this week to last week. Did the change work?
What to improve next:
| Goal | Where to go |
|---|---|
| Billing and client reporting | This playbook |
| Team capacity | Workshop 1: How to Automatically See Where Your Team's Time Goes |
| Client profitability | Workshop 2: How to See Which Clients and Projects Actually Make Money |
| What to automate with AI | Nov 17: How to Identify Automation Opportunities With a Time Audit |
Prompt cheat sheet
| Step | Prompt |
|---|---|
| Capture | What did I work on this morning, by client? |
| Organize | Anything you couldn't match to a client? Check my calendar to figure it out. |
| Organize | From now on, tag anything with [clientdomain.com] to [Client]. |
| Review | Review my week and flag anything that looks off. |
| Review | Every Friday at 4pm, review my week and send me the project update. |
| Improve | What repetitive work took the most hours? Rank it by hours per month. |
| Improve | Every Monday, send me any client over 80% of their hours. |
Frequently asked questions
| Question | Answer |
|---|---|
| Do I need to start and stop timers? | No. Rize tracks your work automatically in the background, AI handles the first pass of tagging, and you review it in Claude or ChatGPT. |
| What should I track first? | Start with the question you want answered, then track the minimum that answers it. For most teams that means clients and projects. Add tasks, labels and contracts later, when you need them. |
| Does this approach work without Rize? | Yes. MCP is an open standard that connects your tools to your AI, so the approach works with or without Rize. Rize adds the automatic tracking, the tagging and the review routines. |
| How do I put the weekly review on a schedule? | Ask your AI to create a Rize Routine, for example: every Friday at 4pm, review my week, flag anything that looks off, and send me a project update with hours by client and project. |