praxic
An AI desktop agent being built to notice the workflows people repeat and turn them into automations employees approve.
- Role
- Co-Founder & CEO
- Dates
- 2026 – present
- Stage
- In development; leading customer discovery
- With
- Krish Punjabi, Co-Founder & CTO, building the observation agent and pattern pipeline
- Links
- trypraxic.com
Try it
praxicLoading demo…Sample app activity: which app, and when. Never messages, documents, or keystrokes.
- 09:05Slack
- 10:20Chrome
- 16:02Outlook (part of the repeated sequence)
- 16:05Excel (part of the repeated sequence)
- 16:09Finance tracker (part of the repeated sequence)
- 16:30Teams
- 16:41Outlook (part of the repeated sequence)
- 16:44Excel (part of the repeated sequence)
- 16:47Finance tracker (part of the repeated sequence)
1 / 5: Observe
The problem
Every company runs on someone who just knows how it works: the spreadsheet only one person understands, the three systems updated in a set order every Friday. None of it is written down, so it can’t be automated, and it leaves when they do.
My contribution
As Co-Founder & CEO, I lead customer discovery, interviewing small and mid-sized business owners and startup employees about how their work actually gets done.
Decisions
- Observe app sequences and timing only: never messages, documents, or keystrokes.
- Bring each repeated workflow to the person doing it, with the evidence, as a suggestion they can approve or decline.
- Run an approved automation once under supervision before it runs on its own, and log every run.
- Keep approved skills in a library the company owns, so a workflow several people repeat can become a shared team skill.
Current stage
In development. The demonstration on this page uses sample data to show the intended workflow; it is not the product and does not observe anything.
