Tell your Proto agent how your offices bill each other, and it builds this for you.
Then it finds the difference and drafts the fix for your finance team to approve.
The report uses recall visits explicitly linked to the original call. Keep that link in the record when a return visit is a recall.
https://t.co/ABItuIiVtl
@protosphinx
A repair finished yesterday has had less follow-up time than one finished last month.
Our Field Service Management docs describe a 30-day window for first-time-fix, with newer calls shown separately.
Going Live with Proto
Autonomous Business:
1. Build your ERP + Systems
2. Build a website + CRM
3. Build Business data
4. Build workflows & rules (eg approval on purchase orders)
5. Run your Business
Start free.
@arpantwts the US had 274 automobile manufacturers operating simultaneously in 1909 - just a few yrs after the invention of automobile
there were celebrities too back then...same as today. Edison was a celebrity too tbh if you read what was going on back then...
Also releasing: Business Bench for teams building AI agents that handle business work.
187 desk tasks across spreadsheets, imports, reports and finance. Public tasks and automatic checks. Success means passing every check.
https://t.co/lOYtbKjW87
YOUR NEXT USER WILL NEVER SEE YOUR DASHBOARD
if agents do most of the work, they don't care about the ui. they care about the data, the permissions and the actions
https://t.co/jTHJgefalz is built for that:
→ proto builds the app from a description → then runs it through the api, no clicking needed → the screen is just where humans check and approve
the demo that matters won't be the dashboard anymore
try it on one process you know: https://t.co/jTHJgefalz
If a gauge is found out of tolerance, its earlier inspection results may need review.
Our QMS page describes flagging inspections that referenced it on or after its last calibration date for Quality Manager review.
tldr: potentially CRISPR for neural networks.
turns out neural nets might already be much more interpretable than we thought. you can identify specific neurons and circuits, understand what they do, and potentially intervene on them.
I GAVE AN AI AGENT ONE PROMPT AND IT BUILT ME AN INVOICE APP.
It's called Proto, from https://t.co/SvBlzcGNjF. Launching today.
It wasn't perfect:
→ It created 15 invoices instead of my 5
→ It noticed on its own
→ It fixed the records before calling it done
Then I gave it a real task:
"Find unpaid overdue invoices above $25,000 and draft the reminders."
It found exactly one. Atlas.
Beacon, under the threshold.
Cedar, not due yet.
Delta, already paid.
Ember, exactly $25,000. Not above.
The reminder was saved as a draft, waiting for my approval.
Nothing sent. Nothing paid.
Build: 9 minutes. Task: 2 minutes. Video sped up.
Pick one process you know. Build it. Judge it yourself.
https://t.co/KWFRehxyAD
Save this, you'll want it at month end.
Proto is now the #2 productivity agent on @OpenRouter
486B tokens today
We think it’s happening because Proto is built on one belief: you shouldn’t be locked into a single model.
A harness tied to one provider bets your whole business on one lab’s roadmap and pricing. Proto runs on any model, so you always get the best one for the job and never pay a lock-in tax to OpenAI or Anthropic.
An agent inside Proto isn’t a loose cannon with a god-mode key that might do anything random. It works with the permissions of a real enterprise user.
It can see only what its role can see, do only what its role can do, and every action goes into the same audit trail as a human’s.
That’s the difference between an AI demo and an agent a company can actually hand its operations to. Any model, real permissions, full accountability. That’s Proto.
I had 0.1% idea of https://t.co/6tIeaKonai last year when i met sphinx.
This launch video just blew my mind! Running a manuf business of a special hardware product means i need custom ERP but i cant buy SAP.
This will change things as hardware companies shroom up everywhere