Pinned Tweet
🚀 🚀 🚀
For 25 years, Salesforce, SAP and enterprise software were built around one assumption: humans operate the software.
We’re changing that.
Introducing ERP•AI and Proto: business software built for agents, and an agent that operates business software.
Describe the system you need. ERP•AI builds it.
Then Proto takes over. It manages records, runs workflows and executes actions inside the software you just created.
This is Headless SaaS.
1: get ERP•AI
2: build your SaaS
3: post below
Best one wins a MacBook. First 100 decent entries get $100.
(a message from Steve Papa / founding team)
Everyone in the replies says they could do what Scott Wu did.
So we made the game. Pick the cards that add up to the target before the clock runs out. 5 rounds. One prompt on Proto. Link Below
“BOM explosion” sounds like a disaster. Here's how to avoid it.
To build one conveyor you need 11 parts, 3 levels deep. This screen breaks it all down, checks every part against stock, and finds you’re 16 kg of welding wire and 164.8 kg of steel plate short.
Factories pay consultants months to build this. We built it with one prompt in Proto.
Canva just shipped a frontier design model. Here’s how to use it without opening Canva.
Proto has a built-in browser that works in the websites you’re signed in to.
Tell it “make 10 Instagram posts for our fall sale in Canva,” and it does the clicking while you do something else.
Introducing Canva Craft 🚀
Extraordinary image generation is now available to everyone, powered by our first frontier model built specifically for design.
With Craft, Canva is now the world's second most advanced AI lab for image generation on @DesignArena's global leaderboard.
See what it can do 👇
“Trust me bro” is not an audit trail.
Your AI agents are approving invoices, claims and contracts. When the auditor asks why, “the AI said so” won’t cut it.
OpenAuditor checks every decision against the actual records and gives you a verdict (true, false or uncertain) with the exact evidence behind it.
openauditor.ai
ERP.AI retweeted
>stop building business software
>we already built all of them
🔥 🔥 🔥
Stop building business software. We already built all of them.
CRM. Payroll. Procurement. Multi-currency accounting. Fixed assets. HR. Manufacturing. Transport. If your company does it, there’s an app for it.
Pick one, click “Get started,” and it’s live. Then tell Proto what to change and it fits it to how you work.
Stop building business software. We already built all of them.
CRM. Payroll. Procurement. Multi-currency accounting. Fixed assets. HR. Manufacturing. Transport. If your company does it, there’s an app for it.
Pick one, click “Get started,” and it’s live. Then tell Proto what to change and it fits it to how you work.
Proto Agent: Make your ERP xitter.cf/i/broadcasts/1qKDzWkRm…
#1 on @OpenRouter in three categories.
> #1 General Chat
> #1 Programming Apps
> #1 Native App Builders
> #12 app overall, daily
> 2.56 trillion tokens used
This week alone was close to a trillion tokens, more than double the week before.
We made GPT-6 Astra 29x cheaper without touching GPT-6 Astra.
A 0.6B model on a laptop answers first. It handled 490 of 500 queries itself, 97.2% accurate overall, in 161 ms. Only the 10 it wasn’t sure about went to Astra.
It’s called SmallDecide. Every number, dataset and checkpoint: smalldecide.com/evidence
“Find bad vulnerabilities in PHP” works when you have the harness.
Try that prompt in most frontier chat apps today and you’ll probably get a refusal.
On Proto, run it on DeepSeek, GLM or any open model, and spin up a swarm of agents across your own codebase. Find your bugs before someone else does.
Frontier models won’t touch your site’s security anymore. Open-source ones will.
Run DeepSeek and the best opensource models on Proto and point it at your own site.
Find the holes before someone else does.
Some stats on #php2shell. In the 10 hours before it was publicly announced, across the @Cloudflare network we saw thousands of attempts to exploit it. That jumped to tens of thousands in the 45 minutes post announcement, hundreds of thousands within 2 hours, and millions now.
DeepSeek’s just outscored GPT-6 sol on real business work
No trivia, no puzzles. The actual work businesses run on:
- reconciling spreadsheets
- building import files
- writing reports
- routine finance ops
Results:
- Proto + DeepSeek V4.1 Flash: 89.5%
- Proto + GPT-6 sol: 88.2%
- Codex + GPT-6 sol: 86.8%
Proto on a Flash model kept pace with GPT-6, and it was the fastest per task
everything is open: tasks, checks and every attempt. MIT licensed. come try to break it
businessbench.org
Tell your Proto agent how you run and bill projects, and it builds this for you.
Then it bills hours the week they're worked and flags the project going over budget.