Show HN: Era – Complete Simulated Companies for Your Agents
console.era.eon.ioHi, I'm Benji. I'm one of the people who built Era a free-for-builders service that creates digital twins for common SaaS services.
We built it for ourselves and got a lot of excitement from research partners - and are giving it away for free.
Hey, I worked on this - ping me if you have any questions this was a really cool experience
Had a chance to try it early, really cool idea and really useful in practice!
Super cool, a digital twin for Abdul company, accurate digital twins for 10s of services
Used an early version of this. Really cool
Thanks a bunch! Thanks a lot for your valuable feedback in the alpha!
How about digital twin of eon.io :)
We have one there! We actually started without it but found it useful for our own development and testing so we added it.
Cool stuff
I've just been setting up some Odoo instances for some side projects that I'm involved with and it got me thinking about whether any of these automated business attempts have used an open source ERP like Odoo as a backend for their endeavours.
It seems to me that it's a good foundation for this kind of experiment and doesn't involve too much reinventing the wheel.
Just let the LLMs do what they're best at and keep the business records inside an established and tested system instead of reproducing one ad-hoc.
We're a large backup vendor so honestly I've had the budget, expertise and time (from backup ventures unrelated to Era which is a community project) to build our own digital twins from scratch that e.g. run SOQL and implement the full API.
Since we have backup offerings at Eon we have the excuse to build really good digital twins for our software factory and the sketching algorithms and data engineering/science experts to make the data realistic.
This was definitely more "reinventing the wheel" than reusing a projects - Era is optimized for reads and a lot of data, we stream/read everything from a datalake so state is mostly immutable and we don't have to worry about e.g. collecting personal user information or scale too much.
The hard part here is obviously the very long and expensive estate generation for companies rather than the digital twins (that are interesting and million of lines of code but less exciting to the scientist in me. If the rsearch is interesting lmk and I'll send the papers over.
Ah yes, this worked so well in professional testing of consumer A.I.s.
I’ll certainly trust this one wholeheartedly.
Hit me up: What concerns do you have and how can I convince you the data is good? I'm obviously biased but I'm personally very surprised by how well it works and how much the sketching/metadata from exabytes of real backups contributes.
We have a lot of interesting results from research partners but most of it is unfortunately still under embargo. If you're in SF you're welcome to say hi (hit me up, my email is user at gmail .com )