Pelladio
Multi Model Collaboration
Few months back I had a cofounder on a different project. Things started fine, then they just stopped showing up, no big blowup, just slowly checked out. I kept doing the work because the project still existed, so I ended up doing both our jobs for a while and it set everything back by months. A bad cofounder turned out to be worse than no cofounder.
So this time I'm building alone. No funding, no team, just whatever hours I can find after everything else.
The actual thing: I got sick of having Claude in one tab, GPT in another, Gemini in a third, depending on what I needed that day. Every tool built for this switches you between models by comparing them, one prompt in, three answers back, pick your favorite. That's not the problem I had. I wanted to pull in the model I needed mid conversation without losing the thread or starting a new chat somewhere else.
Calling it Pelladio. Coordinate, don't compare, is basically the whole idea.
Waitlist's up at pelladio.com if anyone wants to poke at it early. Mostly curious whether this is a real problem for other people or if I've just been doing it wrong this whole time.
5 Comments
The “coordinate, don’t compare” distinction is interesting. The real test will be whether switching models mid-thread actually solves a meaningful workflow problem better than simply using one model well.
Yeah that's fair, and honestly for a lot of people one good model is probably enough. If you're mostly doing one kind of work, switching just adds friction.
Where it's actually helped me is stuff like the project plan example in the post, working through something with Claude, then hitting the part where I need real code and Claude isn't who I'd pick for that. Right now that means either accepting a worse answer or copy pasting my whole context into a new tab. Neither of those feels like "using one model well," it feels like working around a limit.
Might just be describing my own workflow though, not a universal problem. That's kind of why I'm asking instead of assuming.
If you want to try it yourself and see if it actually holds up, join the waitlist and I'll get you in before anyone else.
That makes the distinction much clearer. I’d be interested to hear what you’re seeing from users as they start acting on those checks. If you’re open to continuing the conversation, what’s the best email to reach you at?
Here's my Business Email
maana@torqondotdev
I'm unable to post links, replace dot with .
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.