Thank you all for being part of the launch yesterday, all the people I got to meet, all the experiences throughout the day, the pitches, the competition... that's all part of it, and I couldn't be happier
I also got some really nice insights about Toone from the community. A lot of people showing genuine interest in the tool and in what comes next.
I don't want to spoil too much, but we have a mobile application coming for iOS that will allow you to control Toone, dispatch and watch routines, and talk to an agent as if it were running on your machine. Apple just gotta approve.
We've also already implemented a sharing functionality for routines, allowing people to make their expertise and workflows available to the public.
Can I publish routines as anonymous? Don't know if I want them to be traced back to me haha.
@rajat_gupta51 Hi Rajat! That's a feature request I'll be passing along! Thanks for pointing it out & the feedback!
I also need this kind of service when I work in office!
But I'm also curious, when I made some agents to make them do marketing work for me(since I'm a marketer), where does the expertise come from that each agent has? Like I wanna make done my work with high quality and I want to be satisfied with it but if each agent has unguaranteed skills with expertise, I'd not happy with it.
Happy launch! Interesting product. How do you make sure Grokbot can handle deterministic outcomes?
@davitausberlin Hey Davit, thanks for the feedback and question! It's a mix of some strategies but mostly if a validation happens and the agent can't produce the result it fails the workflow. Thanks for hoping by!
Congrats on the launch! This sounds very promising but I'm having trouble picturing how someone in my industry (book publishing) might use it. A quick explanation would be appreciated. TIA
@eliezer_cernin Hey Eliezer, thanks for the support and kind words! 🙌 And answering your question: It really depends on how your work happens on the day to day. I run lots of automated content creation for various purposes (SEO, Youtube) and one pattern that often comes around is the research -> editorial review loop which basically inside Toone you would setup a set of routines and the editorial review would run with different agents performing reviews taking into account different axis (naturalness, voice matching, source provenance etc). Then we setup the loop and only allow it to pass after a score was achieved (you'd have a scorer step as well). Another interesting pattern that I've been using more frequently is to be after a set of runs from the routines, have another routine run just to grade how the routine is performing, polish some rough edges to make the entire flow better. That's my experience using toone! It's a life saver!
@matheus_paranhos1 Sounds worth exploring, thanks!
This is a cool idea. Can I change the model a workflow is using after I've created it?
@matthew_watson77 Thanks a lot Matt! Indeed you can, even at the more granular level (let's say you have a part in the workflow that runs in parallel, you can have them to run different models (OpenAI or Claude ones, mix and match)
@matheus_paranhos1 - are there metrics for tracking token usage inside Toone or does that depend on your LLM provider?
@matthew_watson77 On the LLM provider, but it's displayed inside Toone. And its system prompt is quite minimal
i always think about how to optimize my own workflow but sharing routines seems like a way to get fresh ideas. i might need to steal a few routines from others first before i get brave enough to share mine.
@therayess 😆 Indeed, and some routines can just be better in the way its crafted and polished, expert work is still valuable! Thanks a lot Ammar for the words
i've used workflow tools like Dify at work, and manually wiring up nodes is pretty tedious. just describing what you need and letting agent to build it is a lot simpler. Congrats on the launch!
@oliverliu98 Indeed, and one of the reasons I can't stand to use n8n or other tools. And having it all integrated in my systems and files makes everything more cohesive. Thanks for the remark!