Just tried it. I think this has a lot of potential and I'm planning to revisit in a few months. Right now I'm running into issues with the orchestrator itself, both bugs and difficulty adapting it to my use case. I found myself spending more time fighting the framework than building my actual agent.
Oh, sorry more out of interest from a business perspective.
Back in the day a toolkit like this would have either been an OSS + services and support play (like spring.io) or an "enterprise alternative" like Weblogic was to Apache.
... I'm wondering how this sort of play scales today?
Oh yes ofc! We are adopting the same OSS + services model, we currently have our Dexto cloud platform in alpha and are onboarding projects on a case-by-case basis. We plan to launch the self-serve version shortly.
Our other product available at www.trytruffle.ai supports creating and deploying stateless agents which are great for simpler use cases.
We plan to launch a similar cloud offering for dexto but with deeper support for long-running agents that handle more complex tasks and integrate more deeply across orgs and apps with support for on-prem setups.
Some of the use cases we showcase are RAG agents, Youtube or any URL summarizer, imagegen tools, email summarizers, etc.
Any use case where you might want to run an LLM+tool in a loop for a multi-step but single-trigger task - but without persistence. It works similar to the ai-sdk but you get the entire agent as a hosted, shareable endpoint.
Just tried it. I think this has a lot of potential and I'm planning to revisit in a few months. Right now I'm running into issues with the orchestrator itself, both bugs and difficulty adapting it to my use case. I found myself spending more time fighting the framework than building my actual agent.
Hey! I'm one of the founders, what issues did you run into? Any feedback here would be very helpful, thanks!
does anyone have a Mumbai-based SaaS orchestrator for my orchestrators?
What's your pricing model?
Its currently open for self-hosting but you can reach out to us if you need help deploying it on your own infra and we can help with a custom setup
Oh, sorry more out of interest from a business perspective.
Back in the day a toolkit like this would have either been an OSS + services and support play (like spring.io) or an "enterprise alternative" like Weblogic was to Apache.
... I'm wondering how this sort of play scales today?
Oh yes ofc! We are adopting the same OSS + services model, we currently have our Dexto cloud platform in alpha and are onboarding projects on a case-by-case basis. We plan to launch the self-serve version shortly.
Our other product available at www.trytruffle.ai supports creating and deploying stateless agents which are great for simpler use cases.
We plan to launch a similar cloud offering for dexto but with deeper support for long-running agents that handle more complex tasks and integrate more deeply across orgs and apps with support for on-prem setups.
Awesome. Can you explain what you mean by simpler use cases? e.g. one shot classification?
Some of the use cases we showcase are RAG agents, Youtube or any URL summarizer, imagegen tools, email summarizers, etc.
Any use case where you might want to run an LLM+tool in a loop for a multi-step but single-trigger task - but without persistence. It works similar to the ai-sdk but you get the entire agent as a hosted, shareable endpoint.
From the site: "Join developers building intelligent applications with Dexto. Open source, local-first, and ready for production."
Note that this code is licensed under "Elastic License 2.0 (ELv2)", so not open source according to OSI.