Chatbot UI API Integration with AI Agent Platforms

Hey guys - I love Bolt and have been building away. My biggest sticking point right now is connecting my Bolt to my AI agent inside LangFlow. I’m using Datastax Langflow and I know their API is new, but I can’t for the life of me integrate the two. Even my Dev couldn’t get Bolt/supabase to do it. I’m going to try to host LangFlow and use their embeded chat next, if i can’t get this to work.

I’d love to see API integrations added for LangFlow & N8n the most. Perhaps the API inputs can be added in the settings? DataStax LangFlow and self hosted LangFlow have different APIs, just FYI.

@ColeMedin Have you used LangFlow at all? Datastax integrated them into Astra DB, which is owned by Nvidia. I’m wondering your thoughts on it.

2 Likes

I haven’t actually used Langflow myself but I’ve heard great things about it so I’ve been meaning to try it for a long time!

And API calls can sometimes get blocked by CORS in the bolt.new/bolt.diy web container which might be the problem you are running into. What is the specific error you are getting when you try to integrate with the Datastax Langflow API?

I’m surprised you haven’t used Langflow yet! I know you’re big on Streamlit - and Streamlit happens to be the only direct component integration that exists right now lol

I’m not a dev. My dev could probably figure it out if I gave him the code and gave him a few days. But I’m trying to get it to work directly with Bolt without needing his intervention.

The main error I was getting from bolt is an error in sending. So it’s on the bolt side. Bolt runs the auth through Supabase so it naturally wants to use Supabase API. And it doesn’t connect to the LangFlow API at all.

LangFlow is fairly new. Tbh so am I to AI Agents but it’s the first tool I found.

Whats interesting about LangFlow is that it’s open source, but has been picked up by DataStax who offer it free inside astra DB. DataStax is an Nvidia brand. And they’re just starting to put out more content.

Like I said they have community built Streamlit component (Listen and Send, on Langflow.store) so I gave it to Bolt to rework for itself which is beyond its scope. My next step is giving the plugins to Cursor and seeing what happens there. I’m ready to just redesign a Streamlit app and call it a day lol

But when it comes to Bolt and any no code dev, I see there’s a real lack of content out there for how to connect their chat functions to AI Agents frameworks like Langflow. I don’t even see content about integrating bolt with N8N other than a few video experiments out there. Not even how to send webhooks from Bolt.

IMO this is CRITICAL for biz owners who want to take these apps to market. We’ve got about 5 prototypes built but can’t seem to close the gap.

Would love to see a video from you about this. You’re great at explaining things.

But ya - go google “DataStax Langflow” and poke around in there. Curious what you find.

2 Likes

I didn’t actually know that about Langflow - both with the Streamlit integration and that it was picked up by DataStax. I appreciate you enlightening me!

Yeah given your description of the issue I do still think it is a CORS problem, the Bolt webcontainer itself blocking connections to some things. Isn’t always an issue but it especially is when you try to connect to things on the localhost like if you’re running Langflow locally.

Seems like a video is in order for simply how to hook into API endpoints in bolt?

1 Like

Yes indeed. And it looks like you put one out two hours ago! Amazing. Thank you.

And yes, DataStax has plans for Langflow. It’s basically a visual editor for Pydantic AI which you also did a video on.

Compared to n8n it’s an infant. N8n has been around for a long time and just went head first into AI agent components. It’s helped them a lot with exposure.

Langflow is maybe a year or two old? They just released v1.1.1 which was a major upgrade (felt like going from beta to production tbh)

I love it. It’s great for working with massive amounts of text and their vector RAG agent chat is excellent.

I’ve gotten to know it really well - but it lacks integrations due to its young age. Theres a huge community pull tho and people have been making a lot of new components recently. They’re all on Langflow.store.

I’d be happy to show you around it on zoom or something! LMK

1 Like

Sounds awesome! I definitely want to check it out sometime. I was even thinking of doing livestreams once and a while where I just build cool things with a bunch of tools like Langflow!