Launching today

PHP Chatbot Assistant

Launching today

Self-hosted AI chatbots, free and MIT-licensed

0 followers

Self-hosted AI chatbots, free and MIT-licensed

0 followers

Visit website

Train it on your own documents and drop one script tag on your site. Answers come out of your content, the conversations stay in your PostgreSQL database, and nobody bills you per seat, per bot or per message.

Free

Launch tags:Productivity•Marketing•Artificial Intelligence

Launch Team

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

The biggest shift is in PageIndex, our vectorless retrieval mode where the model navigates a document's outline directly instead of matching embedded chunks. That only works well if the model can reason through a hierarchy, decide which sections actually matter, and backtrack if the first guess was wrong. With smaller chat models, this worked fine on short or simply structured docs, but fell apart on long manuals or policies with deep nesting. Astra's stronger multi-step reasoning makes outline navigation reliable on the messy, real-world documents our clients actually upload, not just clean examples. The second is AI lead capture. The bot has to naturally steer a conversation toward getting a name, email, and phone number without ever using a form or breaking the conversational flow, essentially exercising judgment about when and how to ask. That's a multi-turn task that requires holding onto intent across a conversation, and Astra's mid-turn steering lets it course-correct naturally

Report

4h ago

Maker

📌

I run a small tech consultancy, and at some point "can you add a chatbot to my site" turned from an occasional ask into basically every third call. Clients wanted the same thing: something that could answer from their own docs, grab a lead's info, and not make people hunt through a FAQ page or wait around for a human to reply. So naturally, I'd start by looking at the SaaS tools out there. And every single time, the pricing made my head hurt a little. Per seat, per conversation, per bot, whatever made sense for one client would fall apart the second I tried to use it for the next. And since I was white-labeling this for multiple clients at once, I needed real multi-tenancy, which most of these tools treat as some premium enterprise add-on. At some point I stopped shopping around and just built the thing myself. Auth, document ingestion, retrieval, the widget, lead capture, all of it. Then I did it again for the next client. And again. Somewhere around the third time, it hit me, I wasn't building three different chatbots. I was building one product and just repainting it each time. So I sat down, pulled everything that kept showing up across those projects into one codebase, actually wrote proper tests this time, and put it out under MIT. These days, when a client wants a chatbot, I'm not quoting two weeks of build time anymore. I clone the repo, run the installer, and it's basically done by lunch. If you're running an agency, freelancing, or just want a chatbot on your own site without a subscription hanging over your head forever, this is the thing I wish had existed when I started doing this.

Report

1d ago

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Report

4h ago

Maker

📌

Report

1d ago