Bring AI features to your Django project

Hello TechForwarders!

In this issue, I'm excited to share how Vinta has been applying our knowledge in AI across two innovative projects.

First, we'll explore how to integrate AI into your Django projects using our latest open-source library, the Django AI Assistant. Next, I'll share our recent experience collaborating with the United Nations to bring AI capabilities to their Social Protection platform.

We hope that these insights inspire you to explore the transformative potential of AI in your projects. Enjoy the read.

Calling All Django Developers: Try Django AI Assistant Now!

In the past few months, Large Language Models have matured, empowering autonomous AI agents that now can form the backbone of new features in existing products and even serve as the entire backend for new startups. Chances are you recently interacted with an LLM and didn't even notice, since it's possible to use them behind the scenes in any web application.

These models now support functions like Function/Tool Calling, Retrieval-Augmented Generation, and vector stores, enabling the creation of autonomous AI agents that can seamlessly integrate into your projects.

Python, with its mature ecosystem of AI libraries, is a prime candidate for these integrations. However, integrating Django with popular AI frameworks such as LangChain and LlamaIndex can be complex and time-consuming. That's where Django AI Assistant comes in.

Django AI Assistant, an open-source lib developed by Vinta, simplifies this process, allowing you to declare AI assistants as Python classes and quickly implement method tools the AI can use. Whether it's accessing the database, checking permissions, sending emails, or calling external APIs, Django AI Assistant enables you to build full applications with LLMs doing the heavy lifting.

For a practical demonstration, watch our tutorial:

Please give it a try. We're excited to hear any feedback on our GitHub Issues and Discussions. Let us know what you've built, too!

Vinta’s + United Nations = AI Product

Vinta recently teamed up with the United Nations Research Institute for Social Development to add some serious AI power to their product. They wanted to make it easier for people to engage over 15,000 resources they had in their dataset.

The project kicked off with a Product Discovery Sprint. Vinta's crew, which included a Product Manager, a Principal Product Designer, and a Lead Developer, teamed up with ONU's project leader. In two weeks, we were able to make a prototype of a user-friendly solution that was scalable and had some cool AI features built in to keep users engaged.

The AI-powered features designed included a summarization tool to help users quickly scan through search results, advanced filtering options, and taxonomic tagging to streamline content submission. We also proposed contextual reading lists, personalized recommendations, and automated categorization upon content submission to enhance the user experience.

Our approach involved a Retrieval-Augmented Generation (RAG) pipeline, leveraging zero-shot learning and vector databases to supercharge the LLM’s generative capabilities with proprietary data. This solution was designed to reduce development costs and time-to-market while enhancing user interactions with more natural and contextually aware responses.

Read the full United Nations case study for more information.

Thank you for joining this TechForward issue. We hope you found our insights on the Django AI Assistant and our collaboration with the United Nations both informative and inspiring.

Don’t forget to leave your feedback on our GitHub Issues and Discussions. We look forward to hearing about how you're leveraging AI in your projects.

Until next time,

Flávio Juvenal

Chief Technology Officer