Blar

Graph-Powered Code Debugging with GenAI See how BLAR App's Code Base Agent uses graph technology to debug and optimize your code effortlessly! In this demo, we’ll sync repositories, address unexpected errors, and demonstrate how Blar allows you to improve your development workflow. Discover the power of graph technology in coding and how it can streamline your debugging process. Guest; José Domínguez, Blar Repo: https://lnkd.in/de2zU3BA Blar: https://blar.io/

Recently at Blar we've been integrating our solution to a diversity of AI-agents. The last of which was an agent that autonomously solves GitHub Issues (SWE-Agent). We later tested its performance. In this Article I wrote we take a deep dive in our results and analyze why Autonomous Software Engineers are Not Quite There …. Yet. Hope you like it. #Neo4j #OpenAI #LlamaIndex

New feature launch at Blar! During the last week, we talked with 30 developers to understand their major pain points around software development. Most of them told us they spent 80% of their day thinking about where to write the code and what implications their modifications would have. To help them solve their problem and save time, we are launching a new feature that leverages Blar's graph technology. It allows developers to read where they should modify the code and understand the impact it has on the repository. Blar uses a combination of GitHub pull requests and the code itself to understand where to suggest the modifications. The information gets automatically written inside the Jira issue for convenience and easy accessibility. Stay tuned!

Meet our Portfolio || Blar is a research-driven artificial intelligence company. The platform helps AI agents deeply understand code repositories by representing imports, exports, function definitions, and function calls as an interconnected graph structure that the AI agents can traverse. With Blar, the agents have all the required context to generate the desired business value outcomes (e.g., debugging, unit tests, refactoring, etc.). 📈 Transverse your code https://blar.io/ #CollideOnPurpose

We are excited to announce the launch of Blar open-source!   Blar helps AI agents deeply understand code repositories by representing imports, exports, function definitions, and function calls as an interconnected graph structure that the AI agents can traverse. With Blar, the agents have all the required context to generate the desired business outcomes (e.g., debugging, unit tests, refactoring, etc.). We included two AI agents as examples, one for debugging and a second for unit tests. However, the applications are endless! Let's explore them together 🚀 Available on GitHub today. https://lnkd.in/ec2dEZ88 We used tools from Neo4j, LlamaIndex, LangChain, and OpenAI to make it happen!

During the last few months, Blar has been working on data retrieval for AI agents. During this time, we have had several interactions with design partners, discovering how challenging and unexplored is data retrieval from code repositories. Some challenges are: - How much context does the agent need to perform the action? - How can we not cut functions while chunking? - How can we keep the context up to date? - How can an agent understand the extensive connections in a code repository? To solve our design partners' challenges, we made a graph builder that represents a code repository in detail. Also, we included an agent capable of traversing the graph. This way, our design partners can connect their AI agents and feed them with meaningful context about the code. To build this we used tools from LlamaIndex LangChain and Neo4j Let us know what you think! Stay tuned ;)

In our last post we talked about how graphs can help us improve our retrieval accuracy. The only problem is that we didn't back our claims with data (my bad :/ ). Many of you asked for this data so we created a benchmark to test our result. In this post we explain how we created the benchmark and the results we obtained.

I'm happy to share my last article on the evolution of information retrieval techniques, the limitations of vector semantic search, and Blar's approach that leverages graph technology. https://lnkd.in/ea4B8QAa Try Blar at: https://blar.io/

My experience with Patricio Daher, the co-founder and CEO of Blar, has been one of the most enjoyable adventures I have had in the world of LatAm startups. Blar uses AI to make databases more accessible to employees of an organization.  You don’t need to know SQL or be an expert on a business intelligence system to get key insights.  Their vision is much bigger and they plan to use AI to transform and optimize how companies build and use data.   But that is a story for another post 😀 I heard about Blar through many sources (Platanus - thank you Paula Enei and Joaquin Stephens , Natalia Gonzalez Vela, direct outreach from Blar, Sebastian Kreis at Xepelin, etc.) but was reluctant to meet them because I was not sure I could really help an AI team focused on a technical audience. My focus is fintech, proptech, and supply chain. However, given all the outreach from people I really respect, I decided to have a zoom call and do my best to help Patricio.  We worked on the pitch, I connected him with investors with more AI experience, got back channels on what these investors thought, and generally did my best to help where I could. Patricio was kind enough to introduce me to Miguel Paredes, PhD who within a few days of the initial intro visited me at my home.  I posted on LinkedIn about Miguel and in passing mentioned Patrcio and Blar.  I immediately got a WhatsApp asking for me to introduce some investors to Blar and provide my perspective and within a few short days/weeks, Blar had their initial institutional investor (thank you Collide Capital)! Since then Patricio came to San Francisco and my daughter (Sofia Montoya) helped him pick the neighborhood for his Airbnb.  And eventually we met in person at the Stanford University Golf course. I am always amazed at the power of a compounding network.   If you act in good faith and support many people, over time that network becomes very powerful.    One of the greatest parts of the Silicon Valley ethos is the idea of “paying it forward” and helping others without expectation of anything in return. I love doing my part to spread good karma and encourage the success of the startup community in Latin America. Congrats Patricio and the entire Blar team! José Domínguez, Benjamin Errazuriz, Elias Mufarech, Brian Hollins, Juan Pablo Silva E., Antonia Rojas Eing, Axel Christensen, Salvador Said, Benjamín Abu-Gosch, Francisco Javier Valdés Morales, Javier Pedrals, Eduardo Beltran Au, David Perez, José Tomás Valdivieso, Evelyn Von Bischhoffshausen, Markus Schreyer, Nicolas Chacon, Camilo Acosta

Big update alert! The time has come for every team in any company to unite on a single canvas. This update promises to reveal unseen insights and decode the narrative behind changing numbers. Stay tuned for this groundbreaking update! And guess what? We've got an attached video with a voice-over to walk you through it all! #InsightfulCanvas