Dan Foley’s Post

Gearing up for your Gen AI Journey? ✨🐼 Check out this brief demo on LLM capabilities within Snowflake Cortex. Sumanth Donthula, Associate Consultant Gen AI, showcases how a set of loan data can be implemented into a Chat Bot that can allow less technical personas to interact with the data. Along with the chat feature there is a summary of stats and visualization component as well. Putting it all together, these features are ready to set you up on a Gen AI roadmap! #snowflake #cortex #genai #chatbot #roadmap

Cracking the code to AI magic with Genix AI! 🌟 Whether you're navigating the maze of data analysis or summoning your inner tech wizard for AI integration, we've got your back! Pro tip: whisper sweet binary codes, and watch your projects transform! 🤓💻 #TechMagic #AISorcery #GenixAI #TechTips #DataWizards 🧙♂️✨

Here is another video showing the power of Tines and its "Automatic Mode." In this quick demo, I walk you through a powerful example of using AI-driven automatic transforms to handle complex data requests in real-time. Recently, during a customer demo, I was asked to demonstrate the capabilities of our AI transform tool with a specific payload. Watch as I showcase how effortlessly we can pull data, execute HTTP requests, and group it by country to display information on specific mentions—such as victims by location. #AITools #DataTransformation #AutomaticTransform #DataAnalysis #AIDemo #RealTimeData #buildeverydaywithtines

CuriousCirkits E27 - Glean - in conversation with Jay Fielding - Glean - The enterprise AI platform for all your company’s data. Follow - CuriousCirkits #genai #chatgpt #chatbot #enterprise

We are super excited to partner with Liucija, Senior Data Scientist on the AI team at Hostinger International, as they work towards leveraging AI for use cases like customer support, website building, and more. If you'd like to learn how Braintrust Data helped Hostinger: - 3x the number of AI features they're working on - Automate over 40% of customer support chat conversations Check out the video and blog post below!

I wrote about what’s important in LLM evaluation, and now I want to share what I'm doing on the HOW part. There’s a number of tools out there already to aid you with LLM evaluation, but I’ve decided to try out Braintrust Data. They make it super easy to evaluate your LLM outputs, using predefined or custom metrics and make data management easy. My highlights so far: 📚 Creating and growing Datasets - this is your “Golden Examples” you want to evaluate against. I’ve gotten product teams to add data they want to test against using the UI directly. 🚀 Validating before shipping: running an experiment on a predefined Dataset before launch gives much more confidence. 🔃 Running experiments when developing and seamlessly comparing performance, speeding up iteration. 💬 Logging from live applications into Braintrust and using metrics + Human review to find problems. Then, right from the logs, create a Dataset for evaluating in Iteration No 2. You can also watch this short demo on how our LLM evaluation framework looks like 🙂

Day 45 of reaching $100K MRR with Nex AI I've been dead silent the past few days.... Nex AI is finally going vertical after numerous feedback! Nex now enables companies to swiftly extract key insights from emails in real-time. We specialize in handling complex, unstructured email data, providing structured metadata that reveals the complete context. Whether you need to identify part codes, SKUs, or product IDs, Nex can provide accurate and actionable insights. This is especially useful for industries where emails' specific information is very difficult to read due to it being in a drawing, sketch, financial statement, or even in the main text body of the email. It's a pretty big pivot considering we have to discount our 7-figure pipeline to nearly $0, but we believe it is definitely the right direction. Would love to get anyone's feedback, connect with me and send me a DM!

Trellis AI (YC W24) is building a swarm of agents to extract and classify thousands of PDFs for you, all at once. Today, they're launching Trellis AI Studio. Engineers and operations teams can now quickly define complex document workflows through a spreadsheet-like interface and scale them across all their tasks in minutes. Critical enterprise data is often stuck in documents, emails, and phone calls. Previously, customers spent months on manual data entry or building their own internal AI solutions—efforts that are challenging to maintain, especially with edge cases and the reliability of LLMs. Trellis offers a robust building block to transform all that data into any system of record, with built-in validations and references. Congrats on the launch, Mac Klinkachorn and Jacky L.! https://lnkd.in/gAtDD5RX

Ellipsis (YC W24) knows what your team shipped last week, do you? We track summaries of code changes in order to power our AI code reviews. You can access those summaries on request (or scheduled, like a standup) using our Slackbot. Here's how that works:

Today in my experiment to rewrite a production codebase with ai help focused on migrating one class and its tests. Success! Went into a spiral avoiding an external service request in test. Carrying over mocking did not work and ai assistants got stuck repeating three approaches - none of which worked. The generalized human intelligence (me) changed the methods to create dependency injection and the bots fell in line. Green tests with 1:1 function coverage.