Bruno.

Using Few Shot examples to boost LLM data extraction by over 50%? If you spent countless hours fine-tuning prompts, testing different parsing libraries, and trying to craft perfect solutions only to get mediocre results, this is for you.

🚀 Preventing Hallucinations in Vision-Based Data Extraction Vision models are coming up as the best way to deal with documents with complex layouts. On the flip side, they are more likely to hallucinate results. How can we address that? With OCR based data validations. 👇

The Takeaway Vision-based models are powerful, but validation frameworks are critical for reliable results. 💡 If you’re building data pipelines, combine extraction with validation to ensure accuracy and trust.

🚀 Preventing Hallucinations in Vision-Based Data Extraction Vision models are coming up as the best way to deal with documents with complex layouts. On the flip side, they are more likely to hallucinate results. How can we address that? With OCR based data validations. 👇

How It Works • Extract raw text using a PDF reader. • Validate each extracted value (e.g., “160.69 €”) by searching for it in the raw text. • Flag values that don’t match as potential hallucinations.

🚀 Preventing Hallucinations in Vision-Based Data Extraction Vision models are coming up as the best way to deal with documents with complex layouts. On the flip side, they are more likely to hallucinate results. How can we address that? With OCR based data validations. 👇

While vision models excel at "reading" layouts, they sometimes invent data. E.g., instead of extracting "2.983 kW" for contracted power, the model returned "2.0 kW"—a made-up value. 😬 How do we prevent this?

🚀 Preventing Hallucinations in Vision-Based Data Extraction Vision models are coming up as the best way to deal with documents with complex layouts. On the flip side, they are more likely to hallucinate results. How can we address that? With OCR based data validations. 👇

🚀 [NEW POST] Show, Don’t Tell: How Dynamic Examples Boosted Accuracy from 61% to 97% Ever spent hours fine-tuning prompts or testing document parsing libraries, only to end up with meh results? What if I told you that one simple change could drastically improve your results?

Check it out here: https://www.limai.io/blog/example

🚀 [NEW POST] Show, Don’t Tell: How Dynamic Examples Boosted Accuracy from 61% to 97% Ever spent hours fine-tuning prompts or testing document parsing libraries, only to end up with meh results? What if I told you that one simple change could drastically improve your results?

That’s when we tried something so simple it felt obvious in hindsight: we added an example. The results were staggering: • With a small model plus the example, accuracy leaped from 61% to 97%. • We achieved this without fine-tuning or complex parsing techniques.

🚀 [NEW POST] Show, Don’t Tell: How Dynamic Examples Boosted Accuracy from 61% to 97% Ever spent hours fine-tuning prompts or testing document parsing libraries, only to end up with meh results? What if I told you that one simple change could drastically improve your results?

Czech utility bills. These documents had: 🌎 Non-English text (a hurdle for many LLMs) 🧮 Values that needed to be calculated (e.g., summing multiple rows for Heating or Cooling) 🎲 A mix of other fields like dates, addresses, and contracts details

🚀 [NEW POST] Show, Don’t Tell: How Dynamic Examples Boosted Accuracy from 61% to 97% Ever spent hours fine-tuning prompts or testing document parsing libraries, only to end up with meh results? What if I told you that one simple change could drastically improve your results?

Interesting paper on Entity Matching using LLMs. I think I'll work on a demo of this soon.

🚀 [NEW POST] The Minimum Viable Model (MVM) Framework The most common mistake I saw in more than 10 years of building AI products? overcomplicating things. You can avoid that with these 5 steps and build AI products that actually work. Here's how 🧵👇

7/ Conclusion AI success doesn’t require complexity. By focusing on MVM, you’ll launch faster, gather feedback earlier, and build products that deliver value. Want to dive deeper into MVM? Check out my full article 👇 https://www.limai.io/blog/mvm

🚀 [NEW POST] The Minimum Viable Model (MVM) Framework The most common mistake I saw in more than 10 years of building AI products? overcomplicating things. You can avoid that with these 5 steps and build AI products that actually work. Here's how 🧵👇

5/ Step 4: Use real-world data Early deployment gives you invaluable feedback. ✅ Monitor real-world failures ✅ Feed use cases & data into your test set ✅ Continuously improve Your users will guide you toward what matters most.

🚀 [NEW POST] The Minimum Viable Model (MVM) Framework The most common mistake I saw in more than 10 years of building AI products? overcomplicating things. You can avoid that with these 5 steps and build AI products that actually work. Here's how 🧵👇

3/ Step 2: Build & evaluate a v0 Don’t overthink it—get your first data point. ✅ Create a simple test dataset ✅ Write basic code ✅ Define a metric Plot v0 performance on a graph (X = iterations, Y = performance). It’s all about starting small.

🚀 [NEW POST] The Minimum Viable Model (MVM) Framework The most common mistake I saw in more than 10 years of building AI products? overcomplicating things. You can avoid that with these 5 steps and build AI products that actually work. Here's how 🧵👇

1/ Why MVM? Too many AI products fail because teams chase perfection instead of value. The Minimum Viable Model framework focuses on launching a simple model, gathering real-world feedback, and improving from there. Here’s the 5-step framework 👇

🧠 Keeping up with research doesn’t have to be painful: LLMs to the rescue! If you’ve ever tried staying up to date with platforms like arXiv, you know how overwhelming it can be. We built a demo notebook to show how LLMs can simplify this process for you. 👇 More details:

At Limai, we’re passionate about making AI accessible to solve everyday challenges. From keeping up with research to scaling your business, AI has the power to transform how you work. Let’s connect to explore how AI can work for you!

🧠 Keeping up with research doesn’t have to be painful: LLMs to the rescue! If you’ve ever tried staying up to date with platforms like arXiv, you know how overwhelming it can be. We built a demo notebook to show how LLMs can simplify this process for you. 👇 More details:

How it works: 1️⃣ Parses arXiv RSS feeds for titles, abstracts, and URLs. 2️⃣ Asks an LLM to evaluate relevance, explain why it’s relevant, and summarize the main takeaway. 3️⃣ Emails you ONLY the most relevant results—no more wasted time!

🧠 Keeping up with research doesn’t have to be painful: LLMs to the rescue! If you’ve ever tried staying up to date with platforms like arXiv, you know how overwhelming it can be. We built a demo notebook to show how LLMs can simplify this process for you. 👇 More details: