For the past two years, we’ve been obsessed with a single mission: turning the messiest unstructured data into clean, reliable, and production-ready outputs. We started with text, helping companies like Fleetio and Ascend to automate mission-critical workflows by structuring the chaos of their documents.

‍

But we always knew that the most valuable data isn't just trapped in text. It's in the real world—in photographs of damaged equipment, in charts from financial reports, and in the complex visual information that drives your business.

‍

Today, we’re taking a massive step forward. We’re thrilled to introduce Analyze, a new core function in the bem platform that lets you embed visual understanding directly into your product.

‍

From Document AI to Visual AI Infrastructure

‍

Traditional AI tools stop at OCR. They turn a document into text and call it a day. But what about a photo of a damaged car for an insurance claim? A picture of a supermarket shelf for a CPG brand? A chart in a financial report?

‍

These are not text problems; they are understanding problems.

‍

Analyze is our answer. It's not a separate product; it's a new, composable primitive that works alongside our existing Transform, Split, and Route functions. Now, you can build production applications that understand both the text and the visual context of your data. It’s full trainable, auditable, and evals run automatically on bem. You get the full suite of confidence scoring that you already get with our document understanding primitives.

‍

You can now give your product the superpower to answer questions like:

• "What’s the product being displayed?"
• "What competitor brands are visible on this store shelf?"
• "Is there visible damage to the front bumper in this photo?"

‍

Analyze turns these visual questions into clean, structured JSON, ready to power your mission-critical workflows.

‍

‍

Built for Your Product

‍

We built Analyze for builders. It’s fully configurable via our simple, schema-driven UI and accessible through our API. Just like Transform, it’s trainable, auditable, and built for the high-accuracy demands of production environments.

‍

This is more than a feature launch. It's an invitation to build the next generation of intelligent products. Ready to see what's possible? Dive into our new use cases below or sign up for free to try Analyze today.

‍

Some recipes to try with Analyze

‍

• Use Case 1: The Intelligent Insurance Product
  • The Problem: Insurance apps need to process photos of vehicle damage to triage claims. This is a slow, manual process requiring human review.
  • The bem Solution: A customer can upload a photo of their damaged car directly in your app. Analyze determines the location and severity of the damage and outputs a structured JSON object that automatically kicks off the claims workflow, providing an instant, magical user experience.
• Use Case 2: The CPG Retail Intelligence App
  • The Problem: CPG brands send reps into stores to manually check if their products are displayed correctly on shelves and endcaps.
  • The bem Solution: A field rep takes a photo of the supermarket endcap using their mobile app. Analyze identifies all visible brands, counts the facings for the brand vs. competitors, and flags if promotional materials are missing, providing real-time competitive intelligence.
• Use Case 3: The Automated Financial Research Platform
  • The Problem: Financial analysts need to extract key data points from charts and tables within investor reports and pitch decks.
  • The bem Solution: An analyst uploads a PDF report to their research platform. Transform extracts the text, and Analyze extracts the data from the charts, providing a complete, structured summary of the entire document that can power downstream financial models.

‍