Introducing Forge: Drop your data. Deploy.

For three years, the AI industry has been obsessed with extraction. Turning documents into data. And it worked. Extraction got really good.

But extraction was never the hard part.

The hard part is everything that comes after. The business logic. The exception handling. The human review loops. The schema enforcement. The production API. The monitoring. The drift detection. That's where every AI project stalls. That's where every proof of concept dies. That's where engineering teams spend months building glue code around a JSON output.

An insurance adjuster opens a claims packet (medical receipts, EOBs, handwritten forms) and starts typing. A logistics coordinator receives a shipping bundle from a new port and spends an hour matching bills of lading against customs declarations. A healthcare administrator cross-references clinical documentation against procedure codes, line by line, because the alternative is a rejected claim and a patient who doesn't get care.

These aren't back-office tasks. They're the operations that run your business. And the reason they're still manual isn't that AI can't read documents. It's that nobody built the production layer between "we extracted your data" and "your operation runs."

That changes today.

Forge

Forge is a new experience in bem that closes the gap between data extraction and a running operation.

Describe your operation. Drop your files (the actual documents your team processes every day, the messy ones, the bundled ones, the ones in three languages with handwriting in the margins). Forge analyzes the data, identifies every document type, and assembles a complete production workflow: splitting, extraction, validation, schema enforcement. It tests the workflow on your files immediately, shows you structured results with a confidence score, and gives you a production API endpoint. One click.

You don't get structured data and a "good luck." You get a running operation.

This isn't a prototype. Forge builds on the same infrastructure that processes millions of documents daily for companies from seed stage to the Fortune 50, across insurance, finance, supply chain, and healthcare.

Watch It Work

Here's a real example. A logistics company drops a single PDF: a bundled international shipping packet containing a Bill of Lading, Customs Declaration, Payment Receipt, and Release Order.

Forge identifies all four document types inside the bundle. It builds a workflow: split the packet, then run parallel extraction on each document with its own schema and field set. It executes a live test on the spot. Minutes later: shipper, BL number, consignee, vessel name, ports of loading and discharge, container numbers, seal numbers, gross weight, customs declarations, tax breakdowns, payment amounts, release authorization. All structured, all validated, at 98% confidence.

One click: API key, cURL, Python, Node.js, webhook subscriptions. Deployed.

That workflow now handles every shipping packet of that type, automatically. And every correction your team makes trains the system, so accuracy compounds over time.

For the Work That's Too Important to Get Wrong

There's a reason we didn't build Forge as a chatbot that generates code, or an agent that figures it out on the fly. When a wrong number means a rejected insurance claim, a missed shipment, or a compliance violation, you need something different.

Every workflow Forge creates runs on bem's deterministic primitives. Each function (Split, Transform, Analyze, Enrich) is individually trainable, observable, and auditable. You see exactly what each step does. You see exactly what data flows through it. If bem can't map data to your schema with confidence, it flags the exception. It never guesses.

Under the hood, bem routes across state-of-the-art vision, language, and embedding models, selected and orchestrated automatically for each function. Evaluations run on every test. Self-healing loops catch drift before it reaches your customers. All of it running on dedicated inference infrastructure built for production throughput, whether you're processing ten documents or ten million.

This is what operations automation looks like when it's built for the real world. Transparent. Trainable. Deterministic.

The Results Speak

Fleetio, a fleet management platform, processes millions of documents weekly through bem. Before: their customers spent 6.5 minutes manually creating each service entry from paper invoices. After: 2 minutes. A 65% reduction in processing time with 100% accuracy on totals, including line items.

"We have not experienced any hiccups so far. The process has exceeded expectations as far as what we thought was going to happen." - Chuck Drews, Group Product Manager at Fleetio

Fleetio isn't alone. A Fortune 50 company runs production workloads through bem daily across their most regulated operations. Teams across insurance, logistics, and healthcare process millions of documents through the same infrastructure Forge now makes accessible to everyone.

Now Available in the EU

Starting today, bem is available in the EU with full data sovereignty. Data processed and stored in-region. Combined with SOC 2 Type 2 certification, HIPAA compliance, and zero-retention options, bem is built for regulated environments where accuracy and data governance aren't negotiable.

Global operations deserve global infrastructure. Now they have it.

Every Kind of Unstructured Data

Forge launches with document workflows, but bem's infrastructure handles far more than paper. Video. Images. Audio. Multimodal data. Dashcam footage for incident reports. Earnings calls for financial analysis. Medical imaging. Handwritten forms in any language.

If your business runs on unstructured data (and every business does), this is where you start.