Introducing Forge

Drop your data.
Deploy.

Forge turns your unstructured documents into production workflows — extraction, validation, schema enforcement — immediately ready to run with a single API call. No glue code.

Start building free

No credit card required. SOC 2 Type 2 certified.

shipping-docs.pdf

AI got great at reading documents. But reading was never the hard part.

The hard part is everything after: the business logic, the exception handling, the schema enforcement, the production API. Most AI products demo well on a clean PDF. Then you try it with your actual data—bundled shipping packets, handwritten claims, invoices in three languages—and you spend the next three months building glue code in an attempt to normalize the data you truly need on the other end.

Insurance claims. Financial reconciliation. Shipping documentation. Healthcare billing. The operations that actually run your business are still manual—not because AI can’t read documents, but because nobody built the production layer between “we extracted your data” and “your operation is complete.” This gap is where every AI project starts to die.

SplitSplit Shipping Documents
Entry
TransformExtract Bill of Lading

bl_number, shipper, consignee, vessel_name, port_of_loading, port_of_discharge, container_details, issu…

TransformExtract Customs Declaration

declaration_number, declarant, total_customs_value, tax_items, liquidation_number, hs_code, origin…

TransformExtract Payment Receipt

receipt_number, payment_date, total_paid, payment_method, reference_code, currency, payer_name…

TransformExtract Release Order

release_number, authorization_date, authorized_by, cargo_description, container_id, port_code…

ValidateSchema Validation
98% confidence
DeployAPI Live
✓ Ready

curl https://api.bem.ai/v1/workflows/ship-docs/run

Forge closes that gap.

Describe the operation. Drop your actual files. Forge identifies every document type, assembles the complete pipeline, and hands you an API key.

How it works

01Describe your operation
Tell Forge what your process looks like. What comes in, what needs to happen, what the output should be. Forge understands your operation, not just your documents.
02Drop your data
Your invoices. Your claims packets. Your shipping bundles. Your contracts. The messy ones. The bundled ones. The ones in three languages with handwriting in the margins. Forge works with what you work with.
03Forge builds the workflow
Forge identifies every document type inside your files and assembles a complete workflow using bem’s production primitives: Split, Transform, Analyze, Enrich. Schemas generated. Extraction configured. Validation wired. Automatically.
04Test on your actual data
Forge runs the workflow on your files immediately. You see structured results, field by field, with a confidence score. Not a demo dataset. Your data, your documents, your edge cases.
05Deploy
API key. Code snippets in cURL, Python, Node.js. Webhook subscriptions. You’re not prototyping. You’re deploying.
shipping-packet.pdf

For the work that’s too important to get wrong.

Not an agent.

Transparent pipelines, not black boxes.

Agents guess. They explore. They figure it out as they go. That’s great for browsing the web. It’s not how you process insurance claims, reconcile invoices, or run compliance workflows. Forge builds transparent, deterministic pipelines.

Trainable.

Self-healing accuracy loops.

Every function in every workflow is individually trainable. Your team makes corrections, and the system learns. Self-healing loops catch accuracy drift before it reaches your customers. The more you use it, the better it gets.

Deterministic.

Schema-valid JSON. Every time.

If bem can’t map data to your requirements with confidence, it flags the exception. It never guesses. Accuracy isn’t a feature. It’s the architecture.

Auditable.

Full visibility into every step.

See exactly what happened to every document. Every decision, every data transformation. Built for environments where “it usually works” isn’t good enough.

From shipping packets to powering your operations. In minutes.

One PDF. A bundled international shipping packet: Bill of Lading, Customs Declaration, Payment Receipt, and Release Order, all in one file.

Document Classification

Forge identified four document types inside the bundle and split them automatically.

Bill of LadingCustoms DeclarationPayment ReceiptRelease Order

Field Extraction

Every field from every document, extracted in parallel with schema enforcement.

Shipper, consignee, vessel, ports of loading and dischargeBL number, container number, seal number, gross weightCustoms declaration number, tax breakdown, declarantPayment receipt, total paid, release authorization

Validation & Confidence

Schema validation passed. Cross-document references verified. Confidence scored across all fields.

98%confidence

Try this yourself → Drop a shipping document and watch Forge build it.

Trusted by teams that can’t afford errors.

We have not experienced any hiccups so far. The process has exceeded expectations as far as what we thought was going to happen.
Chuck DrewsGroup Product Manager, Fleetio
65%
Faster processing time
100%
Accurate on totals
Millions
Documents processed weekly
Fortune 50 companies run production workloads through bem daily across their most regulated operations.

Any input. One output.

From invoices to video footage, Forge handles every kind of unstructured data your operation produces.

Invoices & AP

Vendor info, line items, totals, GL codes. Match against POs automatically.

Insurance Claims

Claimant details, procedure codes, amounts, supporting docs. Ready for adjudication.

Shipping & Logistics

Bills of Lading, customs forms, packing lists. Split bundled documents. Multi-language.

Healthcare

Medical records, lab results, prescriptions, clinical documentation. HIPAA-compliant.

Financial Data

10-K filings, earnings calls, market data. Structured datasets for analysis.

Video & Audio

Dashcam footage, recorded calls, voice memos. Not just documents. Any unstructured data.

The infrastructure behind the magic.

01

State-of-the-art models, orchestrated

bem routes across the latest vision, language, and embedding models—selected automatically for each function based on your data. You don’t pick the model. bem picks the best one and keeps it current.

02

Evaluations on everything

Every test runs accuracy evaluations. Golden datasets. F1 scores. Regression testing. Drift detection. Accuracy is treated like code coverage: measured, monitored, and enforced.

03

Production throughput

Dedicated inference infrastructure. Fast, reliable, built for batch and real-time. Processing millions of documents daily across the most demanding environments.

04

API-first

RESTful endpoints for ingestion. Polling and webhooks for results. Idempotent re-runs for safe reprocessing. Every function versioned with instant rollback.

Enterprise-grade trust and compliance.

SOC 2 Type 2

Certified. Audited annually by independent third parties.

HIPAA Compliant

BAA available. PHI handled with full regulatory compliance.

EU Data Sovereignty

Full in-region processing. Data never leaves your jurisdiction.

99.99% SLA

Enterprise SLA available with dedicated support.

Zero Retention

Options for zero data retention after processing completes.

Flexible Deployment

Multi-tenant cloud, Private Link, Dedicated VPC, or on-prem.

The work that’s too important to get wrong deserves automation that actually works.

Sign up free. Describe your operation. Drop your data. See it deploy in minutes. No sales call. No credit card.

Forge by bem | Turn documents into production workflows | bem