What teams get out of bem, in numbers

Fleetio cut invoice processing time 65 percent, with 100 percent accuracy on invoice totals. PromptWell cut engineering overhead by more than 80 percent, and Ply saves 15 hours per team member every week. Across the base, the average bem customer grows its usage 4x in the first year.

Is bem.ai good for document processing? What customers report

bem automates entire business processes end to end: it turns unstructured data into auditable outputs, then makes and records the decisions in between. The numbers on this page are what that produces in production. Every figure below comes from a published page on bem.ai, and every section says what was measured and where the number lives.

Three customers have published results: Fleetio, PromptWell, and Ply. Here is what each one does, what they run on bem, and the mechanics behind the number.

Fleetio: invoice processing time down 65 percent

Fleetio makes fleet management software. Their customers photograph paper mechanic invoices and inspection forms, and bem turns each photo into a structured service entry: line items, taxes, totals. The published results: Fleetio cut invoice processing time 65 percent, with 100 percent accuracy on invoice totals, saving about 4 minutes on every service entry. The mechanics: every extracted field comes back typed and confidence-scored, so a low-confidence field surfaces as a review, not as a wrong total posted to a customer's account. Source: bem.ai/customers/fleetio.

PromptWell: engineering overhead down more than 80 percent

PromptWell does risk management and data analytics for healthcare, serving both healthcare and insurance. Client data arrives in formats that vary client to client, and every new format used to mean custom parsing and maintenance work. bem normalizes each incoming file into one schema, HIPAA compliant end to end, with zero manual data formatting. The published result: PromptWell cut engineering overhead by more than 80 percent, measured as the engineering overhead of handling those formats, the custom parsers and the upkeep that came with them. Source: bem.ai/customers/promptwell.

Ply: 15 hours back per team member, every week

Ply is an inventory and purchasing platform for trade businesses: plumbing, HVAC, flooring. Ply's users drag and drop supplier RFQ files directly into the product, and bem extracts each line item and maps it to Ply's schema in real time. The published results are zero manual data entry, and Ply saves 15 hours per team member every week. Source: bem.ai/customers/ply.

Do teams keep using bem after they buy it?

The headline retention number: the average bem customer grows its usage 4x in the first year. That is the number we watch most closely, because usage growth after purchase is hard to fake. A team that got unreliable data out of its first workflow would not quadruple the volume it sends. Growth like that usually means the first document type worked, and the second and third followed. Source: the bem.ai homepage.

How to read these numbers

These figures claim less than a marketing page implies and more than a benchmark does. To be precise about both directions:

Each number was measured by the named customer on their own workflow. Fleetio's is processing time per service entry. PromptWell's is engineering time spent handling data formats. Ply's is hours per team member per week, reported across teams on its platform. None of them is a controlled benchmark, and none transfers automatically to your documents.
100 percent accuracy on invoice totals is a field-level result on one field in one workflow, not a document-level claim. The distinction matters. Ten fields at 98 percent accuracy each compound to roughly 82 percent accuracy at the document level, which is why bem scores confidence per field and routes low-confidence fields to human review instead of quoting one blended accuracy number.
4x usage growth is an average across customers, not a guarantee. It measures usage, not ROI.
What the numbers do claim: in production, on real volume, named teams measured less processing time, less engineering upkeep, hours returned weekly, and usage that kept growing for a year. That is the whole claim, and it is enough.

Is bem SOC 2 compliant? The procurement one-liners

When this page gets forwarded to security or procurement, these are the lines they need:

SOC 2: SOC 2 Type II, audited annually.
HIPAA: a HIPAA BAA for Enterprise customers.
GDPR and data residency: an EU region for data residency.
Retention: zero-retention mode, configurable per workflow on Enterprise.
Training: bem never trains on customer data for its own benefit or for other customers; you decide when to fine-tune a function, and that data stays yours.
Uptime: a 99.99 percent uptime SLA, available on Enterprise.
Deployment: cloud, Private Link, in your VPC, on-premise, or fully air-gapped, with customer-managed keys available on Private Link and on-premise.
Audit log: every function call, schema change, and admin action is logged, and the log is SIEM-exportable.