Overview

Product video
Give document reviewers a useful starting point
Suggest two possible document types from English, Spanish or Portuguese business text, then let a person choose the next workflow. The seven categories are purchase order, delivery note, bank statement, utility bill, insurance policy, invoice and other business correspondence.
Use this after your own authorized text extraction, when a reviewer has already established that the input is one supported document. It can help organize intake before invoice extraction, order processing or receipt review. The result is two ranked suggestions, with caller IDs and explicit manual-selection flags.
The CPU runtime includes its weights and needs no model downloads or external services. It rejects over-budget text instead of silently truncating it.
This is an assisted suggestion tool. It does not perform OCR, verify authenticity, detect unknown or combined documents, extract invoice fields, or automatically route or decide anything. Out-of-scope text can still receive suggestions. Rank is not calibrated confidence, and whole-document coverage is unknown.
Training and project evaluation use fictional/model-authored examples. Generated diagnostics are disclosed in the model card; there are no native documents or independent human acceptance labels, and customer-document accuracy is unmeasured. Validate the behavior on your own authorized examples before relying on it.
The software is USD 0.50 per instance-hour on ml.c5.xlarge for real-time or Batch Transform, with a 14-day software trial. AWS infrastructure charges are separate, including during trial. Technical qualification and the complete SDK notebook passed in us-east-1. See the model card for the measured workload and generated-only evaluation limits. The current Marketplace offer is authoritative.
See the review workflow
These demonstration videos use actual frozen model calls with fictional examples in an illustrative UI. They demonstrate manual choice and rejection. They do not establish real customer-document accuracy or independent buyer checkout.
Highlights
- Two ranked suggestions among seven business document types; a person chooses or rejects them.
- English, Spanish and Portuguese text on CPU, with stable IDs and no silent truncation.
- Generated-only validation; customer accuracy is unmeasured. No OCR or unknown/mixed detection.
Details
Features and programs
Financing for AWS Marketplace purchases
Pricing
Free trial
Dimension | Description | Cost/host/hour |
|---|---|---|
ml.c5.xlarge Inference (Real-Time) Recommended | Model inference on the ml.c5.xlarge instance type, real-time mode | $0.50 |
ml.c5.xlarge Inference (Batch) Recommended | Model inference on the ml.c5.xlarge instance type, batch mode | $0.50 |
Vendor refund policy
If the software did not work as described, email info@softwaresushi.com within 30 days with your account ID and endpoint or job name. Software Sushi will refund the software charges for that use. AWS infrastructure charges are separate.
How can we make this page better?
Legal
Vendor terms and conditions
Content disclaimer
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Version release notes
Initial assisted release. Two suggestions for manual choice from extracted text; no unknown/mixed detector. Frozen E5 and original ridge head, bounded JSON/JSON Lines serving and disclosed generated-only evaluation.
Additional details
Inputs
- Summary
One JSON instances envelope containing extracted text and optional id/filename metadata.
- Limitations for input type
- 1-32 records, 256 KiB/envelope, 16,000 Unicode code points/text and 256 complete encoder tokens including prefix/special tokens. No truncation. The caller establishes a single document from one of seven supported types. No files, OCR, unknown/mixed validation or URL/path fetching. IDs must be unique.
- Input MIME type
- application/json, application/jsonlines
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
instances | Complete extracted-text records for manual type suggestion. | 1-32 objects with required text and optional bounded id/filename; reject an invalid batch atomically. | Yes |
Resources
Support
Vendor support
Email info@softwaresushi.com with the version, instance type and request ID. Do not include document contents, private information or credentials. No response-time SLA is advertised.
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products

