Listing Thumbnail

    Tenet PII Engine - In-VPC PHI/PII De-identification

     Info
    Sold by: Tenet Labs 
    Deployed on AWS
    Detect, de-identify and reversibly rehydrate PHI and PII in LLM prompts, documents and tool outputs, entirely inside your AWS account. Data, tokens and keys never leave your VPC; vault keys are wrapped by your own AWS KMS key; every rehydration is audited to your Amazon S3 bucket with Object Lock.

    Overview

    Tenet PII Engine lets you use large language models on sensitive data without sending that data to the model. It detects and de-identifies protected health information (PHI) and personally identifiable information (PII) in prompts, documents, tool outputs and logs, replaces each value with a consistent placeholder, and can reversibly rehydrate the model's response with the original values. Everything runs inside your own AWS account.

    How it fits an LLM workflow

    1. Your application sends text to the detect service's /deidentify endpoint, for example "Follow up with Jane Smith (MRN 4482913) about her 03/14/2026 visit."
    2. It returns de-identified text such as "Follow up with FIRST_NAME_1  about her [DOB_1] visit." plus a session ID. Placeholders stay consistent within a session, so the model can still reason about who and what is being discussed.
    3. You send only the de-identified text to your model: Amazon Bedrock, a self-hosted model, or any third-party API.
    4. When the response comes back, an authorized caller sends it to the separate rehydrate service's /reidentify endpoint, which restores the original values from the encrypted vault.

    Where teams use it

    • Prompt and response protection for chat assistants, copilots and agents that handle patient or customer data.
    • Retrieval-augmented generation: de-identify documents before chunking and embedding, so vector stores and indexes never hold raw identifiers.
    • Agent tool calls: de-identify tool outputs (EHR lookups, CRM records, tickets) before they enter the model's context.
    • Evaluation, fine-tuning and analytics datasets built from de-identified text.
    • Logs, traces and prompt archives stored without raw identifiers.

    Inference that stays in your VPC

    • Detection uses clinical NLP models running on CPU. No GPU and no external model API are required, and the model is packaged inside the signed container image.
    • Measured on a single AWS Graviton c9g.xlarge task: short prompts of about 150 characters take roughly 30 ms each; a 5,250-character clinical note takes about 0.8 s (p50). Scale horizontally by adding tasks or pods.
    • No NAT gateway or internet egress is required. The service reaches AWS APIs (KMS, AWS Marketplace metering, Amazon VPC Lattice, S3, Secrets Manager, ECR, CloudWatch Logs) only through VPC endpoints.

    Compliance and security controls

    • Your data stays in your account. Text, placeholders, token mappings and keys never leave your VPC, and Tenet has no access to your deployment. Usage reporting to Tenet is off by default; if you opt in, only hourly counts are sent (documents per size band, findings per entity type), never text or values.
    • Customer-managed encryption. Each stored value is sealed with a public key, and the matching private key is wrapped by your own AWS KMS key. Your key policy lets only the rehydrate service unwrap it, so the detect service cannot read back what it stored. Tenet has no decrypt path. Vault keys can be rotated without breaking earlier sessions.
    • Privileged, audited rehydration. Rehydration runs as a separate service with its own IAM role, and can be disabled entirely. Callers sign requests with their own IAM roles through Amazon VPC Lattice, with no shared secrets, and you grant each role only the routes it needs. Every rehydration request, including failed lookups, writes an audit record naming the caller to your Amazon S3 bucket with Object Lock. If the audit record cannot be written, nothing is restored. Erasing a session is a separate permission, audited the same way before anything is deleted.
    • Least privilege by design. Separate IAM roles per service, non-root containers, read-only root filesystems, and optional approval tagging plus an MFA-gated break-glass role for rehydration.
    • Usage-based billing through AWS Marketplace. Metering records contain document counts by size band, never payload-derived data, so your auditors can inspect exactly what is recorded.
    • Supply-chain integrity. Images are multi-architecture (arm64 and amd64), vulnerability-scanned, and signed so you can verify them before deploying.

    These controls support HIPAA programs built around the minimum necessary standard, access control and audit logging. Facility and organization names are not reliably detected. As with any de-identification tool, you remain responsible for validating results for your data and use case.

    Deployment

    • Amazon EKS with the provided Helm chart, or Amazon ECS on EC2 with the provided AWS CloudFormation template. A Terraform module is also provided.
    • Deployment guides: platform.tenetlabs.com/enterprise/deploy-ecs and platform.tenetlabs.com/enterprise/deploy-eks. On Amazon EKS, use Graviton4 nodes for now, as the guide explains.
    • Billed through your AWS account based on usage. Private offers are available for custom pricing and terms.

    Highlights

    • Runs entirely in your AWS account: no data, tokens or keys leave your VPC, with your own AWS KMS key protecting the vault.
    • Reversible de-identification for LLM workflows: rehydrate original values through a separate, IAM-gated service that audits every request to your S3 bucket with Object Lock.
    • Deploy on Amazon EKS (Helm) or Amazon ECS (CloudFormation) on AWS Graviton; Terraform module included. No GPU or external model API required.

    Details

    Delivery method

    Supported services

    Delivery option
    Container image for Amazon ECS (CloudFormation)
    Helm chart for Amazon EKS

    Latest version

    Operating system
    Linux

    Deployed on AWS

    Features and programs

    Financing for AWS Marketplace purchases

    AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
    Financing for AWS Marketplace purchases

    Pricing

    Tenet PII Engine - In-VPC PHI/PII De-identification

     Info
    Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time.
    Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator  to estimate your infrastructure costs.

    Usage costs (5)

     Info
    Dimension
    Description
    Cost/unit
    Documents up to 256 characters
    One document of up to 256 characters sent to detect, redact or de-identify.
    $0.000075
    Documents of 257 to 2,048 characters
    One document of 257 to 2,048 characters sent to detect, redact or de-identify.
    $0.000525
    Documents of 2,049 to 8,192 characters
    One document of 2,049 to 8,192 characters sent to detect, redact or de-identify.
    $0.002025
    Documents over 8,192 characters
    One document of more than 8,192 characters sent to detect, redact or de-identify.
    $0.008175
    Each 4,096 characters above 32,768
    One per started 4,096 characters above 32,768 in a document, added to the document.
    $0.00105

    Vendor refund policy

    Usage is billed as it is consumed. To request a refund, email team@tenetlabs.com  with your AWS account ID and agreement ID.

    How can we make this page better?

    Tell us how we can improve this page, or report an issue with this product.
    Tell us how we can improve this page, or report an issue with this product.

    Legal

    Vendor terms and conditions

    Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .

    Content disclaimer

    Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.

    Usage information

     Info

    Delivery details

    Container image for Amazon ECS (CloudFormation)

    Supported services: Learn more 
    • Amazon ECS
    Container image

    Containers are lightweight, portable execution environments that wrap server application software in a filesystem that includes everything it needs to run. Container applications run on supported container runtimes and orchestration services, such as Amazon Elastic Container Service (Amazon ECS) or Amazon Elastic Kubernetes Service (Amazon EKS). Both eliminate the need for you to install and operate your own container orchestration software by managing and scheduling containers on a scalable cluster of virtual machines.

    Version release notes

    The Helm chart can now send usage counts to Tenet, as the CloudFormation template and the Terraform module already can. The runtime, the engine (0.10.0) and the model are the same as in 0.5.3.

    • telemetry.egress in the Helm chart (opt-in, off by default). With enabled: true and apiKeySecret.name, a Secret you create with a Tenet Platform API key, the detect pods send the hourly counts-only usage records to Tenet. Records hold documents per size band and findings per entity label, never text or values. The detect NetworkPolicy gains one rule: tcp/443 to public addresses, with private, carrier-grade NAT and link-local ranges (including the instance metadata service) still closed. Rehydrate doesn't change. The route out, a NAT gateway on your node subnets, is yours to provide. The chart refuses to render with enabled: true and no key Secret, or with an invalid sinkUrl.
    • Left at the default, the chart renders the same resources as 0.5.3. The CloudFormation template and the Terraform module don't change.

    Upgrading from 0.5.3 replaces only the task definitions (ECS) or the image (EKS). No IAM, key or network changes unless you turn on telemetry.egress.

    Additional details

    Usage instructions

    Deploy on Amazon ECS on EC2 (Graviton c9g.xlarge recommended; Fargate is not supported) with the CloudFormation reference template, which creates the customer-managed KMS key, the Aurora PostgreSQL vault, the S3 Object Lock audit bucket, VPC endpoints (no NAT), separate task roles for detection and rehydration, and Amazon VPC Lattice services with IAM auth policies: your applications sign requests with their own IAM role, with no shared secrets.

    Set the template's ImageUri parameter to 709825985650.dkr.ecr.us-east-1.amazonaws.com/tenet-labs/tenet-pii-engine:0.5.4, in any Region. AWS Marketplace serves it only from us-east-1; elsewhere the template's VPC endpoints reach it across Regions, so don't copy it. Full instructions, parameters and the template: https://platform.tenetlabs.com/enterprise/deploy-ecs 

    Verify the image digest and signature: https://platform.tenetlabs.com/enterprise/verify 

    Resources

    Vendor resources

    Support

    Vendor support

    Email team@tenetlabs.com  with your AWS account ID and agreement ID. Support covers deployment, configuration and upgrades of the Tenet PII Engine in your account. Response times and any additional support commitments are defined in your private offer.

    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

    Customer reviews

    Ratings and reviews

     Info
    0 ratings
    5 star
    4 star
    3 star
    2 star
    1 star
    0%
    0%
    0%
    0%
    0%
    0 reviews
    No customer reviews yet
    Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.