LITTLE TOWN LABS

De-identification // Analytics // AI Gateways // AI Implementation

Put AI to work on sensitive data. Without exposing it.

We help healthcare organizations and law firms de-identify data, turn it into reporting they can act on, govern every model call, and roll out AI tools with the safeguards their compliance teams expect.

Start with a de-identification auditSee our services
prompt.in
Summarize the visit for Maria Lopez, DOB 03/14/1961, MRN 00482913
gateway.out
Summarize the visit for [PATIENT_1], DOB [DATE_1], MRN [ID_1]
3 identifiers masked before the model saw them

What we do

Turn sensitive data into insight, and keep it protected wherever AI touches it.

Data De-identification

Remove PHI and confidential client details from structured, semi-structured, and unstructured data so it can feed reporting, analytics, and AI.

  • Validation audits of your current process
  • Pipelines built on Presidio and Philter
  • Deploy in your environment or ours (Azure)
  • Ongoing managed service

AI Gateway Implementation

Put a single, governed front door in front of every model your teams use, with PII guardrails, access control, logging, and cost tracking. Vendor-neutral, on the platform that fits you.

  • LiteLLM proxy deployment and configuration
  • Kong AI Gateway setup and policy design
  • Portkey / Prisma AIRS AI Gateway implementation
  • Prompt redaction before data leaves your network
  • Grafana monitoring for usage, cost, and audit trails

AI Implementation

Roll out AI assistants and agents to real teams, with use cases, safeguards, test plans, and training built around how your people actually work.

  • Use case discovery and pilot design
  • Safeguards for document systems and practice tools
  • Self-hosted agent memory and knowledge bases (OpenViking)
  • Governance aligned to the NIST AI RMF
  • Staff training and adoption support

Data Analytics and Reporting

Turn operational data into reporting leaders trust, from modern data platforms and BI dashboards to privacy-safe HTML reports built on de-identified snapshots, with no live connection to your source systems.

  • Azure data platform design and build
  • Power BI and Grafana dashboards and data models
  • Privacy-safe HTML reporting (Observable Framework, Evidence, Quarto)
  • Data quality, cleansing, and validation
  • Process improvement analysis

Protection at rest and in flight

De-identification secures the data you store and analyze. The AI gateway secures what your people send to models. Together they close the gap most organizations leave open.

At restEHR / case filesDe-identification pipelineDashboards and HTML reports
In flightStaff and appsAI gateway with PII guardrailsClaude, OpenAI, Azure, others

Who we serve

Built for regulated work

Healthcare

Clinics, health systems, and health tech teams that need patient data for analytics and AI without putting PHI at risk.

  • HIPAA Safe Harbor and Expert Determination support
  • EHR data (including eClinicalWorks) to Azure and Power BI
  • Clinical note de-identification
  • We sign BAAs
Healthcare services

Legal

Law firms adopting AI that must protect privilege, client confidentiality, and sensitive case material.

  • Safe AI rollouts inside document management systems
  • Redaction for discovery and protective orders
  • Guardrails on what leaves the firm
  • Practice management integrations
Legal services

How we work together

  1. 01 // Validate

    De-identification audit

    We test your current process against synthetic data with known PHI and show you exactly what leaks and what gets over-redacted.

  2. 02 // Implement

    Build and deploy

    Pipelines, gateways, and AI tooling deployed in your environment, tested, documented, and handed over.

  3. 03 // Manage

    Managed service

    We monitor, tune, and re-validate over time so protection keeps pace with new data and new models.

Benchmarks

We measure de-identification. We don't just promise it.

Our test harness generates realistic synthetic clinical notes with tagged ground-truth PHI, then scores any de-identification tool on recall, precision, and F1.

Read the benchmark report
[RECALL]
PHI caught
[PRECISION]
Redactions correct
[F1]
Overall score
Portrait of Gary Brown, founder of Little Town Labs

About

Gary Brown, Founder

Gary has spent more than 25 years in financial services data and analytics, one of the most regulated data environments there is. He founded Little Town Labs to bring that same discipline to healthcare and legal organizations adopting AI.

  • M.S. in Information Technology, Information Security
  • NIST AI RMF certified
  • SAFe 6 POPM and SSM, SAFe AI certified

Delivered in partnership with Timeless Technology Solutions.

Talk to us about your data

Tell us what you're working with. We'll reply within one business day with next steps, usually starting with a short discovery call.

Please don't include patient or client information in this form.