---
title: "How Concurrence governs clinical AI at a trillion-token scale with Unity Gateway"
description: "Concurrence processes 100.8 billion input tokens and 11.2 million LLM calls every 30 days, at an annualized rate of 1.2 trillion tokens. To govern these workflows, the company consolidates on Databricks with Lakebase for operational state, Unity Catalog for governance, and Unity Gateway for centralized AI access. Patient data is recorded as immutable events in Delta tables, from which a current patient state—called the world model—is computed while preserving provenance. All agents are tested against simulated patients before deployment; simulation traffic is approximately seven times production traffic, using immutable event history to replay states. Concurrence routes all coding-agent traffic through Unity Gateway, attributing each request to the engineer, and enforces compliance-gated model routing where HIPAA requirements are met first."
---

# How Concurrence governs clinical AI at a trillion-token scale with Unity Gateway

[Databricks](https://yomu.fyi/company/databricks) · Ali Khokhar, John Xing, Tony Shi, Kelly Albano · Sep 23, 2026

**Type:** Explainer

## Summary

Concurrence processes 100.8 billion input tokens and 11.2 million LLM calls every 30 days, at an annualized rate of 1.2 trillion tokens. To govern these workflows, the company consolidates on Databricks with Lakebase for operational state, Unity Catalog for governance, and Unity Gateway for centralized AI access. Patient data is recorded as immutable events in Delta tables, from which a current patient state—called the world model—is computed while preserving provenance. All agents are tested against simulated patients before deployment; simulation traffic is approximately seven times production traffic, using immutable event history to replay states. Concurrence routes all coding-agent traffic through Unity Gateway, attributing each request to the engineer, and enforces compliance-gated model routing where HIPAA requirements are met first.

## Context

Concurrence operates clinical AI agents for healthcare, processing over a trillion tokens annually. The company requires strict compliance (HIPAA, GDPR, SOC 2) and must ensure data provenance, agent safety, and governed AI access across batch, real-time, and developer workloads.

## Approach / What changed

Concurrence uses Databricks with Lakebase for operational state, Unity Catalog for governance, and Unity Gateway for centralized AI access. Patient data is stored as immutable events in Delta tables, and the patient state is derived from this event history. All coding-agent traffic is routed through Unity Gateway’s ug CLI with per-user attribution. Model routing is compliance-gated, requiring HIPAA coverage before cost or performance are considered.

## Takeaways

- Concurrence processes an annualized run rate of roughly 1.2 trillion input tokens, with monthly token volume growing about 5x from its late-2025 baseline to July 2026.
- All agents are tested against simulated patients before deployment; simulation and evaluation traffic is approximately seven times greater than production traffic.
- Concurrence routes all coding-agent traffic through Unity Gateway; in July, 14 users generated 35.85 billion input tokens through Unity Gateway, with 95.37% being cache reads.

## FAQ

### How does Concurrence ensure data provenance for patient information?

New information is recorded as immutable events rather than overwriting existing records. The current patient state is computed from this event history, preserving the source and provenance of each piece of information, which allows agents to have a consistent view and history of what is known about a patient.

### How does Concurrence test clinical agents before deployment?

All agents are tested against simulated patients. Because the patient state is computed from an immutable event history, teams can replay that state and test different scenarios without affecting real patient records. Simulation and evaluation traffic is about seven times greater than production traffic.

### How does Concurrence prioritize model selection for clinical workloads?

Model routing is compliance-gated before it is cost-gated. Models must first meet the HIPAA requirements of a workload before Concurrence considers quality, performance, or cost. The same compliance path also runs safety classification for issues like self-harm and medical emergencies.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents), [AI Governance](https://yomu.fyi/topic/ai-governance), [Databricks](https://yomu.fyi/topic/databricks), [Healthcare](https://yomu.fyi/topic/healthcare), [Unity Gateway](https://yomu.fyi/topic/unity-gateway)

- Source: [Databricks](https://www.databricks.com/blog/how-concurrence-governs-clinical-ai-trillion-token-scale-unity-gateway)
- Source URL: https://www.databricks.com/blog/how-concurrence-governs-clinical-ai-trillion-token-scale-unity-gateway
- Ingested by Yomu: 2026-09-23T18:00:08.877Z

[Read original post](https://www.databricks.com/blog/how-concurrence-governs-clinical-ai-trillion-token-scale-unity-gateway)