AIToday
Large Language ModelsAI Business & IndustryDatabricks AI BlogPublished: Nov 19, 2026, 01:00 JST

AUMOVIO centralizes all LLM access on Databricks Unity Gateway

Send on LINE
AUMOVIO centralizes all LLM access on Databricks Unity Gateway

3 Key Points

  1. What happened

    AUMOVIO chose Databricks Unity Gateway as its single route to LLMs after a cross-functional taskforce picked one central gateway; it routes 1adp-* aliases to Databricks-hosted models, Google Vertex AI, Azure OpenAI or AWS Bedrock.

  2. Why it matters

    One shared route gives AUMOVIO a single place for model approvals, usage tracking and cost attribution, so teams stop building their own access paths as AI adoption grows.

WHO IT HITSEnterprise platform and IT teams at large companies that let many departments use multiple AI model providers will see a template here: one gateway with a curated alias catalog, token-level cost chargeback, and provider swaps done in infrastructure code rather than application code.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

AUMOVIO's problem was a familiar one at scale: teams had adopted LLMs through provider integrations and gateways built for specific use cases, and that variety added complexity to operations, cost attribution and consistent governance. A cross-functional AI taskforce with stakeholders from security, data protection, business areas and IT settled on one central AI Gateway as the shared route for LLM access and governance, and the company then had to make that central path useful to teams and their applications.

The build-versus-buy reasoning is where the detail sits. AUMOVIO's engineers argued that operating a gateway is a permanent commitment — provider APIs change, dependencies need patching, security requirements evolve, and users need support — so the company weighed that against buying another gateway and chose Databricks Unity Gateway for five stated reasons, the most important being that Databricks was already the data platform behind 1ADP, the One AUMOVIO Data Platform. On top of the gateway AUMOVIO built cost attribution and chargeback, a curated model catalog, and the 1ADP AI App, which runs in production on the same gateway and gives employees a governed way to use AI without writing code while developers get an OpenAI-compatible endpoint.

The company's own lessons section is unusually direct about what it did not build: the gateway was not genuinely theirs to build, while cost attribution, the model catalog and the employee-facing AI App were, because they encode how AUMOVIO works. Looking ahead, the next phase shifts from model access to agents — sharing agents inside 1ADP projects and connecting knowledge sources, web search and other tools through managed and bring-your-own MCP connections — which raises questions about what an agent may retrieve, which tools it may call and whose permissions it acts under, and AUMOVIO points to having one framework to answer them in as the reason it consolidated in the first place.

FAQ
Why did AUMOVIO choose Databricks Unity Gateway instead of building its own AI gateway?
Because Databricks already powered its enterprise data platform, AUMOVIO could reuse existing infrastructure, Microsoft Entra ID integration and governance controls. That let its teams focus on company-specific work such as cost attribution, a curated model catalog and an employee-facing AI application.
How does the gateway keep AUMOVIO from being locked into one model provider?
AUMOVIO publishes a curated catalog of 1adp-* model aliases, and Unity Gateway routes each alias to whichever provider serves it. The company can re-point an alias to a different provider or a newer model without a single consumer changing a line of code.
How does AUMOVIO track and charge back AI costs?
Unity Gateway reports how many tokens were consumed, and every user assigns consumption to a personal cost center or project. Pipelines join that assignment with Databricks system tables and publish dashboards showing which models are used, by whom and what they cost.
Databricks AI BlogRead Original Article
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleCADMAGEDDON on Qwen3.8-27B beats frontier CAD models across five benchmarks