Tata Steel
Tata Steel rolls out over 300 AI agents in nine months with Google Cloud
Tata Steel and Google Cloud say the steelmaker deployed over 300 specialized AI agents in nine months, built on an internal low-code platform. They report that its HR helpdesk assistant resolves more than 70% of routine tickets and that complaint-handling agents cut average turnaround time by 50%.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
The problem
A global steelmaker runs plants, suppliers, customers and back offices across five continents. Its data sat in many separate systems, and routine work such as HR questions, invoice checks and complaint routing took staff time that could go to more valuable tasks.
What they built
Tata Steel built an internal platform called Zen AI. It lets employees who are not data scientists, such as software developers and frontline managers, build, test and deploy their own AI agents with little code. A second tool, the Tata Steel Digital Assistant, is an internal portal that lets staff ask questions across public data, company systems and their own files such as call recordings and PDFs.
How they did it
The platform is built on Google Cloud's Agent Development Kit and linked to its data warehouse (BigQuery) and storage. The agents run on Google's Cloud Run service. The company says it uses Gemini and other Google models, matched to each task. Agents cover HR helpdesk support, invoice processing, tax classification, contract analysis, equipment maintenance plans and complaint handling, where images and text are used to detect the complaint's intent and defects and route the case. A video agent called Safety EyeQ checks live camera feeds in risky areas against safety procedures. The company reports deploying over 300 agents in nine months.
Results
The companies report that the Digital Assistant helps the HR helpdesk resolve more than 70% of routine employee tickets autonomously. They say complaint-handling agents cut average turnaround time by 50%. The release also describes safety alerts and maintenance planning but gives no figures for them.
Lessons for your project
Let non-specialists build agents. Tata Steel gave software developers and frontline managers a low-code platform, so many small teams could build agents instead of one central team.
Put decades of operational data in one governed place first. The company credits an early investment in consolidated data for the speed of its agent rollout.
Start with high-volume internal work such as HR helpdesk tickets, invoices and contracts, where a share of cases can be resolved without a person.
Match the model to the task. The company uses a catalogue of models rather than a single one.
Caveats
- The figures come from a joint press release published on Google Cloud's site. They are reported by the company and its vendor, and no independent source confirms them.
- The release says the Digital Assistant 'assists' the HR helpdesk in resolving more than 70% of routine tickets autonomously. It gives no ticket volume or period.
- The release names Gemini and a vision-language model it spells 'Palli Gemma', but gives no model versions and says the company has access to over 200 models.
- The release gives no go-live date. The deployed date shown is the announcement date, and the nine-month period ends around then.
- The 50% turnaround figure covers complaint-handling agents only, with no baseline or sample size.
Sources
- Tata Steel Partners with Google Cloud To Deploy a Unified Agentic AI Across its Global Value Chain
Google Cloud Press Corner · 2026-04-22 · Press release · googlecloudpresscorner.com
Summarized in our own words from the sources above. Spotted an error? See corrections.