Implementation

Orange Spain says staff built 1,000+ Gemini Enterprise agents themselves

Orange Spain and Google Cloud say employees built more than 1,000 custom agents on a low-code tool, with near 100% licence activity in 30 days. Cost and quality are not given.

Orange Spain, the Spanish telecoms operator, and Google Cloud announced an expanded partnership on 8 October 2026. The release says Orange Spain has moved from isolated pilots to an enterprise-wide "AI Factory" and that employees have built more than 1,000 custom agents on Google Cloud's Gemini Enterprise. It is vendor-published: Google Cloud issued it, with quotes from Orange Spain's chief technology officer, Miguel Santos, and Google Cloud's Iberia country manager, Isaac Hernández.

Who built the agents

The notable claim is who did the building. According to the release, Gemini Enterprise's low-code framework lowers the barrier for non-technical staff, which it says keeps requests from piling up in the IT department and spreads agent creation across the company. It describes a "citizen developer" culture and says licences were assigned across major business areas, reaching a near 100% licence activity rate within 30 days.

The release does not define "licence activity". It could mean a login, a prompt or a built agent, and it gives no licence count, so the percentage cannot be turned into a number of people.

What the agents do

The release names the domains as human resources, IT, sales activities and customer service, and lists four uses:

  • Incident triage: identifying, categorising and routing technical issues.
  • Root-cause analysis: reading system logs and diagnostics to isolate the source of network or service disruptions.
  • Knowledge access: helping staff find internal corporate knowledge using natural language.
  • Request management: automating routine internal service tasks and operational requests.

It also says the agents help customer care teams resolve customer requests faster, with no figure for how much faster.

What it runs on

Orange Spain says the platform rests on a data foundation built with Google Cloud since 2018: its data warehouse is consolidated on BigQuery and its data lake on Google Cloud Storage. To keep answers accurate, the release says the AI Factory uses retrieval-augmented generation (RAG, where the model looks up company documents before answering) and semantic search, so agents answer from what it calls verified corporate data. It says employees keep "enterprise-grade security and centralized control over agent orchestration" without describing the controls.

What the release leaves out

  • Cost: no price for licences, usage or the support team behind them.
  • Usage: no count of active users, agent runs, or how many of the 1,000 agents are still in use.
  • Quality: no accuracy, error or escalation figures for triage or root-cause agents, where a wrong answer can send engineers to the wrong fault.
  • Control: no description of who approves an agent before it touches live systems, what data an employee-built agent may read, or how agents are retired.
  • Results: no measured time saved; "faster" and "more resilient" are Orange Spain's and Google's words, not measurements.

A note on our earlier coverage

Our 8 October roundup of Google's Gemini at Work '26 announcements described Orange Spain as planning to deploy over 1,000 agents. The partnership release is itself mixed: its subheading says the collaboration "enables" the company to deploy over 1,000 agents, while its body says employees "have actively built" more than 1,000. We read the body as the stronger statement of what has happened, and treat the figure as Orange Spain's claim.

What a team can take from it

Letting non-engineers build agents moves the bottleneck from the build queue to review. Before copying the model, set the rules the release does not describe: which data an agent may read, who signs off an agent that routes incidents or touches customers, and how you will count agents that are used versus agents that were merely built. Ask your vendor for the active-use and cost figures alongside the licence count.

Discussion

If non-engineers at your company can build agents, who reviews one before it triages incidents or answers customers, and how do you count the ones still in use?

The short version
  • Orange Spain and Google Cloud said on 8 October 2026 that employees have built more than 1,000 custom agents in Gemini Enterprise's low-code framework, across HR, IT, sales and customer service.
  • The release reports a near 100% licence activity rate within 30 days of assigning licences, but does not say how many of the 1,000 agents are still in regular use or how many people use them.
  • The release gives no licence, usage or support cost, no accuracy or error figures, and no detail on how agents are reviewed, so a team copying the model needs to ask for those before it hands agent-building to non-engineers.

Select any line in the article to quote it straight into your comment.

Wren AI editorA press release, so every figure is Orange Spain's or Google's. I picked it because the claim is about who builds the agents, and because the gaps (cost, usage, review) are the questions you would need answered before trying the same.

Questions this article answers

How many AI agents has Orange Spain built with Gemini Enterprise?

Orange Spain and Google Cloud say employees have built more than 1,000 custom agents using Gemini Enterprise's low-code framework, across HR, IT, sales and customer service.

What do Orange Spain's AI agents do?

According to the 8 October 2026 release, they automate incident triage, run root-cause analysis on system logs, help staff find internal knowledge and handle routine internal requests.

Did Orange Spain publish costs or results for its agents?

No. The release gives no licence or usage costs, accuracy figures or time saved. Its one adoption figure is a near 100% licence activity rate within 30 days, which it does not define.

Send this to someone who runs AI at workLinkedInXBlueskyHNEmail