AI automation has spent two years being sold on promises. On Monday, one of Europe's biggest telecom companies put a price on it. Deutsche Telekom said it expects AI and automation to save about €2.5 billion in indirect costs by 2030, compared with 2023.

I'm Nitin Kumar, Lead Project Coordinator at Debugged Pro. Our team builds software and AI features for businesses in India and the USA. Clients ask us one question more than any other: will this actually pay off? Telekom's plan is one of the clearest public answers we have seen from a large company, so we went through it in detail. Here is what it says and what agencies can take from it.

Deutsche Telekom headquarters in Bonn. Image credit: Deutsche Telekom / Norbert Ittermann

What Deutsche Telekom Announced About AI Automation

The numbers came out at the company's first investor day focused only on AI, held in Bonn on October 5, 2026. As Reuters reported on Deutsche Telekom's savings target, the key figures are:

About €1.1 billion in gross savings by 2027, outside the United States, compared with 2023.
About €2.5 billion (roughly $2.8 billion) in indirect cost savings by 2030.
AI revenue from business customers outside the US is expected to grow from about €250 million this year to about €800 million by 2030.
Part of the savings will be reinvested in digital transformation and faster fiber rollout in Germany.

One detail stood out to us. The 2027 savings figure does not include the cost of using the AI itself, often called token costs. That is an honest caveat, and it is one every business should keep in mind.

CEO Tim Höttges said, "AI is fundamentally transforming Deutsche Telekom." He added, "What matters most is that technology delivers measurable value, for our customers and for our company."

Where the AI Automation Savings Come From

Telekom is not spreading AI everywhere at once. It is aiming at areas with large volumes of routine work: network operations, customer service, software development and administration. The full agenda is on the Deutsche Telekom AI Investor Day page, which lists sessions on networks, customer service, internal IT and software productivity.

Customer service

Telekom's chatbot, called "Frag Magenta," handled about 2.6 million customer service calls in the first half of the year. In the US, the results are even bigger. At T-Mobile US, customer service calls have dropped by 55 percent, and AI agents now handle 40 percent of all customer contacts.

Network operations

The "RAN Guardian" agent watches the mobile network for congestion, for example during large events. Telekom says the response time has fallen from several hours to about one minute.

Fewer complaints

When new connections are set up, AI is used to spot problems before they turn into complaints. Telekom says early use cases show 30 percent fewer complaints.

Staff training

Under a program called "AI for All," employees get tools such as the in-house AskT, ChatGPT Enterprise and Microsoft Copilot. More than 100,000 employees have been trained. The company says people will set goals, review results and stay responsible for decisions while AI does more of the work.

Why This AI Automation Plan Is Worth Studying

Plenty of companies talk about AI. Few publish targets that investors will hold them to. That is what makes this plan useful for agencies and their clients.

It ties AI to a cost line. The savings are linked to indirect costs and a base year. That gives a clear way to check progress.
It starts with boring, high-volume work. Call handling, network alerts and complaint prevention are not flashy. They are where the hours and money go.
It admits that AI has its own running costs. Leaving token costs out of the gross figure is stated openly, not hidden.
It treats AI as a product too. Telekom is selling AI services, such as its Industrial AI Cloud in Munich and a platform for smaller firms to automate recurring tasks.

The plan also has a people side that should not be skipped. Telekom has not said how many jobs automation will cost. At T-Mobile US, a parallel restructuring will remove about 4,700 positions, which the company links to leaner structures and duplicate roles rather than AI alone.

Two colleagues reviewing an AI agent workflow map on a table, showing how AI automation is governed — Telekom says AI agents need clear roles, rights and approvals. AI-generated image credit: Deutsche Telekom AG

How Telekom Keeps AI Agents Under Control

Savings only last if the systems behind them stay in check. A month before the investor day, Telekom launched two platforms for running AI agents in business. You can read the details in Telekom's announcement of its AI agent platforms.

T-AI Agent Platform is aimed at small and mid-sized firms. It is built on n8n workflow automation and connects to email, CRM and ERP systems, with a clear view of agent costs and actions.
T-AI Agent Hub from T-Systems is for large companies and the public sector. It works like a control room, setting access rules, tracking which model is doing what, managing costs and knowing when to hand a task to a person.

Ferri Abolhassan, CEO of T-Systems, put it simply: "AI needs to pay off in everyday operations." He said the breakthrough "will not come from ever-new models, but when AI agents take on specific work." We agree, and it matches what we see on client projects. If you are planning more than one or two agents, our guide on how to scale enterprise AI agents covers the control side in more depth.

What AI Automation Means for Agencies in India and the USA

Most of our clients are not telecom giants. But the same pattern applies to a 50-person logistics firm in Pune or a retail brand in Texas. Here is how we read it for agency work.

Clients will ask for numbers, not demos

When a company like Telekom publishes a savings target, boards elsewhere start asking for the same. Agencies that can show a before and after on cost, time or error rates will win more work than those showing a clever chatbot.

Integration is the real job

Telekom's platform pitch is all about plugging into existing systems like CRM and ERP. That is the part clients struggle with most. The model is rarely the hard part. Clean data, reliable APIs and good error handling are.

Running costs need a budget line

Token costs can grow fast once an agent handles thousands of requests a day. We now include a cost estimate and a monthly usage cap in every AI proposal, so there are no surprises.

People stay in charge

Telekom says its staff will set goals and own decisions. That is a good default for any client. It also helps when customers or regulators ask who is responsible.

A Simple AI Automation Playbook for Client Projects

Here is the step-by-step approach we use when a client wants to copy this kind of result on a smaller scale:

List the high-volume tasks. Look for work that repeats hundreds of times a week, like support tickets, order checks or invoice matching.
Measure the baseline first. Record how long each task takes and what it costs today. Without this, you cannot prove any saving.
Pick one or two tasks to start. Choose ones where mistakes are easy to catch and fix.
Build the integrations properly. Connect to the CRM, help desk or ERP with proper logging and error handling.
Add human review for risky steps. Refunds, account changes and anything customer-facing should have a clear handoff to a person.
Track running costs weekly. Compare AI usage costs against time saved. Gross savings mean little if the AI bill eats them.
Scale only what works. Expand to the next task once the first one shows a clear, steady result.

For a look at how a large platform vendor is packaging agents for the same kind of work, see our breakdown of ServiceNow's AI agent for delivery teams.

Quick reference table

Area	Telekom result	Lesson for client projects
Customer service	AI agents handle 40% of T-Mobile US contacts	Start with common, simple questions
Network alerts	Response time cut from hours to about a minute	Use AI to spot problems early
New connections	30% fewer complaints in early use cases	Catch issues before customers see them
Costs	Gross savings shown before token costs	Budget AI running costs from day one
People	100,000+ staff trained on AI tools	Train users and keep them accountable

A short example

Picture an online furniture store in Bengaluru with 3,000 support emails a month. Most ask about delivery dates. An AI agent connected to the order system answers those in seconds. Anything about damage or refunds goes straight to a person.

After a month, the store compares time spent per ticket and the AI bill against the old numbers. If the saving is clear, it moves on to returns. If not, it adjusts before spending more.

Risks to Keep in Mind

Telekom's plan is not risk free, and neither is yours. Analysts at Bank of America note that AI agents also make it easier for customers to compare plans and switch providers. In other words, the same technology that cuts costs can raise competitive pressure.

Targets are not results. The 2030 figure is a plan, not a finished outcome.
Running costs can shift. Model prices and usage volumes change, which affects real savings.
Customer trust is fragile. A bad automated answer can cost more than it saves.

If you want help working out which tasks in your business are worth automating, and how to measure the result, our software development consulting team works with companies in India and the USA on exactly this.

Key Takeaways

Deutsche Telekom expects AI automation to save about €1.1 billion by 2027 and about €2.5 billion in indirect costs by 2030, compared with 2023.
The 2027 savings figure does not include AI running costs, a caveat every business should copy.
Early results include 55% fewer service calls at T-Mobile US and network response times cut to about a minute.
Telekom pairs automation with control tools, human oversight and staff training.
Agencies should lead with measurable outcomes, solid integrations and clear cost tracking.

FAQ: AI Automation and Deutsche Telekom's Plan

How much does Deutsche Telekom expect to save with AI automation?

Deutsche Telekom said AI and automation should bring about €1.1 billion in gross savings outside the US by 2027 compared with 2023. It expects that to grow to about €2.5 billion in indirect cost savings by 2030. The 2027 figure does not include the cost of running the AI itself.

Where is Deutsche Telekom using AI automation today?

The company is focusing on network operations, customer service, software development and administration. Examples include the Frag Magenta chatbot, the RAN Guardian agent for spotting network congestion and AI agents at T-Mobile US that handle 40 percent of customer contacts.

Does AI automation mean job cuts?

Deutsche Telekom has not said how many jobs automation will affect. At T-Mobile US, a separate restructuring will remove about 4,700 positions, which the company links to leaner structures and duplicate roles. Telekom says its staff will set goals, review AI output and stay responsible for decisions.

What can a small or mid-sized business learn from this?

Start with high-volume, repetitive work where results are easy to measure, such as support tickets or order checks. Track the cost of running the AI from day one, and keep a person in charge of decisions that affect customers.

How can an agency help clients with AI automation projects?

An agency can map the client's processes, pick the few tasks worth automating, build the integrations with tools like CRM and ERP systems, and set up cost tracking and human review. The goal is a measurable result, not a demo.

About the Author

Nitin Kumar is Lead Project Coordinator at Debugged Pro. He plans software and AI projects for clients in India and the USA, and works with delivery teams on scope, cost tracking and the safe rollout of AI automation in production.