
AI,
efficiency,
CX trends,
technology,
Published on Mon Sep 28 2026
Updated on Mon Sep 28 2026
4 minute read
DeepSeek was only founded in 2023. The company is based in Hangzhou, an important port city in Eastern China that has been a strategic hub along the Silk Roads for thousands of years. DeepSeek has been building large language models (LLMs) since the company was created, but the 2025 release of their R1 LLM challenged all the conventional wisdom around artificial intelligence (AI) and the construction of LLMs.
AI research has always been prohibitively expensive. Companies developing LLMs need very deep pockets and investors who are prepared to keep losing cash - until a future date when so many people need AI systems that the returns start arriving. OpenAI is a good example of the problem. Although they created the most popular and well-known chatbot in the world - ChatGPT - the company still loses money. In fact, even their $200-per-month ChatGPT Pro plan loses money. OpenAI has raised around $20 billion from investors since the company was founded ten years ago. When they released financial data in September 2024 they declared revenue of over $3.7 billion, but still made a loss of around $5 billion. Getting AI to work well is expensive. No matter how much these companies are earning, it seems they are spending more on research and LLM development.
This is the established message on AI that DeepSeek is challenging. The DeepSeek R1 model was trained for less than $6 million. It is no surprise that the research journals have contrasted this with the o1 model launched by OpenAI last September - training cost around $500 million. Importantly, DeepSeek is also offering a model that can be run by individuals on their own computer. It doesn’t require a huge data center consuming vast amounts of energy. There are many technology journalists and software engineers who have written detailed assessments of how DeepSeek has managed to achieve this, and the implication for geopolitics and the global technology industry, but I want to focus here on the implications for customer experience processes.
Many CX leaders read all the articles about how AI is transforming CX and they want a rapid solution. They want some of the innovation and business transformation that this technology promises. A Zendesk survey from last year found that around 80% of executives believe that traditional CX technologies “will be dead inside three years.” But their excitement often turns into disappointment when they find that their data requires a lot of work to make it suitable to train the AI model. It will take time. It will require up-front investment in the cleanup and training work. Analysts such as Gartner also suggest that many AI projects fail.

Created at Mon Sep 28 2026
4 min read
DeepSeek was only founded in 2023. The company is based in Hangzhou, an important port city in Eastern China that has been a strategic hub along the Silk Roads for thousands of years. DeepSeek has been building large language models (LLMs) since the company was created, but the[ 2025 release of their R1 LLM challenged](https://www.techtarget.com/whatis/feature/
This is where DeepSeek could reshape the opportunity for companies to embrace AI, even smaller companies. The ability of DeepSeek to run locally on nothing more than a regular PC also creates interesting opportunities, such as being able to apply the DeepSeek model to your FAQ or internal customer support information. Larger companies may have expected to budget for expensive cloud computing power and high energy costs. Could this be diverted directly into AI projects and other business priorities? An entirely new opportunity is being created for small companies and larger companies may soon find that their AI plans could be far less expensive than planned.
Most of the commentary about AI in CX focuses on automation and the use of chatbots, but there are many different ways in which AI can contribute positively to the broader experience a customer has with a brand. Some ideas include:
DeepSeek, and similar models, may turbocharge interest in AI for customer experience designers. Executives that have been disappointed by the cost or time required to launch their AI initiatives may now find that they have a new path to follow and this can create opportunities for smaller businesses that never even considered the use of their own LLM as feasible.
This could create an important race as brands fight to introduce improvements to their CX that were previously too expensive or difficult to achieve. The gloves are off. Which CX initiative will you plan to implement first?

Created at Fri Sep 25 2026
5 min read
An automated system has answered a customer’s question. The answer is accurate in itself, but it has missed the true reason for the contact. The customer has to explain the situation again, this time to human support. The support representative resolves the issue, but notices something else: the same misunderstanding keeps appearing across multiple customer journeys. When a mistake repeats at scale, who owns the lesson? And what should a global enterprise expect from a CX partner whose job inclu

Created at Mon Sep 21 2026
3 min read
A permanent resignation or a temporary engagement drop? Either way, for BFSI brands, losing their most experienced experts’ efforts for even a moment can turn routine risk into a crisis. For many, CX is the weakest link in this chain. Call centre agent turnover in the UK runs at 31.2% a year and is still rising. This against a European workforce that [Gal