Enterprises are increasingly moving AI beyond experimentation, but turning deployments into measurable business value remains a challenge.
According to Gartner, only 23% of CxOs report confidence in their organisation’s genAI outputs.
With a large gap between deploying AI and realising value at scale, Huawei has outlined a seven-step approach for AI adoption.
The framework, presented by Tao Jingwen, Huawei’s deputy chairman of the supervisory board and chair of global industry business operations, helps enterprises move artificial intelligence from individual use cases into production systems.
“Currently, AI is advancing much faster than people’s ability to adapt to it. AI has become more accessible, yet the challenge lies in how industries, employees, and organisations adapt to and harness it in the real world,” Tao said.
7 steps to AI adoption
This is not about mounting an engine onto an old carriage, but about designing a whole new car. Tao Jingwen
Huawei’s seven-step engineering approach aims to create a complete path from technology to actual implementation.
Tao added: “Future enterprise processes will be designed around collaboration between people and AI. This is not about mounting an engine onto an old carriage, but about designing a whole new car.”
1. Understand the industry’s business logic
The first step is to understand how the industry operates and identify its core business problems.
Moving directly into execution without first understanding these problems can result in a rushed approach to AI adoption.
2. Identify core industry scenarios
The next step is to identify where AI can deliver meaningful value.
Tao said enterprises do not necessarily need more AI models, algorithms or computing power. Instead, they need to identify the right industry scenarios in which those technologies can be applied.
Based on its Huawei projects, the company developed 12 questions to help determine suitable AI applications in complex industries, including what AI can do effectively and where it can create value.
The objective, Tao said, is to move beyond experimentation and use AI to generate tangible business value.
3. Address technical and engineering challenges
Once enterprises identify a suitable use case, they must address the engineering work required to make it viable.
Tao highlighted the need to select appropriate models and algorithms while modernising enterprise data and knowledge engineering.
This stage also involves testing AI technologies against real industry requirements rather than relying solely on theoretical capabilities.
4. Deploy AI in core production scenarios
The fourth step is to move AI from experimentation into production.
“We need to deploy the AI in the middle of the production system,” Tao said.
He added that AI that remains disconnected from production systems and business processes will struggle to generate meaningful business value.
5. Expand across value streams
After establishing AI in a core production scenario, enterprises should gradually expand from point solutions to systematic applications, Tao said.
Tao argued that improving individual functions is not enough if those improvements do not translate into broader operational gains.
6. Build an industry ecosystem
AI deployment also requires collaboration among technology providers, industry specialists, application developers and infrastructure providers.
For Huawei, this means developing a more open ecosystem around industry AI deployments.
7. Continuously raise the level of industry intelligence
The final step underscores an ongoing optimisation process.
Enterprises need to expand AI into additional business processes and continuously improve existing applications as technologies, data and business requirements evolve.
“The seven-step approach provides a complete path for applying AI to core business scenarios, from understanding scenarios to embedding AI into production systems, from single-point intelligence to system-level intelligence, and from individuals to ecosystems,” Tao explained.
Short-chain operations for closer partnerships
According to Tao, one persistent gap to address is the divide between technology specialists and industry practitioners.
Technology experts may understand AI and digital technologies but lack deep knowledge of specific business processes, while industry specialists may understand those processes without having the technical expertise needed to deploy AI.
Organisations must also restructure to adapt to technological advancements.
Short-chain operations allow us to quickly respond to customers’ real needs. Tao Jingwen
To address this, Tao reported that Huawei has strengthened its own industry-specific organisations to further streamline research, marketing, sales, and service activities. The company has established 10 industry groups and 66 sub-industry groups.
“We required experts and scientists to leave the office, to work, to speak, to drink, to serve, and to solve customers’ problems,” he said.
Huawei has also established specialised AI teams, including a computing platform team and industry-specific slim & agile teams that aim to evaluate and deploy emerging AI technologies rapidly.
Tao said such teams helped build China’s first 10,000-card training system in June 2023.
“Short-chain operations allow us to quickly respond to customers’ real needs,” Tao said.
Tao emphasised that Huawei’s operating model is designed to shorten the traditional chain between R&D, products, sales and services. Ultimately, he believes the objective is to make increasingly complex AI technologies easier for enterprises to consume.
“We are committed to packaging complex technologies into other services, building a bridge between algorithms and models, and making artificial intelligence as powerful and easy to use,” Tao said.
