The human foundations of AI: rethinking skills, structure and strategy

This is a guest blogpost by Arunava Bag, CTO, Digitate

Despite the promise of AI, most enterprises are struggling to move from pilot to production at scale, with three in four agentic AI pilots and 95% of GenAI pilots failing. Several factors contribute to this, such as board-level pressure to “apply AI anywhere” without a clear ROI and unclear implementation strategies.

However, the biggest barrier that is often overlooked is the human side of AI implementation. Fundamentally, it’s all well and good to roll out AI tools, but unless teams have the right foundational skills in place, the investment will fall flat.

Preparing enterprises and their teams for AI requires a fundamental rethink of employee training, hiring and job design. Reducing the skill gap and supporting employee skill development and satisfaction all are parts of this puzzle.

The big picture

Re-skilling can be grouped into two broad areas: IT department and general business users.

IT personnel should be re/up skilled to learn the new technologies in the AI domain, so that they can build, deliver and deploy the new age tools.

However, the more difficult and bigger focus is re/up skilling general business users, to allow them to:

• Appreciate the power of the new age (e.g. AI) technologies and its benefits to remove resistance
• Learn the basic features and benefits of the new technologies so that they can guide the IT department on better/optimal deployment of the technologies/products
• Effectively use the new age technology in their day-to-day work

The re-aggregation of skills

Traditionally, IT engineers worked in silos, each focused on specific layers of technology. Today, AI systems can resolve issues across multiple layers simultaneously. This “re-aggregation of skills” marks a major shift in how enterprises must think about their technical workforce. Teams that were once structured around specialisms now need to operate in a more fluid, cross-functional way, collaborating with AI systems that bridge those layers automatically.

As a result, organisations are being prompted to rethink their structures, training, and models of human–machine collaboration. Success depends not just on implementing tools, but on redesigning how people work with them, and how skills are distributed across the enterprise.

The market reflects the shift

The job market mirrors this transformation. Roles such as AI engineer, machine-learning engineer, and data scientist are in growing demand. In the UK alone, there were over 11,000 active automation and AI vacancies in August 2025. That’s a notable increase year on year, but still a fraction of the 717,000 overall job openings currently, and proof that most businesses are still early in their AI journey and working to build the right capabilities.

While some AI pilots are intentionally designed as learning exercises, never meant to scale, the broader trend is clear: more organisations want to move from experimentation to embedding AI into daily workflows.

Beyond the technical roles: AI’s expanding reach

Simply adopting AI tools is not enough. To move from pilot to production at scale, enterprises must prepare their people and processes to unlock real value.

AI’s reach is extending far beyond the technical teams. One of its most powerful effects is democratisation, empowering employees in non-technical/non-IT roles to work more efficiently and creatively.

AI is no longer confined to data science labs or engineering functions. It’s becoming part of everyday workflows across departments, from marketing teams generating content and analysing customer data, to HR teams using AI to streamline recruitment and enhance candidate experience.

Businesses are looking for candidates with AI fluency. This doesn’t necessarily mean they need deep technical expertise, but an understanding of how to work effectively alongside intelligent systems.

To achieve this, two types of reskilling are important:

• Learn the new products and features that employees would use in their day-to-day work effectively (e.g. Co-pilot, text to video conversion tool)
• Learn the broader features (and limitations) of AI technology to appreciate the benefits and reduce adoption resistance, as well as guide the IT department for optimal choice of new tools.

Along with this, employee education on AI guardrails and governance – such as AI bias, data privacy, hallucinations, and system limitations – will ensure better outcomes.

One example is adoption of Agentic AI tools, to drive productivity and resilience. By identifying and resolving faults automatically, downtime reduces and mitigates the ripple effect of system failures across the business. This kind of distributed intelligence helps organisations maintain momentum and focus on higher-value activities, whether that’s product innovation, customer engagement or operational excellence.

As part of this, major players are committing to AI training for their teams, including Santander, where AI training is now mandatory for all employees. Such initiatives underscore the growing understanding that AI capability must be built across the organisation, not concentrated in a single department.

Setting the right foundations

Business leaders should put budgets and efforts towards upskilling and reskilling across three categories, with investment weighted as follows:

• 65-70% in core business areas
• 20-25% in adjacent, business-related areas
• 5-10% in emerging seemingly unrelated but cutting-edge technologies, supported by a focus on a post- training feedback system with idea generation on usefulness of the technologies in the respective business areas

Organisations should plan reskilling and upskilling initiatives around defined goals. Beyond standalone training sessions, programs should include assessments, structured continuous learning pathways, and incentives for skill development to make them more engaging and impactful.

Structure is equally important. Traditional hierarchies can stifle the cross-functional collaboration AI thrives on. As a result, many organisations are moving toward more lateral or matrix-style structures, where teams are empowered to share knowledge and make decisions more quickly. This shift enables faster problem-solving and ensures that AI initiatives are driven by business needs rather than confined to departments.

Finally, training and support must be continuous. Successful AI adoption doesn’t end once a tool is implemented. The world of AI technologies is fast evolving, and new versions and features are released all the time.  The first six to twelve months are critical for building confidence, addressing challenges and refining processes. Enterprises should regularly review their technology stack to ensure seamless integration and alignment with evolving workflows. In essence, scaling AI requires sustained investment in people, not just platforms.

From pilot to scale

Enterprises that succeed with AI share one common trait: they treat it as both a technology and a transformation journey. They recognise that scaling AI requires more than data pipelines and models, and it requires a workforce ready to adapt, adopt the technology and partner with intelligent systems.

The re-aggregation of skills and re/up skilling are not just technical phenomena, it’s a cultural and organisational transformation. To adopt AI at scale, enterprises must build the foundations in people, process and mindset.