The "Digital Intern" mindset: bridging the gap to autonomy

The transition to an autonomous enterprise is often misunderstood as a simple technical switch. While the vast majority of organisations are expanding their AI footprints, Gartner reports that at least 50% of generative AI projects are abandoned after the proof-of-concept phase. This highlights a deeper reality: achieving autonomy is not simply a technology challenge; it is a management challenge that requires organisations to rethink how people and AI systems work together.

 

As I often emphasise to executive teams, technology is only about 40% of what AI delivers - the other 60% is entirely change management. To succeed, organisations must shift their perspective from managing software to managing digital entities. Viewing AI agents as “digital interns”, highly capable, self-learning workers that require onboarding, context, guidance, and human oversight provides a practical roadmap towards autonomy.

 

 

The intern reality

Like a human intern, an AI agent needs the right support to be effective. It needs to understand the business, operate within defined boundaries, and have human oversight to make reliable decisions. The challenge is not just developing the technology but creating the right environment for people and AI to collaborate successfully.

 

The best place to start is with focused tasks that solve real business problems, rather than attempting to automate entire departments at once. For example, finding information in complex documents may take an employee 15 minutes, while an AI agent can complete the same task in seconds. Solutions such as IFS Loops support these practical use cases by connecting digital workers with existing business systems and workflows, helping organisations demonstrate value and build trust.

 

Crucially, AI adoption must be linked to clear operational outcomes. In heavy-asset industries, applying AI workflows to processes such as supplier management or field service activities has helped companies like Kodiak Gas save thousands of hours annually. Demonstrating these measurable outcomes is what moves AI from an experimental IT project into a scalable business capability.

 

 

From software to digital workers

Many enterprise AI initiatives stall because organisations treat AI agents like traditional software: something to purchase, deploy, and leave running in isolation. Moving from pilot projects to production requires a different mindset, one that treats digital workers as capabilities that need management, governance, and integration.

 

Enterprise AI agents must operate within complex business environments, where they interact with operational systems, company processes, and large volumes of information. This requires clear rules, appropriate access controls, and mechanisms for human escalation when decisions fall outside expected boundaries.

 

Building an AI agent is only the beginning. The real challenge is ensuring that it can operate safely and effectively within the organisation. By treating AI as something to manage rather than simply a tool to install, leaders can combine the speed of automation with the judgement and oversight of their people.

 

 

The org chart of 2026

Organisations are moving towards hybrid teams where digital workers handle repetitive, high-volume tasks, while humans focus on strategy, judgement, problem-solving, and innovation. The key question is not simply “AI versus jobs,” but how AI can help people work differently and create new ways of delivering value.

 

Success will depend on a cultural shift. When employees see AI agents as partners that support their work rather than threats to their roles, adoption becomes much easier. By taking on time-consuming analysis and routine activities, AI can help teams focus on higher-value work.

 

This shift also requires organisations to rethink how they measure performance. Instead of focusing only on activity or hours worked, leaders will need to focus on outcomes and how effectively humans and AI work together.

 

Traditional automation was built around fixed workflows: define the steps once, and the system repeats them indefinitely, regardless of whether conditions change. Digital workers need to be judged differently. Rather than measuring success by whether a predefined process was executed, organisations should measure whether the intended outcome was achieved, even as the steps to get there evolve. This is a subtle but important shift: an AI agent isn’t just running yesterday’s workflow faster; it’s continuously working towards a business result, adjusting its approach as context changes.

 

The future of AI adoption will not be defined by flashy demonstrations, but by whether AI becomes a trusted part of everyday operations. When organisations move beyond experimentation and begin managing AI as a core capability, it becomes a fundamental part of how work gets done.

 

---

 

Somya Kapoor is CEO of IFS Loops

 

Main image courtesy of iStockPhoto.com and demaerre