AI Changes the Possibilities, Not the Organisation

There is a comforting idea circulating through business: buy some AI tools, train the staff and wait for productivity to arrive.

It is attractive because it treats transformation as a procurement exercise. Procurement has forms. Transformation has people, politics and the occasional department defending a spreadsheet from 2011.

The more difficult truth is that AI creates little value if an organisation continues to work in exactly the same way.

Julian Phillips, who leads XTG inside AVI-SPL and works on the company's AI strategy, compares the current shift with earlier general-purpose technologies. Electricity did not instantly transform factories. Businesses initially applied electric power to systems designed around steam. The larger gains emerged when managers redesigned factories, processes and roles around what distributed electricity made possible.

AI is likely to demand a similar organisational response, compressed into a much shorter period.

Automating the Wrong Thing

Most organisations contain processes that exist because they existed yesterday. A task passes through several departments, gathers approvals and generates duplicate records. Nobody designed the whole system. It accumulated.

Adding AI may accelerate individual steps without improving the result.

A summary appears faster. A report is drafted automatically. An email is written with impressive politeness. Yet the customer still waits because responsibility remains fragmented and the final decision requires approval from somebody attending a quarterly off-site.

The useful questions are therefore not limited to what AI can automate. Leaders should ask:

• What outcome is the process meant to produce?
• Which tasks genuinely require human judgement?
• Where do delays and errors enter the system?
• Which steps could disappear entirely?
• What new service becomes possible when the cost of analysis or creation falls?

The last question is especially important. Efficiency is valuable, but widely available tools rarely produce a lasting advantage if every competitor uses them in the same way.

Protect the New from the Old

Phillips learned another useful lesson while building specialist propositions inside larger businesses: new ventures often need protection from the operating rhythm of the core company.

A sales organisation rewarded for today's established products will naturally focus on those products. It may support innovation warmly, provided innovation does not interfere with the quarter.

This informed the creation of XTG as a specialist experiential-technology group. Its work can require early collaboration with architects, corporate real estate firms, content creators and clients planning major projects years in advance. That is a different commercial model from responding to a conventional integration opportunity.

The specialist group gains scale and delivery capability from the larger company. In return, it develops expertise and opportunities the core business might not create alone.

AI initiatives may need the same structure: a focused team with authority to challenge current processes, access to operating expertise and responsibility for commercial outcomes.

This should not become an innovation theatre company furnished mainly with beanbags. The team must eventually move useful changes into the business.

Get Big, Get Niche, or Get Out

Phillips has long advised AV integrators to get big, get niche, or get out.

The phrase applies to many fragmented industries.

Large providers can offer geographic reach, broad services and investment capacity. Specialists can dominate a valuable category through expertise and focus. Medium-sized generalists risk being neither scalable nor distinctive.

AI may intensify that pressure. Large companies can distribute investment across more operations and customers. Niche firms can use AI to deepen scarce knowledge and improve specialist delivery. Generic providers may simply produce generic work more efficiently.

That is useful, but not necessarily defensible.

Acquisition is not an automatic answer. Phillips observes that one plus one in a merger often equals one. Revenue overlaps, customers leave and expected efficiencies prove unexpectedly shy. A strong acquisition should add capability, access or reach that would be difficult to build organically.

Scale without strategic purpose is just additional administration.

The Human Work Starts Now

AI strategy is ultimately a human redesign project.

Leaders must decide where the technology belongs, what risks are acceptable and how work should change. They must also give employees a credible explanation of what productivity gains are for.

If the entire proposition is that workers should document their jobs so fewer workers will be required, enthusiastic cooperation may be limited.

The more constructive approach is neither fantasy nor panic. Map the work. Find commercially meaningful problems. Build protected capacity to redesign processes. Involve the people who understand how those processes really operate. Then decide where the company intends to be distinct.

AI changes what is possible. It does not choose a strategy, repair a weak business model or remove the need for judgement.

Apparently, humans still have a role. It is simply the inconvenient one.

Further reading

• The Enterprise AI Blueprint (The Economist) - Explains why firms need redesigned workflows, skills and organisational structures to realise returns from AI.
• Generative AI at Work (Stanford Graduate School of Business) - Provides field evidence on how generative AI affects productivity, learning and workers with different experience levels.
• The Productivity J-Curve: How Intangibles Complement General Purpose Technologies (National Bureau of Economic Research) - Examines why technologies such as AI require complementary investment in processes, business models and human capital.
• The Frontiers of Corporate Innovation (The Economist) - Considers why efficiency alone is unlikely to create a durable AI advantage and why firms must pursue genuine innovation.
• Why Some Workers Are Embracing AI While Others Won't Use It, According to a New Gallup Poll (Associated Press) - Explores the practical, ethical and privacy concerns shaping employee adoption of workplace AI.

THE CONVERSATION BEHIND THE ARTICLE

Watch the interview.