To understand why AI has not yet delivered the productivity leap many expected, it helps to look back at the arrival of electricity in factories.

The popular version of the story says that electricity arrived, factories remained unchanged for thirty years, and productivity did not improve. The central idea is useful, but the claim is too absolute. Electricity delivered earlier gains in lighting, energy use, machine control and safety. What took longer was its broad impact on productivity.

Because at first, the motor changed. The factory did not.

The old system was running on new energy

In factories powered by steam or water, energy came from a central source and was distributed to machines through shafts and belts. The building and the placement of equipment were designed around that system.

When electric motors arrived, many factories kept the existing arrangement. New motors began driving the old transmission system. The technology had changed, but the way work was organised had not.

According to economic historian Paul A. David, electric motors accounted for less than 5% of mechanical drive capacity in US manufacturing in 1899. It took roughly two more decades for electrification to approach 50%, and its visible impact on manufacturing productivity emerged in the early 1920s. Source

The real transformation came when each machine could have its own motor. Once factories no longer had to accommodate heavy shafts and belts, equipment and material flows could be redesigned around production rather than the power source.

Electricity’s real value was not a better motor. It was a factory design that had not been possible before.

We are changing the motor with AI

Today, many companies approach AI by asking, “Which step in our existing process can we add a tool to?”

The email is written faster, but the approval chain remains. The report is produced faster, but the data is still unreliable. A customer request is summarised, but ownership and decision rules remain unclear.

In other words, we replace the steam engine with an electric motor while continuing to run production by the rules of the steam age.

This is not useless. It can save time on individual tasks. But lasting productivity requires new processes, decision rules, human capabilities and organisational design. The economic idea known as the “Productivity J-Curve” explains why the benefits of a general-purpose technology often appear only after these complementary investments have been made. Source

In travel technology, the real measure is time-to-value

Consider the journey of onboarding a hotel or travel agency partner. Commercial terms, contracting, payment checks, product configuration, integration testing and operational readiness may all sit with different teams.

Adding AI to each team can make individual tasks faster. Proposals, contract reviews, technical documentation and meeting summaries can all be produced more quickly.

But if the work still moves sequentially and decisions remain stuck in queues between teams, the partner’s time-to-value may not improve.

Because making every station faster does not make the whole line faster.

Real transformation means redesigning which steps can run in parallel, which checks can be automated, which exceptions require human judgement and who owns the end-to-end outcome.

AI’s value is not that every team produces documents faster. It is that a partner begins creating value earlier, with less friction.

Electricity became transformative when it allowed machines to move within the factory. AI will become transformative when it allows information, decisions and accountability to move within the organisation.

The real question

The central question of an AI strategy should not simply be, “Which tool should we use?”

It should be:

If intelligence is becoming more accessible, why are we still designing work like the old factory?

Is your organisation using AI merely to replace the motor, or to redesign the factory?