Who benefits from AI: your company or your service partner?
Today, a company’s AI development can be brought to a halt by something quite ordinary: an outsourcing agreement. Juha Kujala challenges leaders to consider whether the way they buy services supports the redesign of work, or whether the resulting improvements and financial benefits end up with the service partner.
“Good idea, but this work has been outsourced. There’s no point in us developing it, at least not for the next three years.”
I hear comments like these increasingly often in conversations with business leaders. We identify a work process that could be redesigned with AI. The technology itself is not the obstacle, and there is no shortage of ideas. Yet development comes to a halt because an outsourcing agreement has locked the work into its current model for years.
I am not suggesting that AI will make outsourcing entirely unnecessary. It will, however, fundamentally change what companies should outsource and what their service agreements need to specify.
Before your next service procurement, it is worth considering three questions.
1. Is your outsourcing agreement preventing the redesign of work?
Finnish companies have traditionally outsourced a great deal of work. In IT services, for example, we buy multi-year support services and other work from partners, even though digital workers can already perform much of it largely independently.
Digital workers will soon be followed by autonomous IT systems. These systems will solve problems and carry out work independently. They will also seek to prevent the same problems from recurring, effectively correcting themselves.
This means that every outsourced stage of work is also a potential AI development project. What matters is who accumulates the intelligence and automation capabilities created through that work. Are they being built outside your own company, leaving you less able to reinvent the business yourself?
One of our Danish clients made a deliberate choice. The company retained an external partner to provide its finance services, but the work was carried out in the client’s own systems and the resulting knowledge remained with the client. This allowed the company to examine the work, reorganise it and transfer tasks from the partner to its own digital workers. Thanks to ticket-based pricing, service costs fell to a fraction of their previous level as the work shifted back to the company.
2. Are you lighting light bulbs or building a power grid?
Many companies are still looking for quick wins from AI. One stage of work is automated and takes less time. But if the work then piles up elsewhere, the overall process has not changed. The bottleneck has simply moved.
A single AI agent can be compared to a light bulb. Edison’s invention was revolutionary, but light bulbs alone did not transform society. The real change began with the power grid, when strings of lights started illuminating entire cities.
The same applies to a digital workforce. I visited a Norwegian organisation with 3,000 employees and a total of 7,000 AI agents in use. At first glance, the figure looked impressive. It soon emerged, however, that the organisation had run a campaign encouraging every employee to build at least one agent. The experiment was undoubtedly useful for learning, but it may not have changed the work itself. Agents could have been performing the same tasks across different parts of the organisation.
What matters is not the number of agents, but what they are used for. Can something built once be used elsewhere? Are the agents merely personal assistants, or are they digital workers that manage entire workflows? And can they work together?
3. Do leaders know what their digital workforce is doing?
Before transferring outsourced tasks to AI, remember that a digital workforce also needs to be managed.
Not every personal assistant requires a heavy centralised governance model. But when a digital worker independently performs work that is important to the business, access rights, data security, costs, legal requirements and human oversight must be defined from the outset.
Digital workers should be built according to a shared model. Common processes and ground rules speed up implementation, as the same checks and solutions do not have to be repeated each time. If a company plans to transfer outsourced work to digital workers, it cannot wait until the outsourcing agreement is about to expire before considering the necessary structures and management. Preparations need to begin now.
When drawing up your next outsourcing agreement, look beyond price and service levels. Can the work be reorganised during the agreement period? Does the intelligence generated through the work accumulate within your company or with the service partner? Can tasks gradually be transferred to your own digital workers, and will your costs fall as the amount of externally purchased work decreases?
Your next outsourcing agreement will determine who redesigns the work – and reaps the benefits of AI.
How should you prepare for the next outsourcing agreement?
AI makes it easier to build, but that makes choosing harder
AI solved one problem, but it exposed another
