Thinking Companies

Companies lack a thinking mode.

A process lets a company reuse decisions about how work should be done. People can act without reconsidering every assumption each time. The process at a company can keep running even when the assumptions behind those decisions no longer hold.

AI and agents change the economics of making new and revisiting past decisions. Work that used to be too costly to re-open can now be re-examined as part of day-to-day operations, using what the business learns in real time.

@Konstantine Buhler puts the economic opportunity in The Cognitive Revolution:

Most problems on Earth currently go un-thought-about, because thinking is expensive.

That is a larger opportunity than finishing today's work faster. A company now might investigate a recurring problem that never justified a project, compare more ways to serve a customer, or offer a service whose cost previously made it impractical. Cheaper cognitive work expands what becomes worth attempting. It also makes it easier to produce analysis that nobody needs.

Choosing where to spend that capacity becomes part of management. An important decision may deserve much more investigation; a routine one may need no additional analysis. The useful measure is whether the effort changes a decision or improves a result.

I think of a Thinking Company as an organisation built to make those choices, carry them through and use the results to improve later work. People and agents contribute to that ability. The company gets better at deciding what deserves attention as well as at doing what it has already decided to do.

Designing it starts with a business commitment. What has to become true for a customer or for the company? What context and capabilities does the work require? Who can act, and who answers for the consequences? The arrangement might be a recurring process, a permanent team or a group assembled for one difficult question. There is no reason every answer must resemble a job description.

Consider a business that repeatedly has to find replacement stock because a supplier delivers late. A fast response to a late delivery matters less if sales is still promising a date that purchasing cannot meet. Giving each department an assistant can speed up several tasks while leaving that contradiction intact.

A different process would compare actual deliveries with the assumptions used to make each promise. Agents could connect that record with the cost of emergency purchases and investigate whether the planning assumption still holds. The process owner could then change the assumption, choose a different supplier or accept the existing trade-off. Price, reliability and the cost of holding more stock all affect that choice.

If the evidence supports a longer lead time, the process owner needs to change the dates sales quotes to customers. The team can then check whether deliveries meet those promises. Routine orders can follow the new rule without another approval; unusual commitments may still need someone authorised to make them.

This is where the job bundle becomes less fixed. A job combines knowledge, time, execution and judgment with responsibility and authority. Some of those capabilities can now be supplied separately. That gives us more freedom to design a process, but it leaves us responsible for deciding who can commit the business and who answers for the result.

The thinking mode is the ability to make these judgments part of running the process. Companies already analyse and improve their operations. The opportunity with agents is to reduce the effort needed to do it often enough, and in enough places, that worthwhile questions no longer have to wait for a special initiative. That includes questions whose answers might justify an entirely different service or way of operating.

The same opportunity is opening up the market for services. AI companies can take on work that customers previously did themselves or bought from a services firm. @JulienBek captures that move in Services: The New Software:

A copilot sells the tool. An autopilot sells the work.

A services firm already sells the work. Its opportunity is to change how that work is produced. Whether the provider started in software or services, it still has to answer for the result. Bérenger Teboul-Danguin makes that distinction in Heartcore Insights:

What the client buys is not only the work, but someone who answers for it.

The firm can also become more capable through delivery. @JakeSaper and the @emergencecap team develop this argument in The AI-Native Services Playbook: completed work should improve the ability to perform future work. An expert's correction is particularly valuable when the firm can establish why it was needed and turn the lesson into a method that works again. The next team then has less to rediscover.

That is a different economic prospect from adding people every time demand grows. It depends on tested improvements surviving outside the original assignment. Client information stays within its boundaries; reusable methods can still become better. The gain has to remain after the effort of checking work, running the systems and sustaining quality. A larger archive on its own does not increase delivery capacity.

At The Thinking Company, the process is the unit of transformation. We start with the result a business needs and work through what has to change to achieve it. That can involve decisions and work spread across departments; speeding up one task can simply move the bottleneck. Senior people and agents work together on the research, design and implementation needed to change the whole process. The client should finish with a working process its own team can run and improve.

Runstate is the product behind our bet that TTC can become a firm on the scale of the Big Four with far fewer people. We are building it so agents can perform substantial transformation work, from investigating a business problem to helping put a changed process into operation. Senior people remain responsible for the choices and the result.

Each improvement to Runstate should increase what our teams can deliver across engagements. We can invest in the product to expand the firm’s capacity, with agents taking on more of the work. A client's existing transformation budget could then buy more substantial change, and work previously too expensive to commission could become viable.

The consequence for a client goes beyond the price of a project. If reviewing and changing a process costs less, the company can afford to do it more often as demand, costs and customer expectations change. That gives TTC a market for continuing transformation, supported by the same product we invest in between engagements.