Every company I speak to is trying to understand what AI means for its business. The conversations start in different places: some leaders want to move faster, others want to reduce costs, and others are trying to understand whether the tools they have already bought are making a difference. But sooner or later, they all arrive at the same question: how do we turn a few promising experiments into a truly different way of working across the company?
We ourselves have gone through the agentic transition in-house and I can tell you that it changes how work is organized, how decisions are made and how teams collaborate. In engineering of course, it changes the relationship between the work a person does and the work that can be delegated to an agent. For me, the most important point is that it also changes what leaders need to know about their organization: If an agent can complete a task in minutes that previously took a team days, you need to understand what that means for delivery, quality, cost and the way you plan your next investment.
Companies are looking for help with all of this. Some are building that expertise internally. Others are turning to consultants, technology vendors or specialist engineering firms. Most will end up working with a combination of them. Each brings something valuable, but they approach the problem from a different starting point:
1) Building the transition yourself
For companies with strong engineering organizations, the first instinct is often to build. Their platform and developer experience teams are experimenting with models, creating internal agents, connecting them to existing systems and developing guidelines for how teams should use them. Some are also bringing this work together in centers of excellence, where teams can share what they learn and establish common practices.
There is a clear advantage to this approach: Nobody understands a company’s systems, constraints and culture better than the people working inside it. An internal team can adapt quickly, focus on the problems that matter most to the business and build capabilities that become part of the organization’s long-term advantage.
The challenge is that the people who are best placed to lead this work are often the same people responsible for keeping everything else running. They need time to experiment, access to the right expertise and a way to share what works across teams. A successful pilot is a beginning, but it does not automatically become an organization-wide transformation. Someone still has to decide what to scale, where to invest and how to know whether the changes are producing better results.
2) Consultants and systems integrators
Consulting firms and systems integrators are helping companies develop AI strategies, select technologies, train their people and manage large-scale rollouts. For a business facing a complex transformation, that support can be valuable. These firms can bring experience from other organizations, help coordinate work across functions and provide the capacity to move faster than an internal team could on its own.
They also help companies deal with a reality that is easy to underestimate: adopting a new tool is relatively simple; changing how thousands of people work is not. The transition involves leadership, processes, security, training and governance. It requires decisions about which teams should move first, what they need in order to succeed and how the rest of the organization can learn from them.
But a transformation program also needs a way to determine whether it is working. Training people and deploying tools are necessary steps. However, they do not, on their own, tell you whether teams are delivering more, whether quality is improving or whether the investment is producing a return. Companies and their consulting partners need a common view of those outcomes so they can adjust the program as they learn.
3) AI vendors and tool adoption
Model providers and AI tool vendors are playing a major role too. They are investing in customer teams, forward-deployed engineers and partner networks to help companies put their technology to work. They bring deep knowledge of what their products can do and can help customers move from experimentation to practical use cases.
That expertise matters since technology is changing so quickly that most companies cannot be expected to understand every new capability on their own. Working closely with a vendor can help a team get started, avoid mistakes and discover applications it might not have considered.
At the same time, every vendor sees the transition through the lens of its own product. That is natural. But companies need to make decisions across an entire organization, where different teams may use different tools, models and combinations of human and agent work. A dashboard that shows how much one tool is being used can be useful, but it only tells part of the story. Leaders also need to understand what that usage produces and how it compares with the other options available to them.
4) Specialist firms
Engineering service providers and specialist firms are also adapting. Many are incorporating agents into the way they build software, helping customers introduce AI into their engineering workflows and developing expertise in areas such as governance, security and implementation.
This shift raises an interesting question for the industry. For years, much of the services market has been organized around the number of people assigned to a project and the hours they work. As agents take on more tasks, that model becomes harder to reconcile with how value is created. If a smaller team can deliver more with the help of agents, the conversation needs to move toward the quality and value of what is delivered.
That change will take time. But I believe it creates an opportunity for service providers and their customers to build a more productive relationship: one where both sides have a clearer understanding of the results they are achieving and can make decisions based on those results.
A shared way to understand progress
From my point of view, the agentic transition will be a shared effort: Internal teams, consultants, vendors and engineering partners all bring different strengths, and most companies will need a combination of them. But whoever helps lead the change, they need a common way to understand whether it is working or not.
To learn what it is delivering, companies need to measure the work produced, its complexity and quality, and what it costs, and then compare those results over time and against the market.
This is where we see Pensero’s role: giving companies and their partners an objective view of performance across human, AI-assisted and agent-driven work, so they can understand what is paying off, decide what to scale and move forward with confidence. As in evolution, the companies that thrive will not necessarily be the biggest or the strongest, but the ones that learn and adapt the fastest.