Everyone is a builder now: scaling AI across non-technical teams

On the surface, people wrestle with apprehension: fear of losing a job, fear of losing control over work that is part of their professional identity, and paralysis about where to start in a landscape that reshapes itself every few weeks.

Below the surface, the first movers build. They create bespoke workflows, logic, and automations at speed. Without horizontal governance, distribution, and evaluation, that energy turns into churn and technical debt. Quality and confidence erode. That erosion feeds the apprehension, and the organization stalls short of what is actually possible.

The way through is not a bigger mandate. When I step back and look at how teams get past this, I keep seeing the same three pillars, worked in order: the human element, scalability, and innovation. Each has operational, organizational, and technical dimensions. The order matters. Skip the first and the other two collapse under their own weight.

Pillar one: the human element

If you are trying to scale AI in your enterprise, I promise you are going to run into apprehension in some form. People's internal voices ask, will I have a job, and will I have control over what was once specialized for me. Disciplines that took years or decades to build suddenly face disruption, and people do not know where they stand. Apprehension also shows up as not knowing where to start, because every entry point feels like it might soon be obsolete.

You cannot pave through this with mandates, town halls, or a channel full of links. People need to feel they still have agency over their own future, and a transition to an agentic-first world has to be designed around that. We treated AI as an amplifier of our people, and we were intentional about how we demonstrated it.

So charter a horizontal enablement team whose mandate is to meet people where they are and shepherd them through the transition. They onboard people in large numbers. They run focused team sessions, hold office hours, debug in real time, and work through use cases alongside the people who have ideas but no idea how to start. Their goal is to raise the floor and the ceiling at the same time. They have to be both an expert student of the space and an expert teacher to the people they guide: patient, empathetic, organized, and masterful communicators. These people do not grow on trees, but they exist. Find them, and deputize them.

Translation is one of the most valuable skills in this transition. The people who can take something highly technical and make it feel approachable are some of the best positioned for this future. Treat onboarding as a product, not an event. Invest in it, observe where real users struggle, and turn every friction point into a fix, a guide, or a proactive check. The message worth coming back to is simple: your job is going to change no matter what, and you have the agency to define how. Once people internalize that, and once they have a tool in hand and a person they trust to help when they get stuck, apprehension starts to fade and curiosity takes its place.

This transformation needs one non-technical catalyst: space and authority. Replace committees and decision docs with a roadmap built on trust and purposeful action. That empowers a flat, collaborative group that ignores traditional titles, where roles blur because everyone handles infrastructure, governance, and prototyping at once. It relies on a shared vision and the psychological safety to disagree and be wrong. You can replicate a framework, but you cannot copy the trust or the chemistry. Recognize when you have those human ingredients, and move quickly.

Pillar two: investing in scalability

This is the pillar where most large organizations quietly fail. If you have addressed apprehension, you now have a new problem: hundreds of newly enabled builders, each with their own workflows. No governance for best practice, no way to measure impact at scale, inconsistent sources of truth, and automated workflows that lack the checks and balances of the manual ones they replaced. You have replaced one bottleneck with a thousand small ones.

Scalability requires you to build a platform, and just as importantly an ecosystem, that lets your builders build freely while capturing what they build so it can be shared, reviewed, and elevated. The architecture that works is a central platform with deliberate distribution, where org leaders have full control over their zones. Each team owns its own skills, agents, and knowledge bases, promoted as the source of truth for that domain. Builders work with org leaders and the enablement team to move a workflow that lives on a laptop, or in someone's head, into the shared platform, where it becomes part of the organization's collective intelligence.

The single biggest unlock is letting domain experts build for themselves. A finance manager who knows the reporting cadence will build a better forecast tool than any central team. A marketer who has launched a hundred campaigns knows where the quality gaps are. An analyst who has lived inside the data model knows which tables to trust. Give them the infrastructure, a building framework, and a governance process, then get out of the way. The platform team's job becomes governance and quality, not authorship. That shift, from central authorship to distributed authorship with central stewardship, is the move that changes everything.

The promotion process for workflows, skills, and agents is governed by org leaders, not in a way that stifles invention, but in a way that ensures what gets codified actually represents best practice for the function it lives inside. Make the catalog feel like a product, not a wiki. People should see automations organized by functional zone, not by underlying technology, along with who built each one, when it was last updated, and how often it is used.

Underneath the platform, you need the infrastructure that makes any of this measurable: telemetry, prompt tracking, evaluation frameworks, data layers, and distribution for every kind of user. Without it, you cannot tell whether an agent is making good decisions, whether a skill produces high-quality output, or whether a workflow is saving time or quietly creating new risk. The best thing you can do for a scaling plan is to get the tracking right immediately, then put the numbers and the evaluation engine back in the hands of org leaders so they can optimize their own space. The mental model I keep coming back to is community: best practice flows up from the people doing the work, leaders codify what works, and the horizontal team shepherds the design.

Pillar three: position for innovation

If the human element raises the floor and the ceiling, and scalability builds the shared infrastructure, innovation is about leveraging that durable platform and builder community to capitalize on what comes next. First, you need partnerships that did not exist before. Engineering, security, legal, and functional leaders have to be at the table, because unlocking real agentic capabilities, agents that act on behalf of the business rather than chatbot-level workflows, requires safe infrastructure, considered policy, and a shared understanding of where the boundaries are. The partnerships themselves are part of the platform.

Governance, including risk, legal, and security, should be engaged early in the design of high-risk workflows. Rather than slowing progress, their involvement, supported by pre-check systems, builds safe infrastructure and speeds deployment by focusing human judgment only on the genuinely ambiguous cases.

Second, you need a serious answer to evaluation. How do you objectively evaluate a model, an agent, and even the team that evaluates the agents? How do you set direction in a landscape that reshapes itself every few weeks? These are organizational questions as much as technical ones, and they demand people whose job is to stay nimble and hold a clear point of view on what is working and where to lean in next.

Third, solve problems that are a mile deep, not a mile wide. The horizontal enablement team is essential, but its strength is breadth: scale, durability, education, and governance. That is not the same skill set as living on the bleeding edge. Consider a second, dedicated team deputized to go deep, work in sandboxes, push fully agentic capabilities into real business problems, and stay in continuous contact with how the industry is changing. A forward-deployed team that integrates with functional leaders and pursues specific, high-value problems at the frontier, while the horizontal team keeps the rest of the organization moving forward together. Both are required, and both accelerate the other. Sandboxes only work if people are genuinely trusted to experiment in them. If your innovation team has to go through a six-week review for every prototype, you do not have an innovation team, you have a committee.

Final thoughts: the goldilocks zone

I keep coming back to one picture. On one end of a spectrum is a one-person company built almost entirely on agents, a single founder orchestrating a stack of automated capabilities that a year ago would have needed a team of fifty. Those companies scale exponentially early, then hit a wall, because there are decisions, judgments, relationships, and ambiguities a fully agentic system cannot navigate on its own. They eventually need humans to scale further.

On the other end is an all-human company with no agents. They have all the judgment in the world but cannot scale, because every additional unit of output costs another unit of headcount. The work compounds linearly when it could be compounding exponentially.

In the middle is a goldilocks zone: the intersection of agentic capabilities that execute, self-learn, and operate on behalf of the business, and humans in the loop whose role is increasingly to make decisions, exercise judgment, and shape direction. The work is shifting from doing the task to directing the system that does the task. That is a new role, and helping people feel empowered to take it on is one of the most important things leaders can do right now. What it means for job design, performance, and career growth is genuinely unresolved, and pretending otherwise is not a strategy.

You cannot scale a platform inside an organization paralyzed by apprehension, and you cannot innovate on top of a platform that has no governance or measurement. Enterprise AI adoption is a human challenge first. The technology is the straightforward part; success depends on a willingness to embrace disruption inside organizations designed for stability. There is no universal playbook. Effective adoption needs the space and authority to adapt to a specific culture, set of systems, and politics.

What does transfer are the team structures: curious, scrappy groups with strong relationships and a blend of technical and human skill. Builders should adapt best practices rather than import them, treating adoption as a behavioral shift first. The work is to build the bridge between bottom-up ground truth and top-down executive narrative. Be intentional about it, maybe with a framework like this one, and that is probably how we all get to that goldilocks zone.