What Anthropic and OpenAI reveal about enterprise AI adoption
As OpenAI and Anthropic move deeper into enterprise services, the message is clear: AI value depends less on access to powerful models and more on an organization’s ability to redesign work, build capability, and learn at scale.
Nicholas Massad, AI Practice Lead, and Shiv Kamal, Senior Marketing Consultant
The market catches up to the adoption reality
The news that OpenAI and Anthropic are moving deeper into enterprise AI services is a signal that the market has entered a more serious phase.
For the past two years, much of the enterprise AI conversation has focused on access: access to models, agents, and tools that can reason, write, code, analyze, summarize, and act. But the next phase is about real adoption, and whether organizations are structured for it.
That is the real story behind the latest moves from the major AI labs. Anthropic has announced a new AI services company designed to help bring Claude into core operations. The company says Anthropic applied AI engineers will work alongside the new firm’s engineering team to identify where Claude can have the greatest impact, build custom solutions, and support customers over the long term.
OpenAI is reportedly moving in a similar direction, preparing a new enterprise AI services venture with the aim of creating a new channel for enterprise AI deployment. Both ventures point toward an embedded-partnership engineering model, where more engineering resources are assigned directly to customer implementation.
The gap between AI capability and AI implementation
The timing here matters. They are acknowledgements that enterprise AI transformation has a last mile implementation problem. The last mile is never actually a mile; it’s often the hardest part of the journey.
Agentic AI is becoming more capable, more autonomous, and more deeply integrated into the systems where work happens. But enterprise implementation is not advancing at the same speed.
It is one thing to buy access to a powerful model. It is another to redesign workflows, clarify decision rights, govern data access, manage security risk, align leaders, influence culture, train teams, update performance expectations, and build trust in outputs that may increasingly be produced or supported by agents.
It’s important to remember: the model may be new. The organization around it is not.
Most enterprises are not designed for a world where intelligence can be so easily embedded into workflows, delegated to agents, and scaled across functions. They have existing processes, roles, systems, approval paths, incentives, risk models, and management practices. AI does not simply plug into that environment and create value automatically.
This is why the conversation is shifting from AI adoption to AI absorption. Adoption asks whether people have the tool. Absorption asks whether the organization can turn the tool into a new way of working.
The bottleneck: Organizational readiness
Microsoft’s 2026 Work Trend Index puts data behind what we are already seeing: the constraint is organizational readiness.
The report states that “people are ready” but “the systems around them are not.” It also finds that organizational factors, including culture, manager support, and talent practices, account for more than twice the reported AI impact of individual mindset and behaviour.
Microsoft calls this the Transformation Paradox: workers are ready, but their organizations are not. According to the report, only 19% of AI users are in the “Frontier” zone, where individual AI capability and organizational readiness reinforce each other. Another 10% are in “blocked agency,” where individuals have built strong AI skills but lack the organizational conditions to apply them. Half sit in the “emergent” middle, where both individual practice and organizational conditions are still taking shape.
That “blocked agency” category may be one of the most important findings in the report. It describes a pattern many organizations will recognize: motivated employees experimenting with AI, discovering better ways to work, and then running into old metrics, unclear rules, cautious managers, disconnected systems, or cultures that reward current performance more than reinvention.
Organizational readiness also depends on the foundations that make AI usable in the first place. As we often say at Levio “bad data, bad AI.” For many enterprises, the barrier is fragmented systems, poorly documented processes, inconsistent governance, and information trapped in silos. Even the strongest AI adoption methodology will struggle if the organization’s data environment cannot support reliable, secure, and scalable use cases.
The shift from adoption to embedded partnership
As pressure builds to show results from AI investment, many organizations are trying to move quickly from experimentation to impact. But not every adoption model creates the conditions for lasting transformation.
One common approach is “Buy”. Purchase the platform, announce the rollout, host the training session, and expect value to follow. It can create early excitement, but usage is not the same as transformation. Without a clear connection to workflows, governance, business priorities, and management expectations, AI risks becoming another tool employees are encouraged to use but not truly enabled to operationalize.
Another approach is centralized AI delivery, or “Build”. AI is placed inside a dedicated team, innovation group, or technical function while the broader organization remains at a distance. This can help create expertise and reduce risk, but it can also isolate the transformation from the people who understand the work best. The result is often a portfolio of pilots, but not a redesigned operating model.
The model that Levio practices, and recent moves from Anthropic and OpenAI support, is partner-driven transformation. It is about bringing in a partner that can work inside the organization’s reality: embedding with teams, understanding how work actually happens, connecting business and technology, and helping build the internal capacity to keep adapting.
That distinction matters. AI transformation cannot be fully outsourced because the most valuable knowledge is embedded inside the organization such as the internal knowledge of time, friction, risk, exceptions, trust, and priority.
A strong partner brings engineering depth, delivery discipline, governance models, industry experience, and acceleration. But the organization still must learn. In fact, the right partner makes that learning part of the work.
Embedded partnership is becoming the standarde
This is where Anthropic’s announcement aligns with Levio’s strategy to execution approach. Its stated model is not simply to advise from a distance. A typical engagement begins with a small team working closely with the client to understand where Claude can have the biggest impact, then building systems tailored to that organization’s operations.
Effective AI consulting is consultants embedding with teams, learning the work, redesigning workflows around real constraints, and leaving behind systems that can keep improving after the engagement ends.
This is not a new idea in enterprise transformation, and large organizations have seen mixed results from embedded delivery models before. The difference now has to be incentive alignment. Embedded partnership only creates value when the partner is accountable not just for delivery activity, but for capability transfer: helping the client learn, govern, operate, and improve the system after the engagement ends.
This is the difference between delivering an AI solution and building AI capacity.
What this means for leaders in organizations
For leaders, the message is uncomfortable but clear: buying AI will not be enough, and delegating AI will not be enough. Organizations need to answer practical operating model questions, including where agents should participate in workflows, where human judgment must remain central, how quality standards are defined, how agent performance is reviewed, how workflows are governed, how local wins are scaled, and how managers are expected to support new ways of working.
These are not abstract questions. They require a mix of capabilities that rarely sit in one place: engineering, business analysis, change management, data governance, cybersecurity, process redesign, product thinking, user experience, and executive alignment.
That is why deep partnership matters.
For a firm like Levio, this is the heart of the opportunity. Organizations do not need another layer of AI hype. They need a partner that can work inside the complexity of real enterprises: legacy systems, regulated environments, human workflows, security requirements, public accountability, technical debt, stakeholder alignment, and operational risk.
They need teams that can connect strategy to delivery. They need builders who can sit with business teams, understand how work happens, design responsible AI-enabled workflows, and create the conditions for adoption to become value.
They need implementation partners who understand that AI transformation is not only about the model. It is about the work around the model.
The last mile belongs to learners
OpenAI and Anthropic moving into enterprise services is a validation of what real transformation partners already know: AI value is created at the nexus of technology, people, process, and change.
The next phase of enterprise AI will separate organizations that experiment from organizations that learn. The companies that win will be the ones that can turn AI into institutional capability: governed, repeatable, measurable, and continuously improving.
The last mile of AI will be solved by organizations willing to redesign how they work, and by partners who enable the capability to keep changing.