For many companies, the first phase of enterprise AI was about experimentation. Teams tested chatbots, generated content, and built small pilots to understand what the technology could do. The harder question is now becoming what happens when those experiments need to operate inside the systems a business relies on every day.
That shift changes the role of AI in the enterprise. A model can generate an answer in seconds, but deploying an AI agent that interacts with financial systems, HR platforms, customer data, or software infrastructure requires a different level of engineering, security, and oversight.
Research from Deloitte shows how quickly that transition is taking place. Its 2026 State of AI in the Enterprise report, based on 3,235 business and technology leaders across 24 countries, found that worker access to approved AI tools increased by 50% in 2025. At the same time, only 21% of organizations reported having a mature governance model for agentic AI, highlighting the gap between adoption and operational readiness.
The gap is becoming particularly important as companies move from generative AI toward systems that can take action. Instead of simply producing information, AI agents can be connected to business processes and instructed to complete tasks, make decisions within defined boundaries, or coordinate work across different systems.
That creates an engineering problem as much as an AI problem.
Building AI into the enterprise
The latest move in this direction comes from Anthropic’s Claude Partner Network. Gorilla Logic, an engineering company focused on enterprise technology, has been named a Select Services Partner in the network’s Services Track.
The partnership will focus on bringing Claude and Claude Code into enterprise software development, as well as building agents and automated workflows for business functions such as finance, operations, compliance, HR, and sales.
The distinction matters because the technology is being positioned inside existing enterprise infrastructure rather than as a standalone AI application.
Gorilla Logic plans to use its Construct system, which brings together engineering playbooks, reusable components, accelerators, and tooling developed through two decades of client work. The company says that approach will allow Claude to be introduced into projects using engineering practices that have already been applied in previous engagements.
“Anyone can build an agent. Which makes engineering standards matter more than ever,” said Drew Naukam, CEO of Gorilla Logic. “A lot of the companies we talk to already pay for Claude. What they want is help getting it into the work their teams do every day, in a way their security and IT teams will sign off on.”
As a Select Services Partner, Gorilla Logic will work across two areas. The first is AI enabled product and platform engineering, using Claude across software development to design, build, test, and modernize products and platforms.
The second is enterprise agent and workflow engineering. That work involves connecting AI agents to business systems and workflows, including through integrations such as the Model Context Protocol, or MCP.
The governance challenge behind agentic AI
The move toward autonomous systems is also forcing companies to rethink how they manage AI.
Deloitte’s research found that 74% of surveyed organizations expect to be using AI agents at least moderately by 2027, while only 21% currently report mature governance capabilities for agentic AI. The consulting firm’s research also points to the need for clear boundaries around which decisions agents can make independently, alongside monitoring and audit trails.
That makes the infrastructure around an AI system increasingly important. An enterprise agent may be able to complete a task, but companies still need to determine what information it can access, what actions it can take, when a human needs to intervene, and how its decisions can be reviewed.
Gorilla Logic’s approach is to treat AI as part of a broader engineering environment rather than a separate initiative.
“We don’t treat AI as a standalone initiative,” Naukam said. “We treat it as an extension of the product engineering, platform engineering, quality engineering, and cloud modernization work our clients already depend on us for.”
Anthropic’s partnership with Gorilla Logic reflects that change in priorities. As enterprise AI moves into production, companies will need more than access to capable models. They will need the engineering layer that connects those models to the systems, workflows, and controls already running their businesses.
The next phase of enterprise AI may therefore be less about proving what models can do and more about determining where they can safely do real work.
Disclosure: This article mentions a client of an Espacio portfolio company.
