AWS: governance and unclear AI value are holding back enterprise deployment, financial sector leads according to Reimagine report
According to AWS (report Reimagine: Turning AI into Value), the main concerns for companies are AI governance and unclear value of deployment. The financial sector leads in adoption thanks to documented rules, while automotive and manufacturing are redesigning processes with agentic AI.
Matthias Patzak, AWS executive in residence, summarized findings from the company report AWS Reimagine: Turning AI into Value (published in September) in an interview for AI Business. According to him, the most common concern among AWS customers is governance and responsible use of AI, followed by the fact that a large share of companies do not see clear value from AI deployment. Financial services are, according to Patzak, the most advanced in adopting and scaling AI because they have their rules and policies well documented – however, AWS points out that simply inserting these documents as context for an agent is not a best practice. A better approach, according to the company, is to build governance as a separate infrastructure or platform into which the agent is deployed and which sets its boundaries.
The automotive industry and manufacturing, especially in Germany, are actively pursuing the redesign of entire business processes using agentic AI, partly in response to price pressure and competition from China. Patzak described specific examples: an expert in a leasing department or at a factory created an agent that was subsequently used by more than 200 colleagues, without the company first having to formally approve a budget. Another example involved the CIO of a large German bank, who took 50 of his most capable people off-site for a week with the task of designing a completely new mortgage process from scratch, because the existing process was slow and complicated.
On the term "AI native," Patzak stated that it means minimizing human involvement in the process and maximizing automation. For orchestrating multiple agents, he said the current best practice consists of deploying a separate orchestration agent that oversees other agents in individual roles with limited knowledge and context – today this orchestration is handled as individual software components that companies and developers have to assemble themselves, though Patzak expects ready-made product solutions to appear soon.
Why it matters
For companies planning to deploy AI agents, the interview shows a concrete path for handling governance – not as a document inserted into the agent's context, but as a separate platform with defined boundaries, which reduces the risk of agents misbehaving and simplifies scaling. At the same time, examples of bottom-up agent creation without formal budgets suggest that AI value can also be generated by individual employees, not only by centrally managed projects.
Two audiences, two different impacts
What this means
For individuals
According to the cases described in the source interview, individual experts can create an agent for their process without having to request a budget, and dozens of colleagues can then use it – digitalization can thus emerge from the bottom up, not only as a top-down corporate project.
For a business
According to AWS, governance and unclear AI value are the main obstacles to enterprise deployment; the financial sector is the most advanced because it has its rules documented and can build them in as infrastructure for agents, while automotive and manufacturing, driven by competitive pressure, are actively redesigning entire business processes using agentic AI.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.