← back
Most Enterprise Agentic Projects Are Doomed, Here's Why — Jess Grogan-Avignon & Jack Wang, Accenture
Takeaway
Enterprise agent adoption requires changing delivery infrastructure and funding models alongside the technology.
Summary
- Accenture argues enterprise approval, governance, and deployment processes still run at human speed even as agents accelerate implementation.
- One example took two weeks to build but another 12 months to reach production because infrastructure, security, data, and application teams had to align.
- The speakers advocate automating operational controls and investing in CI/CD so faster code generation does not simply create downstream queues.
- Rigid business cases assume scope, value, and delivery costs are known upfront; they propose portfolio-style experimentation to discover new AI opportunities.
enterprise-aiorganizational-changedeployment
Original description
Jess Grogan-Avignon and Jack Wang at Accenture built an agentic application in two weeks. Getting it to production took another 12 months. Not because the code was wrong. Because the infrastructure team, the security team, the AI gateway team, the data governance team, and the application team all had to align before anything could ship. That is not a technology problem, and fixing the code does not fix it. The deeper issue is that GitHub averaged 275 million commits per week in 2025 on track for 14 billion by year end, while the approval infrastructure was never designed for that throughput. They name five tensions that predict whether an enterprise agentic project succeeds before it starts: human approval chains need to become executable code rather than longer signoff meetings; finance should back a portfolio of AI bets the way a VC would rather than demanding committed returns from each project; delivery should run on hypothesis driven loops not milestone programs; trust is built by graduating through shadow mode, advisory mode, and controlled autonomy with each step gated by outcome evidence not by project plan completion; and the real moat is not the data in your ERP but the living memory your product builds from real customer signals every day. Speaker info: / jessicaannbiggs / jackxwang