AI is pressuring technology teams to move faster and handle more requests across the organization. Yet many CIO organizations still split understanding the business problem, building the solution, and owning the applications among different teams.
That model made sense when requests usually ended inside one application and software was expensive to produce. With coding agents, approved APIs, and shared components, one capable person can now build and ship smaller cross-system applications that once required several handoffs. Integration, permissions, testing, and adoption still take time, but the analyst-to-developer relay is no longer always necessary.
The Business Applications Team Is About to Change - The main essay. CIO organizations need a blended delivery lead: a domain expert with enough AI-assisted development skill to build, ship, and improve solutions across a workflow. This is not a requirements or coordination role. The lead needs real development capability, plus the judgment to choose between ordinary automation, an AI-assisted workflow, and an agent, and to decide whether the work is worth doing at all.
Product and engineering are already converging as faster code production makes customer context, prioritization, and product judgment more important. Corporate technology teams face the same shift. Some analysts can learn to build production applications, while some developers can learn a business domain deeply enough to make informed product decisions. One person can then stay with a problem from the first conversation through release and improvement.
Organize that work around the workflow rather than one application. Employee onboarding, for example, crosses HR, identity, the intranet, training, service tickets, and payroll. An HR-aligned delivery lead can own the complete solution, bring in platform and security specialists where needed, and measure whether access arrives faster, setup errors fall, and support tickets decline.
Delivery leads should remain inside the CIO organization and follow shared intake, security, testing, release, and support processes. Business leaders own the process goal and adoption. Application, data, and identity owners retain control of official records and access. A central AI Operations function sets standards for evaluation, monitoring, cost, and escalation. The delivery lead owns building, releasing, measuring, and improving the solution.
Also in this issue:
- The Wire - Workplace transcription is turning informal conversations into permanent records, while companies are choosing models according to what each workload is worth.
- Quick Hits - Enterprise agents are failing without current context, data centers are being redesigned for community acceptance, banks are testing with synthetic customers and data, and IBM customers are shifting technology budgets.
- Meanwhile... - ElevenLabs recreated actor Eric Dane's voice as ALS took away his ability to speak, showing the potential of voice restoration when the person or family has clearly chosen its use.
- What I'm Consuming - Measuring useful intelligence per dollar, risk-based AI architecture, token economics, agentic frameworks, and the relationship among NIST, ISO/IEC 42001, and the EU AI Act.
- After Hours - Remarkably Bright Creatures in book and film form, with the novel offering the richer version of Tova, Cameron, and Marcellus's story.
CIOs do not need to redraw the org chart first. Start with one domain, one workflow set, and one person who combines business knowledge with AI-assisted development skills. Give them approved tools, APIs, knowledge sources, clear decision rights, and specialist support, then measure whether the workflow actually improves.