Agentic AI Becomes an Operating Model: Lessons from PepsiCo, Dell, and Accenture at Dreamforce 2025
AI is moving from pilot projects to the core of how companies run. At Salesforce's Dreamforce 2025, three leaders made it clear: by 2026, being "agentic AI-first" won't be a slogan - it will be an operating standard.
PepsiCo set the pace. Dell issued a warning. Accenture called it a leadership test. Here's what they said, and what to do about it.
PepsiCo: Agentic AI-First Across the Entire Business by 2026
PepsiCo's executive vice president and chief strategy and transformation officer, Athina Kanioura, announced the company is on track to operate as an agentic AI-first enterprise by the end of 2026. The shift spans strategy, innovation, and day-to-day operations through Salesforce's Agentforce 360 platform.
"We will be agentic in every part of the business by the end of 2026," Kanioura said. The goal: connect processes end to end - from marketing and sales to service, field, and supply chain - and equip every employee with a growth mindset.
PepsiCo credited Salesforce as a core partner embedding AI across the organization, including Slack for collaboration and AI-driven workflows. This isn't just tooling; it's operating rhythm.
Dell: Become the Disruptor Before You're Replaced
Dell Technologies chair and CEO Michael Dell urged leaders to avoid complacency. The rise of powerful models in late 2022 created a new path to make use of the "dark data" sitting in systems.
His directive to his team was blunt: reset, reimagine, and operate like the company that could put you out of business. Standing still is the biggest risk.
Accenture: Reinvention Is the Leadership Job
Accenture chair and CEO Julie Sweet framed AI transformation as a daily act of reinvention. If leaders can't point to a clear difference between how work was done and how it will be done, they're not moving fast enough.
The expectation is a visible delta in process, speed, quality, and outcomes - not small tweaks. Reinvention is the standard.
What "Agentic AI" Means in Practice
Agentic AI combines autonomous agents, trustworthy data, and integrated workflows to plan, act, learn, and coordinate across systems. It's not a chatbot on top of legacy processes; it's AI plugged into your stack with clear goals and guardrails.
Think: sales agents that qualify, route, and follow up; service agents that resolve, summarize, and escalate; supply agents that forecast, plan, and trigger actions - all with audit trails and metrics.
A Practical Playbook to Get to 2026
- 1) Set the thesis and metrics: Define where AI must move the needle: revenue, margin, cycle time, CSAT, error rate. Tie each to quarterly targets.
- 2) Fix data and identity first: Establish a single view of customers, products, inventory, and policies. Add observability: logs, feedback loops, and event streams.
- 3) Map the first 10-12 agent workflows: Prioritize by value and feasibility across sales, service, field, finance, and supply chain. Scope the system of actions, not just the prompt.
- 4) Build on a trusted platform: Use platforms built for agents, security, and integration. For example, Salesforce Agentforce connects data, apps, and workflows with policy and governance.
- 5) Govern with intent: Define model usage, data access, human-in-the-loop checkpoints, evaluation standards, and incident response. Track cost and quality per task.
- 6) Upskill everyone: Train managers to redesign processes, engineers to ship agent workflows, and frontline teams to work with AI copilots. Consider structured paths like role-based AI training and certifications. Explore AI courses by job role to accelerate adoption.
What This Means for Your Role
- General: Learn to work with AI as a teammate. Expect AI in your daily tools, tasks, and performance goals.
- Management: Redesign KPIs around throughput, accuracy, and cycle time with AI in the loop. Fund platform work, not isolated pilots.
- IT: Prioritize data quality, access control, auditability, and integration. Standardize model endpoints and evaluation methods.
- Development: Shift from features to outcomes. Build agent workflows with clear triggers, actions, recovery paths, and tests.
90-Day Plan to Start Now
- Select two functions (e.g., Sales and Service) and define four high-value agent workflows each.
- Stand up the data contracts, tools, and governance needed to run them safely in production.
- Ship to a real user group, measure impact weekly, and scale what works.
Why This Matters
The signal from Dreamforce 2025 is direct: PepsiCo is moving enterprise-wide, Dell is rebuilding to stay ahead, and Accenture is pushing leaders to reinvent, not iterate. The companies that win will act like disruptors with clear KPIs, system integration, and disciplined execution.
If 2024 was about pilots, 2025-2026 is about operating models. Choose where agents drive measurable outcomes, then build the system to support them.
Related
Your membership also unlocks:
