Exclusive: Salesforce says Agentic Maturity Model is key to CIO success

Today's CIO faces the task of shifting an organisation's information to artificial intelligence readiness. Salesforce says that the right starting point is to focus on identifying process-driven, manual work that's performed every day and working upwards with a tiered, agent-based AI integration plan.

Shibani Ahuja, SVP of Enterprise IT Strategy at Salesforce, says that integrating agentic AI into an organisation can be successfully implemented through a four-step roadmap that helps effectively scale the technology. It's called the Agentic Maturity Model.

"I've spoken with hundreds of CIOs around the world - from SMBs to multinational enterprises - and the number one question I'm asked is how to move past the AI inertia to AI adoption driving business results. That's why we developed the Agentic Maturity Model," says Ahuja. "It helps organize agentic AI use cases by maturity so they can start small while dreaming big–without having to re-architect the entire IT stack every time. It also helps CIOs understand the incremental technology lift necessary to scale over time, so they don't over invest at the beginning."

According to Salesforce, 84 per cent of CIOs believe AI will be as transformational to business as the internet. That's a strong consensus, but deployment is uneven. Many organisations have dipped their toes into chatbots and co-pilots. Still, few know how to think about scale, governance, risk, or orchestrated agents that collaborate with humans (or with each other) across different systems.

Salesforce's model attempts to address this problem by outlining progression over four levels, each representing a deeper stage of agentic intelligence and increasing autonomy. These stages are not only technical, but organisational, requiring corresponding advances in data readiness and security. It also outlines human–machine and machine-to-machine collaboration.

The first level involves information retrieval agents that can extract data and provide contextual suggestions based on that information. For example, a customer support system recommending articles in response to an inquiry. While the agent assists the human, this stage marks a step forward in AI reasoning and interaction, paving the way for more autonomous functionality.

"Focus on identifying process-driven, manual work that's performed every day and start there. Sometimes these are the simple and mundane customer service requests like basic inquiries or password resets. This way you have not only removed a task your teams would rather not do, and you've demonstrated that agents can be a benefit, helping build confidence in their use among teams," says Ahuja.

The second level of the model introduces basic orchestration within a single business domain. Here, AI agents not only provide recommendations but also take actions such as automating structured workflows. This could include scheduling meetings or triggering emails based on internal rules and calendars. Although these agents remain confined to one business function, they signal the beginning of autonomy.

The third level represents complex orchestration across multiple domains. Agents at this stage operate beyond individual silos, integrating data and processes across sales, service, marketing, finance, and other departments. 

Finally, at the fourth level, multiple agents can collaborate to solve complex problems, respond to customer behaviour in real time, or generate insights that span entire business ecosystems. Salesforce states that, at this time, very few enterprises have yet achieved this highest level of maturity. 

"One of the biggest pitfalls I see is leading with technology instead of business outcomes. Without a clear problem to solve, AI projects are more likely to stall or have minimal impact," says Ahuja. "I like to remind business leaders that AI only creates value when it's woven into the way work actually gets done."

The complexity is considerable, but the potential payoff could be significant. According to Salesforce, enterprises that reach this stage can unlock dramatic improvements in customer experience, employee productivity, and decision-making speed.

For Ahuja, among all else, she thinks CIOs need to keep a clear plan and transparency with agentic implementation.

"CIOs should be asking: Are my AI initiatives isolated pilots or are they connected to strategic priorities? Do we have clean, accessible data? Can team members explain and trust AI-based decisions? Those answers will help determine where the business is today and what to focus on next."