Beyond the AI hype: the questions organisations should be asking about agents, data, processes and the future of CRM.
The conversation around AI is changing.
For the past few years, businesses have focused on how generative AI can help employees work faster: generating content, summarising information, answering questions or providing recommendations.
The next stage is different.
Salesforce is increasingly focused on the Agentic Enterprise: organisations where AI agents can reason, interact with business systems and execute tasks across workflows, working alongside people rather than simply assisting them. Salesforce's latest Agentic Enterprise Index reports that the average number of activated agents per organisation nearly tripled over the past year, while the average time to create an agent fell by 53%.
But the rise of AI agents raises a more important question than “What can these agents do?”
What should your organisation actually ask them to do?
The answer requires more than choosing a technology. It requires looking at processes, data, people, governance and business outcomes together.
An AI assistant helps a person complete a task.MAn AI agent can potentially complete the task itself.
That distinction sounds simple, but it changes the way organisations need to think about automation.
An agent might qualify a lead, resolve a customer query, update a record, trigger a workflow or coordinate several steps across different systems. Salesforce's own Agentic Enterprise Index describes increasingly complex deployments, from high-volume, task-specific agents to more versatile agents capable of handling multistep workflows and cross-functional business logic.
The opportunity is therefore to ask: Which work should be performed by people, which work can be augmented by AI, and which work could eventually be delegated to agents?
That is a much more strategic conversation.
In our Dreamforce 2026 webinar, three redk experts will unpack the key developments from the perspectives of AI and technology, marketing, and sales and business. Helping you understand what the announcements could actually mean for your organisation.
One of the biggest mistakes organisations can make is starting with the technology.
Before deciding where to deploy an agent, it is worth looking closely at the process itself.
Is it clearly defined? Are the rules understood? Are there unnecessary handoffs? Are exceptions documented? Is ownership clear?
If the answer is no, introducing an autonomous agent may not solve the problem. It may simply make an inefficient process run faster, or create new risks at a much larger scale.
A useful starting point is to identify processes where:
This is where CRM expertise becomes particularly important.
An agent doesn't operate in isolation. It sits inside a process involving data, systems, people, rules and customer interactions.
The quality of the process will ultimately constrain the value of the agent.
An intelligent agent with poor context is still a poorly informed decision-maker.
Salesforce's expansion of Headless 360 reflects this challenge. The platform is increasingly designed to allow authorised agents and applications to access Salesforce data, automation and business logic across different experiences, while maintaining identity, security and governance.
Salesforce is also positioning Data 360 as a foundation for trusted agentic experiences, with its latest Agentic Enterprise Index highlighting the importance of managed data in supporting reliable AI at scale.
For businesses, this means an AI strategy should start with some fairly unglamorous questions:
AI can expose weaknesses in a data strategy very quickly. If customer information is fragmented, outdated or poorly structured, agents will struggle to deliver reliable outcomes.
The future of agentic CRM therefore depends as much on data foundations as it does on AI capabilities.
More autonomy doesn't automatically mean better outcomes.
Some tasks are highly structured and repetitive. Others involve judgement, empathy, risk or significant consequences for the customer or the business.
The right question is: “Where does automation create more value than human involvement?”
Salesforce itself highlights human oversight, governance and responsible AI as important components of the agentic era. Its Trusted AI work covers areas including agent guardrails, auditability, testing, governance and human oversight.
For organisations, this means defining clear boundaries.
An agent may be able to handle a routine customer request independently, while a complex complaint should be escalated to a person. An agent may prepare a sales recommendation, while the final commercial decision remains with the account executive.
Agentic transformation isn't about removing humans from processes. It's about making better decisions about where humans add the most value.
A successful AI initiative shouldn't be judged by how impressive the demo looks. It should be judged by whether it improves a meaningful business outcome.
Depending on the use case, that might mean:
Salesforce's Agentic Enterprise Index is already moving towards this type of measurement through its Agentic Work Unit, which measures a discrete task completed by an AI agent. Salesforce reports that Agentforce output measured through this metric has been growing at a 15% compound monthly growth rate.
But organisations should go one step further.
The question it's whether that work contributes to the outcome the business actually cares about.
An agent that resolves thousands of low-value tasks may be useful. An agent that improves customer retention, accelerates revenue or frees skilled employees to focus on higher-value work may be transformational.
If you want to hear our perspective once the announcements are out, listen to our Dreamforce 2026 webinar with our experts across AI and technology, marketing, and sales and business.
The rise of AI agents also changes the role of CRM.
CRM has traditionally been about giving people access to customer information and helping them manage processes.
The agentic model introduces another layer: software that can use that information and execute parts of those processes on behalf of people.
This makes CRM architecture, data quality and process design even more important.
Organisations need to think about:
Data: What customer context should agents be able to access?
Processes: Which workflows are suitable for agentic execution?
Technology: Which capabilities need to be connected?
Governance: What can agents do, and what requires approval?
People: How will roles and responsibilities change?
Measurement: How will we know whether the investment is working?
This is why agentic transformation shouldn't sit exclusively with IT.
Marketing, sales, customer service, operations and technology teams all have a role in deciding where agents can create genuine value.
The most interesting part of the agentic shift isn't that AI can do more.
It's that organisations now have to rethink how work gets done.
As agents become capable of handling increasingly complex tasks, businesses will need to decide which processes to redesign, which decisions to automate, where humans should remain involved and how the resulting outcomes should be measured.
That is a very different conversation from simply asking which AI feature to switch on.
And it is why we believe the organisations that get the most from agentic AI won't necessarily be those that deploy the most agents.
They will be the ones that are most deliberate about where agents belong.
You don't need to automate everything. A better starting point is to identify a small number of processes where the combination of clear objectives, trusted data, manageable risk and measurable outcomes makes agentic AI genuinely useful.
From there, organisations can test, learn and expand, rather than introducing AI simply because the technology is available.
Dreamforce 2026 will undoubtedly introduce new capabilities and accelerate the conversation around the Agentic Enterprise. But the technology is only one part of the equation.
The more important questions are organisational:
At redk, we believe the path to agentic transformation starts with strategy, processes and data, and only then technology.
