From AI Experimentation to Execution: What Salesforce Leaders Need to Get Right Now
Artificial intelligence is entering a more serious phase. The conversation is no longer only about pilots, prompts, demos, or productivity experiments. It is now about execution: measurable business outcomes, governed automation, trusted data, redesigned work, and enterprise adoption.
Slalom’s latest AI research captures the problem clearly: organizations are increasing AI investment, but many are still early in their ability to measure value, scale use cases, modernize operations, and prepare their workforce. The message is simple but important: AI ambition is growing faster than AI execution.
For organizations using Salesforce, this is especially relevant. Salesforce is no longer just a CRM platform where teams manage accounts, opportunities, cases, campaigns, and reports. With Agentforce, Data Cloud, the Einstein Trust Layer, ChatGPT integrations, and Claude as a preferred model for regulated industries, Salesforce is becoming an execution layer for AI-powered work.
The question is no longer: “Can we use AI?”
The better question is: “Are our data, processes, people, and governance ready for AI to act?”
AI Is Moving From Assistant to Agent
The first wave of generative AI in business was largely about assistance. Employees used tools like ChatGPT or Claude to summarize text, draft emails, generate ideas, analyze documents, or accelerate individual productivity.
That still matters, but the next wave is different. Agentic AI is about systems that can understand intent, retrieve business context, reason through a task, and take action within defined guardrails.
This is where Agentforce becomes important.
Agentforce is Salesforce’s AI agent platform. Its value is not simply that it can answer questions. Its value is that it can operate inside the Salesforce ecosystem, where customer data, business processes, permissions, workflows, and audit requirements already exist. In practical terms, an Agentforce agent can help sales, service, marketing, commerce, financial services, and operations teams move from “AI as advice” to “AI as action.”
That distinction matters.
A chatbot may tell a sales representative what to do next. An AI agent can prepare the account brief, identify risks, summarize recent interactions, update CRM records, recommend next steps, and trigger a workflow — provided the organization has designed the process, permissions, data model, and governance correctly.
The Salesforce AI Opportunity: Execution, Not Hype
The biggest opportunity for Salesforce customers is not to add AI on top of old processes. It is to redesign work around trusted customer data and measurable outcomes.
For example:
A sales team should not only ask AI to write better follow-up emails. It should redesign account planning, opportunity qualification, meeting preparation, and pipeline hygiene.
A service team should not only ask AI to summarize cases. It should redesign case intake, triage, knowledge retrieval, escalation, compliance review, and customer follow-up.
A financial services team should not only ask AI to summarize client notes. It should redesign advisor preparation, consent tracking, portfolio review, service workflows, and regulated client communications.
A nonprofit or public sector organization should not only use AI for productivity. It should redesign citizen, donor, client, or beneficiary journeys with better data quality, clearer accountability, and transparent governance.
In other words, AI value comes from process transformation, not tool adoption alone.
ChatGPT and Salesforce: A New User Experience for CRM
One of the most important developments is the growing connection between Salesforce and ChatGPT.
The Salesforce and OpenAI partnership points toward a future where users can access Salesforce context directly through ChatGPT-style conversational experiences. OpenAI’s Agentforce Sales app, for example, is designed to bring CRM context into ChatGPT so sales teams can research deals, prepare meetings, summarize notes, and enrich Salesforce records without constantly switching screens.
This is a major change in user experience.
For years, CRM adoption has been limited by manual data entry, screen navigation, inconsistent updates, and low perceived value for front-line users. If ChatGPT becomes a conversational surface for Salesforce work, the CRM experience becomes less about “where do I click?” and more about “what outcome do I need?”
That creates a major opportunity, but also a major responsibility.
Organizations must define what AI can read, what it can write, what it can recommend, and what still requires human approval. Otherwise, conversational CRM can create the same old problems at greater speed: poor data, inconsistent processes, and unclear accountability.
Claude and Agentforce: Trusted AI for Regulated Industries
Claude’s role in Salesforce is particularly important for regulated industries such as financial services, healthcare, cybersecurity, and life sciences.
Anthropic and Salesforce have expanded their partnership to make Claude available as a preferred model for Agentforce in regulated environments. The significance is not only model performance. It is trust architecture.
In industries where data sensitivity, compliance, explainability, and security are critical, AI cannot simply be “powerful.” It must be governable. It must respect customer data, privacy obligations, internal controls, consent rules, audit trails, and role-based access.
This is where Salesforce has a strategic advantage: AI can be connected to CRM data, metadata, permissions, workflows, and governance controls. Claude can then be used within that enterprise context to support use cases such as advisor meeting preparation, case summarization, financial services workflows, claims support, policy analysis, and regulated customer communications.
For Salesforce professionals, this means the future is not only about knowing prompts. It is about understanding data models, security, consent, process design, user permissions, validation, auditability, and change management.
The Real Barrier: Data and Process Readiness
Slalom’s research highlights a major issue: many organizations are investing in AI, but fewer are measuring ROI or scaling use cases enterprise-wide. That gap is familiar to anyone who has worked on CRM transformation.
AI exposes the same weaknesses that already exist in business systems:
Duplicate customer records.
Incomplete account data.
Unclear ownership.
Poorly documented processes.
Weak governance.
Legacy integrations.
Manual Excel workarounds.
Low user adoption.
Inconsistent reporting.
Unclear KPIs.
AI does not magically solve these problems. In many cases, it amplifies them.
If Salesforce data is fragmented, AI recommendations will be unreliable. If business processes are unclear, agents will automate confusion. If permissions are poorly designed, AI access becomes a risk. If ROI is not defined, leaders will struggle to prove value.
This is why modernization matters.
Data Cloud, integration architecture, Salesforce security, consent management, reporting governance, and workflow redesign are not secondary topics. They are the foundation of successful enterprise AI.
What Salesforce Teams Should Do Now
To move from AI ambition to AI execution, Salesforce leaders should focus on five practical shifts.
First, define measurable AI use cases. Avoid vague goals like “improve productivity.” Instead, define specific outcomes: reduce case handling time, improve first-contact resolution, increase pipeline hygiene, reduce manual reporting, improve advisor preparation, accelerate onboarding, or increase campaign personalization.
Second, modernize the data foundation. AI agents need reliable, permission-aware, and well-structured data. This means investing in data quality, deduplication, integration, Data Cloud strategy, consent management, and clear ownership.
Third, redesign the workflow before automating it. Do not use Agentforce to reproduce broken processes. Map the current state, identify friction, redesign the target state, and only then decide where AI should assist, recommend, or act.
Fourth, govern AI like a business capability. AI governance should include security, privacy, model selection, prompt design, action permissions, audit trails, exception handling, and human approval points.
Fifth, reskill the workforce. Salesforce administrators, business analysts, product owners, architects, and business stakeholders all need new skills. The most valuable professionals will be those who can translate business needs into trusted AI-enabled processes.
The New Role of the Salesforce Professional
AI will not eliminate the need for Salesforce professionals. It will change what strong Salesforce professionals are expected to do.
The next generation of Salesforce value will come from people who can connect business strategy, CRM data, automation, AI agents, governance, and adoption.
A Salesforce Business Analyst will need to document AI-ready user stories, acceptance criteria, risks, controls, and measurable outcomes.
A Salesforce Administrator will need to understand permissions, data access, prompts, flows, knowledge sources, testing, and monitoring.
A Salesforce Architect will need to design secure, scalable, integrated AI operating models.
A Product Owner will need to prioritize AI use cases based on value, feasibility, risk, and adoption.
A Change Manager will need to help employees trust AI without over-trusting it.
The best Salesforce professionals will not be the ones who simply say “we can use Agentforce.” They will be the ones who can answer: “Where should Agentforce create value, what data should it use, what actions should it take, what controls are required, and how will we measure success?”
Conclusion: AI Transformation Is a Leadership Test
AI transformation is not primarily a technology race. It is a readiness test.
Agentforce, ChatGPT, and Claude create powerful new possibilities for Salesforce customers. They can improve productivity, customer experience, decision-making, and operational speed. But they will only deliver meaningful value when organizations modernize their data, redesign their workflows, govern their AI, and prepare their people.
The future of AI on Salesforce will not belong to organizations that run the most pilots.
It will belong to organizations that turn AI into trusted, measurable, enterprise execution.
References
[1] Slalom — AI Research Report: The ambition-execution gap is widening.
[2] Salesforce — Agentforce: The AI Agent Platform.
[3] Salesforce — Einstein Trust Layer and trusted AI guidance.
[4] Salesforce and OpenAI — Strategic partnership connecting Agentforce 360 and ChatGPT.
[5] OpenAI — Agentforce Sales app in ChatGPT.
[6] Anthropic — Salesforce and Anthropic expanded partnership bringing Claude to Agentforce for regulated industries.
[7] Salesforce Developers — Supported models and BYOLLM guidance for Agentforce.
[8] Reuters — Salesforce acquisition of Fin and continued investment in agentic AI.
