When you need an AI-ready CRM
AI in your CRM is only as good as the data, permissions and processes underneath. Most companies come to us when someone is pushing for AI and they want results without breaking what already works.
Sound familiar?
- Your board or a vendor is pushing AI.
- You want a service agent that answers customers.
- Your teams want to use Salesforce from Claude.
- Automations are being generated faster than anyone can review them.
What we deliver
Agentforce agents that work on real processes, the foundations they need, and clear rules about what AI does and what people decide.
- An AI-readiness check: data, architecture, permissions and running costs.
- Agentforce agents: customer-facing agents on a knowledge base, order-status answers once the customer is verified, and employee agents for process help and onboarding.
- Preparing for Salesforce in Claude: a pilot with one task and a small group, an access check first, and the running costs measured.
- Governance: what AI controls, and what stays under human review.
How we introduce AI
We start with one useful task, a small group of users and a pilot we can measure. If it saves more than it costs, we roll it out. If it doesn’t, we stop, and you’ve learned that cheaply.
- Readiness check
- Pick one useful task
- Pilot with a small group
- Measure the time saved against the cost
- Roll out, or stop
An agent is only as good as the data and knowledge base it works from. That’s usually where the real work is.
Mistakes we see
AI can do a lot inside a CRM. These are the problems we see when it runs ahead of the foundations.
- Automations nobody can maintain. Flows generated with AI pile up faster than anyone documents them, and nobody knows what they do.
- A working CRM switched off because an AI pitch said so. Before replacing anything, check what runs on it: integrations, invoicing, reporting. A CRM audit shows exactly that.
- AI that can see more than it should. A field hidden on a page isn’t necessarily restricted. An AI tool may still read it.
Where should AI start in your Salesforce?
Start with a free 20-minute health check. We’ll look at one task you’d like AI to help with, and what your data and permissions need first.
What it looks like in practice
A typical Agentforce use case: customers ask where their order is. The agent answers once the customer is verified, using only the order data it’s meant to see, and passes anything else to a person.
The hard part is rarely the agent itself. It’s the knowledge base, the data behind it and deciding carefully what may be shared outside the company.
Frequently asked questions
Where do we start with AI in Salesforce?
With one everyday task, a small group of users and a pilot you can measure. A readiness check first tells you whether your data and permissions are ready for it.
What does “AI-ready” actually mean for our CRM?
Clean, connected data; clear permissions; documented automations; and someone who owns them. If those are in place, AI tools have something reliable to work with.
What will Agentforce or AI usage cost us?
Agentforce is priced by usage, with Flex Credits. Salesforce also plans to charge for agent requests made through external tools, with rates still to be announced. That’s why we measure real tasks in a pilot before any rollout.
Can our team use Salesforce from Claude safely?
Yes, if permissions are checked first. A field hidden on a page isn’t necessarily restricted, so an AI tool may still see it. We check access before a pilot, not after.
Do we need Data 360 for Agentforce?
It depends on where your data lives. If what the agent needs is already in Salesforce, you may not. We check before recommending it.
Should we replace our CRM with an AI tool?
Usually not. We’ve seen a working CRM switched off because an AI pitch said so. Check what runs on it first: integrations, invoicing, reporting. Then decide what AI should add.