← Back to the blog
AI agents

CRMs now sell ready-made AI agents. When they are enough and when you need your own

Salesforce has launched a line of ready-made agents for sales, service and operations. They handle standard work well. Processes that span several systems still need a different approach.

Central node connected to several cards, with an approval mark, in dark line art on a light background, next to the glass shape with a blue core

Until recently, having an AI agent in your CRM meant building everything from scratch. The big vendors are changing that.

What Salesforce launched

In September, Salesforce introduced a set of "job-ready" agents in Agentforce for sales, service, commerce and internal teams. One of them, an outbound sales agent called Hunter, pursues a goal over weeks, builds a plan and defines when it can act on its own and when it needs the seller's approval. According to the company, it was in pilot, with general availability planned for November 2026.

The market's message is clear: agents are no longer an experiment, they are a catalogue item.

Flow in which a quote request leads the agent to check CRM, ERP and stock, suggest the next action, wait for a person's approval and log everything in the history

When a ready-made agent is enough

  • The process is standard in your industry and lives entirely inside the CRM.
  • The data the agent needs sits with the same vendor.
  • Your team already uses the CRM consistently, with stages and fields filled in.

When it falls short

  • The process spans systems. Follow-up depends on stock in the ERP, payment at the bank and the conversation on WhatsApp.
  • Your CRM is a different one. Many companies use a local CRM, spreadsheets or an in-house system.
  • The rule is yours. Deadlines, discounts and approvals that only make sense in your operation.

What to check in any agent

  1. Clear limits on what it can do on its own.
  2. Human approval for discounts, purchases and sensitive decisions.
  3. A history of every action, with when and why.
  4. A success measure defined before the pilot.

These criteria apply to ready-made and custom agents alike. The difference is being able to connect the agent to the systems your company actually uses.

Agents inside your systems

The AI agents for CRM and ERP we build work inside the systems you already use, through APIs, with per-agent limits, human approval and an auditable history. We start with a pilot in which a person checks every action.