Salesforce names its AI agents: a decision guide for SMEs

On September 11, 2026, days ahead of its annual Dreamforce conference (San Francisco, September 15-17), Salesforce unveiled seven named AI agents — Casey, Paige, Carter, Hunter, Marshall, Piper and Fin — as part of its Agentforce 360 platform (Salesforce). Around the same time, Gartner reiterated that by the end of 2026, 40% of enterprise applications will ship with a task-specific AI agent, up from less than 5% in 2025 (Gartner). At Evicron, an AI and custom software studio based in Barcelona, this week gave us a good excuse to answer a question that comes up earlier and earlier in client meetings: does your business need one of these off-the-shelf agents, or a custom-built one?

What Salesforce actually announced

The seven agents cover very different jobs: Casey handles customer service across voice, SMS, WhatsApp and web chat; Paige resolves IT and HR requests inside Slack and internal portals; Carter helps shoppers compare and buy online; Hunter works an outbound sales pipeline over weeks rather than a single conversation; Marshall automates back-office and supply-chain work with an auditable record of every action; Piper qualifies inbound leads from websites and inboxes; and Fin resolves customer-experience workflows across any channel.

Six are already generally available; Hunter remains in pilot, and it’s the first agent built on a new “long-horizon runtime” that lets it pursue a goal over days or weeks instead of one chat session. Alongside the agents, Salesforce introduced the Trusted Enterprise AI Harness, a governance layer aimed at companies already juggling several AI agent platforms at once.

The Gartner number that changes the question

What matters here isn’t just one vendor’s announcement: Gartner has spent over a year warning that 2026 is the year AI agents stop being a proof of concept and start shipping inside everyday business applications, an eightfold jump from 2025. Whether or not it’s Salesforce, your competitors — or your own suppliers — are very likely evaluating one of these agents right now.

Off-the-shelf agent or custom build: how to decide

Before signing anything, answer five concrete questions:

1. Are you already on the vendor’s platform, or would you have to migrate just to get the agent? A Salesforce agent makes immediate sense if your CRM is already Salesforce; if not, the real cost includes the migration, not just the agent’s license fee.
2. Is your process generic, or does it have rules specific to your sector or your operation? Casey and Fin are built for generic customer service. If your booking flow, your time-tracking process or your stock management has quirks no competitor shares, an off-the-shelf agent falls short — or needs expensive configuration to fake it.
3. How does pricing scale with users or conversations? Large platforms typically charge per seat or per interaction: fine if you’re starting small, expensive once your volume grows. A custom build costs more upfront but doesn’t scale linearly with every new employee or customer.
4. What happens if you want to switch providers in two years? Ask explicitly how your data and configuration export before you sign, not after.
5. Does your data carry specific residency or audit requirements? If you work with the public sector, healthcare or financial data, check where the agent’s conversations are actually processed before committing.

There’s no universally right answer: an SME with repetitive customer service that’s already deep in a large vendor’s ecosystem can come out ahead with an off-the-shelf agent. A business built around a process that’s genuinely its own — the thing that actually differentiates it from competitors — usually needs something built for that specific case.

How we approach this at Evicron

Our AI consulting for businesses always starts with that full picture — what you already have, what’s generic and what’s genuinely yours — before we recommend anything, something we cover in more depth in our guide to choosing an AI consultancy. When the diagnosis points to a custom build, our applied AI work puts the agent on top of your existing stack — your CRM, your ERP, your business WhatsApp — without depending on someone else’s seats or product roadmap.

The bottom line

Salesforce has given its agents names and job titles right as Gartner confirms 2026 is the year agentic AI stops being an experiment for most businesses. The question that actually matters for an SME isn’t which agent to pick — it’s whether the problem you’re solving is generic enough that an off-the-shelf agent will do, or specific enough to your business to justify a custom build.

Want an honest read on which option fits your case? Get in touch: the first consultation is free and we reply within 24 hours.