Salesforce releases 7 pre-built AI Agents ready for action!
Salesforce just announced a portfolio of 7 job-ready AI agents, pre-built to actual roles, ready to start work on day one.
ProQuest has spent the last two years getting Australian businesses live on Agentforce. Every rollout taught us the same lesson: the AI was never the hard part. The hard part was everything around it. Configuration. Legacy data. Getting an agent to actually understand how your business runs, not just how to chat.
And with this announcement, Salesforce is closing a big chunk of that gap.
Why we're paying attention (and why you should too)
Salesforce built the agents on Agentic Work Units (AWUs), their measure of real work completed by AI agents in live deployments. The count just passed 7 billion, with 3.2 billion landing in a single quarter alone.
That's the result of years of stress-testing at a scale most companies will take time to reach on their own, now baked into the agents themselves.
As a partner who's been in the room for a lot of those Australian deployments, this is the moment that track record becomes something businesses here can plug into directly.
Meet the new hires
Salesforce released 7 pre-built AI agents, each one built for a specific job:
- Casey (ex-Help Agent) handles customer support within Salesforce across voice, SMS, WhatsApp, and chat, pay-per-resolution
- Fin (ex-Intercom) resolves complex customer service issues, laying over third-party solutions, across every channel, pay-per-resolution as well.
- Hunter runs your sales pipeline, from research to outreach.
- Piper (ex-Qualified) turns website and inbox traffic into qualified sales pipeline.
- Marshall automates back-office and supply chain work, with a full audit trail.
- Paige clears IT and HR requests for your staff.
- Carter helps shoppers find, compare, and buy.
And it's already working. Paige resolves 70% of Autism Queensland's admin requests. Fin closes 79% of Anthropic's conversations with zero human involved.
You still shape them to fit your business, including giving them their own names, but you're no longer starting from zero.
Trained on how businesses work. Not on your data.
Salesforce has talked publicly about training its enterprise AI on simulated workflows, not customer data. The same principle holds here.
Once deployed, every agent operates inside your own business rules, permissions, and security, exactly as they stand today. Nothing bypasses your existing controls. The agent adapts to your business. Your business doesn't hand over its data to train the agent.
For anyone who's sat through a security review before an AI rollout (and we've sat through plenty), that distinction is the whole ballgame.
AI Agents that don't clock off after one chat
Salesforce also gave agents a longer attention span. A new runtime lets them chase a goal over days or weeks, not a single conversation. A sales agent can now work a deal for a month, remember where it left off, adjust as new information lands, and only loop in a human when it genuinely needs one.
Where this puts us, and where it puts you
We're not reading this announcement from the sidelines. We helped build the muscle memory Salesforce is now packaging into these agents, through hundreds of hours in the room with Australian customers getting Agentforce from pilot to production.
What excites us is the speed. Two years of AWU volume means the configuration work we used to spend months on is largely done before we start. Our job shifts from building the agent to tuning it for your business, fast.
We're testing that shift on ourselves before we bring it to you. We're about to become one of the very first Fin partners in Australia and configuring Fin for our own customer support.
The agents are job-ready. Getting them working for your business, that's still where the right partner earns their keep. If you want to get started, let’s have a chat.
