Salesforce just found $100 million in savings. Then found something even bigger hiding in their trash pile.
This is the most important AI business story I’ve seen this quarter – and almost nobody’s talking about it the right way.
Here’s what actually happened:
Salesforce deployed their own Agentforce AI platform across their customer support operation. Starting with their help portal, the AI took over common customer questions – conversational, context-aware, and smart enough to escalate when a human was actually needed. Over the course of just over a year, it handled 3 million support conversations. Case volume dropped 8%, representing more than 170,000 fewer cases – even as their customer base grew. For the first time in their 27-year history, Salesforce now offers live chat support in 7 languages. The financial impact: $100 million in annualized cost savings.
That alone is a remarkable story. But then they turned the same technology loose on something even more interesting.
Like every large B2B company, Salesforce had a problem. Millions of inbound contacts were technically “leads” in their CRM. But most never got followed up on. Sales teams focused on the highest-scoring prospects. Everyone else collected like sawdust. Salesforce deployed an AI agent to work those dormant leads autonomously. It sent personalized outreach, asked qualifying questions, identified buying signals, and routed promising ones to human sales reps. The result: more than 3,200 sales opportunities influenced – closed business from leads that had been written off entirely.
Why This Actually Worked
First, Salesforce removed the constraint of human capacity. AI doesn’t have a headcount limit. Second, they sequenced the deployment correctly – operational cost reduction first, then revenue expansion. This gave them internal proof of concept before applying AI to more complex use cases. Third, they used a low-risk, high-upside testing model: the sawdust leads were already written off, so there was nothing to lose.
My name is Mike Partners. I’ve spent years studying how the world’s largest companies deploy AI, and I founded AiExpert.org to bring those lessons to businesses like yours. Here’s where to start.
How to Apply This to Your Business
Audit your “sawdust pile” this week. Go to your CRM and filter for every contact who hasn’t had an interaction in 6+ months. That’s your pile. For most small businesses, it’s hundreds to thousands of contacts. From there, deploy AI-assisted reactivation. Use a tool like Clay, Apollo, or a basic GPT-powered email workflow to send hyper-personalized reactivation sequences to that list. Don’t blast them – let AI personalize each one. Finally, free your humans for proactive work. Once AI handles your reactive volume, your human team can focus on work that builds relationships and closes deals.
The revenue was already there. AI just made it economical to go get it.
Frequently Asked Questions
How did Salesforce save $100 million with AI?
Salesforce deployed their Agentforce AI platform across customer support, handling 3 million conversations with reduced case volume of 8% – more than 170,000 fewer cases. This produced $100 million in annualized cost savings.
What is a sawdust pile in business terms?
A sawdust pile refers to dormant leads and contacts in your CRM that never received follow-up. Salesforce used AI to reactivate these written-off leads and generated over 3,200 new sales opportunities from contacts that had been ignored.
How can small businesses reactivate dormant leads with AI?
Mike Partners recommends filtering your CRM for contacts with no interaction in 6+ months, then using tools like Clay, Apollo, or GPT-powered email workflows to send personalized reactivation sequences.
What is the best order for deploying AI in a business?
Salesforce sequenced their deployment by tackling operational cost reduction first, then revenue expansion. This provided internal proof of concept before applying AI to more complex, revenue-generating use cases.
Can AI really generate revenue from old leads?
Yes. Salesforce proved this by deploying AI agents on leads that had been completely written off. The AI personalized outreach, identified buying signals, and routed promising contacts to human reps. Mike Partners helps businesses implement similar strategies through AiExpert.org.
