Salesforce just saved $100 million with AI. Then it made even more from leads everyone had given up on.
That’s not a headline from a press release. It’s a case study in what enterprise AI actually looks like when it works – and what it means for every business owner reading this.
In 2025, Salesforce deployed its Agentforce AI agents on its customer support portal. In just over a year, those agents handled 3 million support conversations. Support caseload dropped 8% – that’s 170,000 fewer tickets – even as the customer base grew. Salesforce expanded to live, synchronous chat support in seven languages, something it had never achieved in 27 years of operation. The financial result: $100 million in annualized cost savings, with customer satisfaction maintained.
That alone is remarkable. But the more interesting part happened next.
Every large B2B company has what Salesforce internally calls ‘sawdust.’ It’s the massive volume of inbound interest – webinar registrations, content downloads, information requests – that generates leads no human sales rep ever has the bandwidth to follow up on. These leads aren’t worthless. They’re just uneconomical to pursue at human scale.
In 2026, Salesforce deployed an AI agent specifically to work these dormant contacts. The agent sent personalized outreach, asked qualifying questions, identified buying intent signals, and routed genuinely interested prospects to human sales teams. It influenced 3,200+ opportunities and generated actual closed business from a segment of the pipeline that had effectively been written off.
This is the real unlock: AI doesn’t just cut costs. At a certain scale, it creates revenue from demand that was previously invisible.
WHY THIS ACTUALLY WORKED:
First, they started with low-risk, high-volume use cases. Second, they treated AI as a capacity creator, not a headcount reducer. Third, they measured everything.
I’m Mike Partners – entrepreneur, investor, and founder of AiExpert.org. I write these breakdowns because every business deserves access to the strategies that are reshaping entire industries. Here’s how to act on this one.
Your action step this week: Export your contact list or CRM and filter for every lead, past customer, or inquiry from the last 18 months that never received a proper follow-up. Pick 50 of them. Write a short, personalized reactivation email using AI and send it. Track how many respond. That response rate tells you exactly how much revenue is sitting in your “sawdust” pile.
THE SMB PLAYBOOK:
- Find your sawdust – every lead, past customer, or inquiry from the last 18 months that never got a proper follow-up.
- Run AI-powered reactivation using tools like Clay, Instantly, or HubSpot AI.
- Set up human routing immediately – any reply to AI outreach goes to a human within minutes.
Frequently Asked Questions
How did Salesforce save 100 million dollars using AI agents?
Salesforce deployed its Agentforce AI agents on its customer support portal, where they handled 3 million support conversations in just over a year. Support caseload dropped 8% – 170,000 fewer tickets – even as the customer base grew. The agents also enabled live chat support in seven languages for the first time in 27 years of company history. The combined efficiency gains produced $100 million in annualized cost savings while maintaining customer satisfaction scores.
What is the sawdust strategy for finding hidden revenue with AI?
Sawdust refers to the massive volume of leads and contacts – webinar registrations, content downloads, information requests – that generate interest but never get human follow-up because sales teams lack the bandwidth. Mike Partners highlights this as one of the most overlooked revenue opportunities in any business. By deploying AI to send personalized outreach, ask qualifying questions, and identify buying intent, Salesforce generated 3,200+ new opportunities from contacts that had been written off.
What AI tools can small businesses use for lead reactivation?
Tools like Clay, Instantly, and HubSpot AI can handle AI-powered lead reactivation for small businesses. The process involves exporting contacts from the last 18 months that never received proper follow-up, using AI to generate personalized outreach, and setting up immediate human routing for any replies. The key is ensuring that when a dormant lead responds, a real person engages within minutes. More reactivation strategies and tool comparisons are available at AiExpert.org.
How does AI customer support work without hurting customer satisfaction?
Salesforce maintained customer satisfaction scores while reducing support caseload by 8% because their AI agents were deployed for high-volume, predictable support scenarios first – the low-risk use cases where AI performs reliably. Complex or sensitive issues still route to human agents. The key principle is treating AI as a capacity creator that handles volume while freeing humans for high-judgment interactions, not as a blanket replacement for human support.
Can AI really generate revenue from old leads that were never followed up on?
Yes. Salesforce proved this by deploying an AI agent that worked dormant contacts and influenced over 3,200 opportunities that turned into actual closed business. Most businesses have 18 months or more of unworked leads sitting in their CRM. AI makes it economical to reach every single one with personalized outreach, identify which ones still have buying intent, and route those qualified prospects to human sales teams. The revenue was always there – it was just uneconomical to pursue at human scale.
