Salesforce just used AI to save $100 million. Then they used AI to make money from leads they had already given up on.
The second number is the one to remember.
Here’s the story. Over the past year, Salesforce deployed Agentforce – its own AI agent platform – across customer support. In just over twelve months, those agents handled 3 million support conversations. Year-over-year case volume dropped 8% – more than 170,000 fewer cases – even as the customer base grew. Live chat expanded from one language to seven, with plans for fourteen. The annualized cost savings: $100 million.
That alone is one of the cleaner enterprise AI ROI stories in the market. Most companies would put it in the keynote, declare victory, and move on.
Salesforce didn’t.
Like every B2B company on earth, they had a problem they’d long ago given up on solving: a vast inbound funnel of leads no sales rep ever worked. Content downloads. Webinar registrations. Demo requests where someone never picked up the phone. Sales teams quite reasonably prioritize their highest-scoring prospects, so the long tail of lower-priority leads sat untouched. Internally, the sales team called it ‘sawdust’ – too small to be worth a rep’s time, but real demand nonetheless.
Salesforce pointed an AI agent at the sawdust pile.
The agent did what a junior SDR might do, except across hundreds of thousands of leads simultaneously: sent personalized outreach, asked qualifying questions, responded based on context, identified signals of real buying intent, and routed warm prospects to humans.
The result: 3,200+ opportunities influenced. New pipeline. Closed revenue from prospects the company had effectively written off as uneconomical.
Why This Actually Worked
First, AI is the only thing that makes long-tail demand economical to pursue. The math has always been the same for sales orgs: a rep costs a number, and that number forces every lead to clear a quality threshold. AI agents collapse the unit cost of outreach. The threshold moves down. Demand that was always there, but invisible, becomes visible.
Second, the agent didn’t replace anyone. It worked the segment humans would never have reached. The economics aren’t ‘fire the SDR.’ The economics are ‘the SDR now closes the warm ones the AI surfaces.’ That’s an additive model, not a substitution model, and the math is more favorable.
Third – and this is the strategic insight – Salesforce ran AI twice. Once for cost, once for revenue. Most companies stop after the first deployment because the savings already justify the budget. They miss the second-order play: the same automation infrastructure, redirected at growth.
I’m Mike Partners. I founded AiExpert.org because I believe the strategies behind billion-dollar AI deployments should be accessible to every business owner. Here’s how to put this one into practice.
How to Apply This to Your Business
Start by counting your sawdust – pull every dormant contact in your CRM, email tool, or even your inbox. Old quote requests, ghosted leads, trial users who never converted, webinar attendees you never followed up with. Most owners underestimate this pile by 10x, and knowing the size is half the move. Next, set up one AI follow-up sequence by picking the top 1,000 dormant contacts and using an existing AI tool to send personalized re-engagement based on what they originally asked about – don’t generic-blast, make the prompt good, because the personalization is where this stops being spam. Then track conversion against zero, because these leads are worth zero in your forecast right now, and anything you recover is pure upside. One percent conversion on 10,000 contacts equals 100 new customers at essentially no human cost. The biggest AI mistake I see operators make right now is treating it as a cost-cutting story only. The companies pulling away are the ones running it twice – once on cost, once on growth.
Frequently Asked Questions
What is Salesforce’s Agentforce platform?
Agentforce is Salesforce’s AI agent platform deployed across customer support and sales operations. In twelve months, it handled 3 million support conversations, reduced case volume by 8% (170,000+ fewer cases), and expanded live chat from one language to seven, generating $100 million in annualized cost savings.
What does “sawdust leads” mean in Salesforce’s AI strategy?
Sawdust refers to the long tail of lower-priority leads that sit untouched because sales reps reasonably prioritize higher-scoring prospects. These include old content downloads, webinar registrations, and demo requests where nobody followed up. Salesforce used AI to work this pile and generated 3,200+ new opportunities from leads they’d written off.
How did Salesforce use AI for both cost savings and revenue growth?
Salesforce ran AI twice: first for cost reduction ($100 million in support savings), then redirected the same automation infrastructure at revenue growth (3,200+ new opportunities from dormant leads). Most companies stop after cost savings – the second-order revenue play is where the strategic advantage lives.
How can small businesses work their dormant leads with AI?
At AiExpert.org, Mike Partners recommends pulling all dormant contacts from your CRM, selecting the top 1,000, and using AI tools to send personalized re-engagement based on each contact’s original inquiry. Track conversion against zero since these leads are currently worth nothing in your forecast – even 1% conversion on 10,000 contacts means 100 new customers at no human cost.
Does AI lead outreach replace sales development reps?
No. Salesforce’s model is additive, not substitutive. The AI works the segment humans would never have reached – leads below the economic threshold for human outreach. SDRs then close the warm prospects the AI surfaces. This expands the addressable market rather than replacing existing sales capacity.
