#12 Only 6% of companies trust AI agents with core business...here's what they built differently

💰 THE EXECUTION

How AI turns contact centers from cost centers into revenue engines

Martin Kalinov, CMO at Voiso, thinks most contact center leaders are asking the wrong question entirely.

The leaks nobody's measuring:

"Follow-ups that happen hours too late or leads worked out of priority order eat away at conversion rates," Kalinov explains. "Leaders often don't know in real time which conversations are stalling, which queues are backing up, or where agents are struggling until it's too late to do anything about it."

The deeper issue is visibility. "Companies tend to focus on the top of the funnel, while the bottom – exactly where customers are ready to buy – gets ignored, or is expected to sell itself."

Real-world impact: Kalinov pointed to Airpaz, a global online travel agency handling high inbound volumes. Prior to Voiso, a significant share of calls were abandoned before reaching an agent. "After implementing routing through our Flow Builder tool, call abandonment dropped from 50 percent to 10%. Local caller ID increased answer rates five times over. The revenue impact didn't come from AI selling, it came from more calls reaching the right agents faster."

Same team. Better results. No new hires needed — just better infrastructure that eliminates the friction killing conversion rates.

 💼 DELL'S $25 BILLION AI BET

How Dell's CFO uses AI agents to run finance — and built a $25B business from zero

Dell pulled off something almost no company of its size has managed: In two years, it built a $25 billion AI infrastructure business from scratch and posted record revenues of $113.5 billion.

The numbers: Dell did $34 billion in AI-optimized server orders in Q4, fourth-quarter AI server revenue surged 342% to $9 billion, and it's guiding for $50 billion in AI server revenue for fiscal 2027—103% growth year-over-year.

CFO David Kennedy: "The fear of being left behind is becoming more powerful."

Inside finance: "I've started to deploy agents to do reconciliations, do accounting journal entries. We've launched digital twins in our supply chain. We have our own internal sales chat CRM model, which has handed back multiple hours per week to our sales force."

 The cost: Dell's headcount fell 10%—about 11,000 employees—in fiscal 2026. The company spent $569 million in severance.

Kennedy's take: "You're only as good as the data you have. And an agent wants to work 24/7."

💼 THE STRATEGIC IMPERATIVE

AI-first companies will spend 20-30% of revenue on tech …and win anyway?

While most companies are trying to cut costs with AI, BCG's research reveals a counterintuitive truth: AI-first transformation requires increasing IT spend, not decreasing it.

The payoff? Well-positioned AI-first firms can unlock value from both low costs and high revenues, with illustrative P&L impacts showing 15-25% of costs shifting to tech infrastructure in retail banking and 20-30% in consumer packaged goods.

Here's the reality check from GAI Insights: Firms face a common challenge determining how many AI productivity tool licenses to purchase, with options ranging from standalone tools (ChatGPT, Claude Code) to embedded AI (M365 Copilot, AgentForce in Salesforce). Total annual cost per employee equals the license fee plus training costs, which range from $0 to $1,000 depending on frequency and format.

The bottom line is blunt: "Your top talent and new recruits expect access to a secure, enterprise AI chatbot. Yes, IT cost per employee will permanently increase."

But here's what separates winners from losers:

AI-first organizations operate with an AI-first operating model that streamlines operations via reusable AI workflows and reduced duplication, supported by specialized, scalable talent: lean, high-performing teams focused on strategy, judgment, and AI collaboration.

The companies getting this right aren't adding more tools. They're consolidating around platforms that enable AI workflows while eliminating duplication.

🍺 AI IN THE WILD

How one man used Claude to call 3,000 Irish pubs and changed Guinness pricing!

After paying €7.80 for a Guinness in Dublin, Matt Cortland discovered Ireland stopped tracking the price of its national beer in 2011.

So he built an AI agent. Using AI voice platform ElevenLabs, he created Rachel—equipped with a Northern Irish accent—which made more than 3,000 calls across Ireland inquiring about pint prices.

He then used Anthropic's Claude to create the "Guinndex," a living consumer price index for Guinness across Ireland that allows bartenders and drinkers to contribute prices.

The results: Average price: €6.01. Most common: €5.50. Cortland's €7.80 pint was 30% above average.

Real beer impact: One pub owner lowered his Guinness price by 0.40 euros and updated the Guinndex himself.

The human element: Most bartenders didn't realize they were talking to AI. "Listen, they're normally 6.20 [euros], but if you can't afford one, we'll buy you one. We'll look after you," one bartender told the AI.

Cortland's next targets: prescription drug prices in the U.S. or pizza slices in New York City.

🎙️ THE TAKEAWAY

BCG's warning is clear: IT costs will hit 20-30% of revenue in AI-first organizations. The strategic question isn't whether to increase infrastructure spend—it's whether you're consolidating or sprawling.

Contact centers managing 3.9 vendors can't scale AI. Companies layering ChatGPT, Copilot, Claude, and Gemini without governance create execution nightmares. But unified platforms win: Airpaz cut abandonment from 50% to 10% through consolidation, not capabilities.

HBR's governance framework—defined identity, bounded authority, trusted sources, audit trails—separates the 6% who trust agents with core processes from the 94% still piloting. McKinsey's insight completes it: AI-native experiences require rethinking work paradigms, not bolting chatbots onto legacy UX.

The strategic imperative: Consolidation over expansion. Unified platforms over best-of-breed sprawl. Governance frameworks over pilot theater. Your IT costs will permanently increase. The question is whether you're building systems agents can execute within—or adding to the disconnected pile killing 7 hours per week.

That's it for now, talk soon — Avaamo Team