The Hardest Tickets Don’t Text. They Break.
Salesforce just spent $3.6 billion to acquire Fin, the AI agent platform formerly known as Intercom. It’s the biggest AI customer service deal of 2026. Every major platform — Salesforce, Zendesk, ServiceNow, Freshworks — is now running the same playbook: deploy AI agents that can autonomously close the majority of support tickets without a human ever touching them.
They’re right. For most tickets, it works. Billing questions, password resets, policy lookups, refund requests, basic troubleshooting flows — an AI agent handles that at scale, instantly, around the clock, for a fraction of the cost of a human. The ROI is real. The efficiency gains are real. But for physical failures, broken equipment, and installations that won’t initialize — visual support for field service is the missing capability that $3.6 billion doesn’t address.
The 20% That Breaks Everything
The hardest support tickets don’t come in as text. They come in as a machine that won’t start. A device that’s doing something the manual never described. An installation that looks right on paper but is failing in the field. A piece of equipment that a field tech is standing in front of, alone, not sure what they’re looking at.
You can’t resolve those tickets with a chatbot. A knowledge base won’t help either. Even a perfectly trained AI agent that has ingested every support doc your company has ever written can’t close these — because the problem isn’t information. The problem is visibility.
The customer has the answer. It’s right in front of them. They just can’t transmit it through text. And you can’t help them until you can see what they see.
That’s the gap that $3.6 billion in AI infrastructure doesn’t solve.
The Tickets That Cost the Most
Here’s what that 20% looks like on your support team’s dashboard:
- Average handle time: 3–5x higher than a standard ticket
- Escalation rate: near 100%
- First-contact resolution: near 0%
- Truck rolls, on-site dispatches, repeat contacts: expensive
- Customer satisfaction: your lowest scores, consistently
These are the tickets your AI agent flags as “needs human review” and your human agents dread seeing in the queue. They generate the most back-and-forth. Someone says “can you send a photo?” and the customer sends a blurry JPEG that tells you nothing. Eventually the ticket ends in a scheduled callback, a dispatch, or a quietly churning customer six weeks later.
Moreover, the industry’s entire bet on AI is optimizing the 80% while leaving the 20% exactly where it was. That 20% — the physical, visual, “I need to show you something” tickets — is where your highest churn, your highest cost, and your most damage to brand trust is concentrated.
Visual Support for Field Service: The Problem Is Physical, Not Textual
Consider what actually happens when a complex support issue arrives. A homeowner’s HVAC system is making a sound it’s never made before. A retail associate’s point-of-sale device is throwing an error code that isn’t in the documentation. A field installer is at a job site and the equipment isn’t responding to initialization the way it should.
In every one of these cases, the customer or technician knows something is wrong. They can see it, hear it, feel it. Unfortunately, what they can’t do is describe it with enough precision for a remote support agent to diagnose it. “It’s making a weird noise” is not a diagnostic input. “There’s a red light” could mean fifty different things. “It just stopped working” tells you nothing about state.
The resolution path for every one of these tickets is the same: someone with expertise needs to see it. Not read about it. See it. That’s precisely why visual support for field service exists as a category — and why it matters more than ever in an AI-first support stack.
That’s what Viewabo is built for. Live video support — not a video call where both parties are fumbling with Zoom links and waiting on IT approval — but a one-tap session where a support agent can see exactly what the customer sees, in real time, without requiring an app download. The agent sees it, diagnoses it, and guides the customer through the fix. One session. Problem closed. Teams already using this approach are cutting unnecessary truck rolls before they ever happen.
What the Salesforce Deal Actually Tells Us
The Fin acquisition isn’t just a competitive move. It’s a signal that the industry has decided text-based AI is mature enough to bet the company on. Salesforce isn’t buying Fin to experiment — they’re buying it to make it the default resolution layer for hundreds of thousands of businesses on the Salesforce platform.
Interestingly, that’s actually good news for visual support. When AI agents handle 70–80% of tickets automatically, the tickets that remain are, by definition, the harder ones. The edge cases. The issues that required escalation. The moments where something physical went wrong and a human needs to intervene. AI triage will surface those tickets faster and route them more efficiently — but it won’t resolve them. It will hand them off to a human agent who needs to actually understand what’s broken.
The best-in-class support stack in 2026 isn’t AI or humans. Instead, it’s AI agents handling what they’re good at, and humans with the right tools handling what they’re not. Visual support for field service isn’t a workaround for the AI era — it’s what makes human escalation worth it.
Visual Support for Field Service: The Gap Is the Opportunity
Customer service has spent the last two years racing toward automation. Every platform is promising ticket deflection rates, resolution accuracy scores, CSAT improvements from AI. The ROI is real for the easy stuff. However, the industry is also quietly building a problem: when AI handles the routine and fails gracefully on the complex, what happens to the customer standing in front of a broken machine at 9 PM?
They’re still waiting on hold. Their callback is scheduled for next Tuesday. They’re still trying to describe a problem in words that don’t match what they’re seeing. And they churn quietly, six weeks later.
The hardest tickets don’t text. They break. And until someone can actually see what’s breaking, they stay broken.
That’s the problem we’re solving. And it’s the one the $3.6 billion deal didn’t.