The Device Play Nobody Asked For
OpenAI president Greg Brockman just announced the company is building "a family of devices" for its AI chatbots. Not a single device. A family. This isn't about creating one breakthrough product — it's about solving a distribution problem that OpenAI doesn't want to admit it has.
The AI model itself isn't the bottleneck anymore. ChatGPT is already powerful enough to handle most tasks users throw at it. The problem? Getting people to actually use it consistently, in the right contexts, at the right moments. Hardware is OpenAI's answer to the question: how do we make AI unavoidable?
Why Software Companies Build Hardware
Software companies turn to hardware for one reason: they've hit a ceiling with pure software distribution. Amazon built Alexa devices because people weren't opening shopping apps enough. Meta built Quest headsets because social media engagement was plateauing. Now OpenAI is building devices because app downloads and web visits aren't sticky enough.
The pattern reveals something important about AI adoption. The barrier isn't capability — it's habit formation. People understand ChatGPT is useful. They just forget to use it when it matters most.
This matters enormously for customer service. If OpenAI — with its massive brand recognition and first-mover advantage — struggles to get consistent usage, what does that tell us about AI adoption in business contexts?
The Real Distribution Channel
Here's what OpenAI is missing: the most effective AI distribution isn't through devices you carry. It's through workflows you already have.
Customer service represents the perfect case study. Companies don't need their support teams to buy new hardware or remember to open an app. They need AI that lives where conversations already happen — in existing chat platforms, email systems, and phone infrastructure.
When a customer sends an email at 2 AM, they don't care what device your support team uses. They care about getting a helpful response. The AI that wins isn't the one with the coolest hardware. It's the one that intercepts the conversation before a human has to.
This is the AI-first approach: asking not "what new behavior can we create?" but "what existing behavior can AI enhance invisibly?"
The Hardware Distraction
OpenAI's device strategy betrays a fundamental misunderstanding of how businesses adopt AI. Hardware creates friction:
- Training overhead: Every new device requires onboarding and training
- Integration complexity: Physical devices need to connect with existing systems
- Maintenance burden: Hardware breaks, needs updates, requires support
- Replacement cycles: Devices become obsolete; software evolves continuously
Compare this to AI that integrates directly into existing workflows. A customer service team using an AI workforce doesn't need to learn new hardware interfaces. They don't need IT to provision devices. The AI simply appears in their existing support queue, handling conversations through familiar channels.
The best technology disappears into the background. OpenAI's hardware push does the opposite — it makes the AI more visible, more physical, more present. That's not evolution. That's regression to a model where technology requires dedicated attention rather than ambient assistance.
What Customer Service Already Knows
Customer service teams learned this lesson years ago. The winners in customer support software weren't the companies that built proprietary hardware terminals. They were the companies that integrated with email, Slack, Zendesk, and every other channel customers already used.
Intercom didn't build a special device for customer conversations. They built software that lives on your website. Zendesk didn't require custom hardware. They created a platform that connects to existing communication channels.
The AI workforce follows the same principle. Businesses don't want AI that requires new infrastructure. They want AI that makes existing infrastructure smarter. An AI agent that answers customer emails doesn't need special hardware. It needs access to your email system, your knowledge base, and the judgment to know when to escalate.
The Device Trap
There's a deeper issue with OpenAI's hardware ambitions: it reveals a company optimizing for consumer novelty rather than business value. Devices make great demo moments. They photograph well. They generate buzz at launch events.
But businesses don't buy AI for the launch moment. They buy it for the ten-thousandth customer conversation, six months after implementation, when nobody's paying attention anymore.
That's where the real test happens. Can the AI handle the edge cases? Does it integrate smoothly enough that teams forget it's there? Does it reduce response times and increase resolution rates without requiring constant human oversight?
These questions don't get answered by hardware design. They get answered by diving deep into the actual problem — understanding customer conversations at a granular level, seeing where humans add unique value and where AI can handle routine patterns.
The Workforce Model
The alternative to OpenAI's device strategy is treating AI as a workforce, not a product. Your human customer service team doesn't use special devices. They use computers, phones, and whatever tools get the job done. Your AI workforce should work the same way.
This means:
- Channel-agnostic operation: The same AI handles chat, email, and phone
- Existing infrastructure: No new hardware required
- Invisible integration: Customers can't tell (and don't care) whether AI or humans respond
- Continuous learning: The AI improves based on real conversations, not firmware updates
When you think of AI as workers rather than devices, the whole approach changes. You don't ask "what hardware do they need?" You ask "what conversations can they handle today, and how do we expand that tomorrow?"
The Future OpenAI Won't Build
OpenAI will launch its devices. Some will sell. Tech enthusiasts will buy them. But six months later, most will sit in drawers next to old Fitbits and unused smart speakers.
The real AI revolution won't come from new hardware. It'll come from AI that integrates so seamlessly into existing workflows that you forget it's there. Customer service shows the path: AI that handles the routine, escalates the complex, and lets humans focus on conversations that actually need human judgment.
That's not a device problem. That's a delegation problem. And delegation doesn't require new hardware. It requires AI smart enough to be trusted with real work.
The companies building that future aren't making devices. They're building AI workforces that make your existing infrastructure smarter, one conversation at a time.