Ant Group pushes terminal entry points toward an Agent foundation—is it worth betting on?

Last Wednesday afternoon, I was reading investment memos in the office using RayNeo iO glasses. The glasses prompted translation, the mobile app required permission confirmation, and the PC notes weren't fully synced. In those few minutes, one thought popped into my head: when will this thing become as ubiquitous as water and electricity, allowing agents to safely work across devices? The next day, seeing Ant Group upgrade GPASS to "Lingying," my first reaction was that this feels more like a positioning adjustment.

The value of Lingying mainly depends on whether Ant Group can turn payments, identity, risk control, and trusted connections into the ledger for AI terminals. News says Lingying is positioned as an Agent-native operating system and open platform for new terminals like AI glasses, providing foundational capabilities around natural interaction, agent execution, multi-device collaboration, and security/trust, while opening up to chip partners. Natural interaction and multi-device collaboration bring experience; security and trust determine the moat.

I've been using Claude a lot recently, mostly to read news and break down clauses. Using Anthropic models for about four weeks, and LLMs for just one week, but enterprises fear most when models don't know their boundaries. Once an Agent can look up info, book trips, adjust contracts, and issue commands, the problem shifts from generating content to executing actions. Executing actions requires authorization, auditing, and accountability. This happens to be Ant Group's core business. Payment systems handle user authorization, risk scoring, and transaction links daily. Moving this capability to AI terminals makes logical sense.

Valuation logic depends on what it ultimately extracts. If it only provides trusted connections for glasses manufacturers, Lingying is more like middleware with a limited revenue ceiling. If it becomes an Agent-native OS, allowing agents to call services between glasses, phones, and car units, completing identity verification, permission checks, and billing settlement in each call, then it extracts transaction and service fees, making its valuation closer to a platform.

Regarding business models, I'm watching whether chip and terminal manufacturer integration forms a hardware-side standard, whether the Agent runtime environment allows developers to upload task flows and generate usage fees or shares, and whether payment and identity settlement enable agents to complete consumption, insurance, travel, etc., allowing Ant Group to take fees. The path is complete, but risks are obvious. Hardware manufacturers may not be willing to hand over entry points to Ant Group, and phone makers have their own Agent strategies. I've used Huawei ADS 5 for two months, and one feeling is very clear: terminal manufacturers won't easily let external OSes steal the experience, and glasses manufacturers aren't necessarily willing to let external platforms take over user intent.

So Lingying's real competitive moat is trusted execution. AI glasses are close to the body and close to the wallet. Whether users dare to let it act on their behalf depends on whether it can turn every action into a traceable, revocable, and accountable record. I've used RayNeo iO for three weeks. What annoyed me most was unstable device handoffs, too many permission prompts, and opaque operation logs. Only if Lingying turns these fragmented problems into foundational capabilities does it qualify to talk about being a platform.

This matter isn't suitable for short-term hardware shipment-based valuation. AI terminals are still early-stage; the concept phase is easily pushed by news. A steadier view is to observe whether it can make trusted connections the default option for developers. Infrastructure success depends on who developers think of first when facing identity, payment, and audit issues. Office entry points are red oceans; terminal Agent entry points are equally crowded, with phone OSes, car OSes, and glasses OSes all competing. Ant Group's advantage is financial-grade risk control and payment networks; its disadvantage is that the hardware ecosystem isn't in its hands.

What's most noteworthy about Lingying is whether it has turned trust into a measurable asset. When agents order food delivery, buy insurance, or call enterprise systems for users, they all need identity, authorization, logs, and dispute resolution behind them. If Ant Group packages these capabilities into an open platform, the valuation logic shifts from terminal shipment volume to system-level solution capability. I just started touching zHBM five days ago. As hardware upgrades enter vertical integration, valuation looks at the overall solution.

The battle for entry points heats up slowly; the battle for standards is even slower. Ant Group upgrading GPASS to Lingying this time lifts the previous trusted connection onto an Agent-native OS. Next, we'll see if chip partners are willing to connect, if developers are willing to come, and if users dare to hand over actions.