Don't believe the hype. But do believe the direction.

Not long ago, OpenAI said something interesting about the pace at which Agentic eCommerce is emerging. After much initial fanfare, a spokesperson said the company was evolving how it approaches commerce in ChatGPT to better meet merchants and users where they are, prioritizing search and product discovery, with the Agentic Commerce Protocol serving as the connective infrastructure across the shopping journey.

That got me thinking about the bigger picture. Every few years, our industry agrees on a technology that will change everything. Sometimes it does. More often, the transformation is real but slower and messier than the consensus predicted — remember cash's supposed death, or the High Street's after Covid. I say this not to dismiss agentic AI — which I think is genuinely significant — but to keep a note of healthy caution in the conversation.

Agentic systems capable of completing a transaction without human involvement are real and advancing fast. But the vision of autonomous agents handling the majority of consumer retail? Still feels further away than the hype suggests. Payments infrastructure, trust frameworks, and simple human preference for control will all act as friction, so the checkout page isn't going anywhere soon for a large share of online commerce.

Here's what I've been confident about throughout: in any ecommerce transaction, whether human or agent-driven, the customer's address has got to be right. When a person reviews their order, there's an informal quality check built in. Agents don't pause to review — so one inaccurate address record isn't one failed delivery, it's every automated transaction that follows, until someone happens to notice.

That conviction is what led us to build GBG Reach, live now at agents.gbg.com as part of GBG for Agents. Reach takes Loqate's address, email and phone verification and makes it available natively to AI agents and agentic workflows: composable, API-native, built on MCP.

It would have been easy to bolt an AI-friendly wrapper onto our existing API and call it done — stand up an MCP endpoint in front of Verify, ship it, claim the agentic box is ticked. We chose not to, deliberately, because that approach misunderstands what an agent actually needs.

A person filling in a checkout form makes a binary decision: accept the address suggestion or don't. An agent making that same call doesn't have the surrounding context a human brings — the glance at the map, the “that doesn't look right” instinct — and it doesn't have a human waiting to review the outcome afterwards either. What it needs instead is something a simple valid/invalid response was never designed to give it: a confidence score it can reason over, and a policy it can apply consistently, at machine speed, without a person in the loop.

That's the real shift. Non-deterministic systems can't be handed deterministic answers and be expected to make good decisions with them. They need to be handed the tools to weigh a decision themselves, inside guardrails we've defined. Reach exists so that when an agent needs trusted address, identity, or contact data, it doesn't just find us — it finds an answer it can actually act on.

Our 3 billion+ annual address validations are a continuously updated record of how addresses actually behave across 240+ countries — at a scale no general-purpose AI can replicate. That knowledge was always valuable. It becomes foundational the moment autonomous systems start acting on it without a human in the loop to catch the mistake.

Our strategy was never a bet on one outcome. We continue to invest in the onboarding, checkout and reverification experience that serves customers today — not defensively, just because it's the smart thing to do. Alongside that, we've built the agentic layer rather than just talked about it: platform integration ambitions with the likes of Shopify, Salesforce and Oracle continue, on top of a foundation — address intelligence an agent can call, decide with, and trust — that's already live.

The platform giants are building address layers into their own ecosystems, some with us. But 240+ countries of coverage, deep local knowledge where generic models fail, and the independence to serve every retailer on every platform without conflict of interest — that's not replicated in a product cycle, or a wrapper. We've navigated market shifts before, always by deepening expertise rather than defending a position, and we're leading through this one the same way. Twenty-five years into location intelligence, I'm genuinely excited to be at the forefront of how this era develops — and to have something live to show for it, not just a roadmap slide.

The hype will settle. The direction won't change.

We're not waiting to find out which one wins. Reach is live now, discover how GBG Reach provides the verified address and contact intelligence agents need to act with confidence.