A customer contacts Verizon after a connectivity problem. An upgrade campaign may already be in market. A useful AI system would go beyond composing a polished answer. It could interpret the request alongside account history, current service conditions, and campaign context, then place the next sensible action in front of the customer or a care representative.

That precise journey is a constructed scene, not a documented Verizon outcome. But it captures the product direction sketched by Google Cloud and Verizon’s new partnership. The announcement brings Gemini Enterprise, unified data, customer-experience tools, marketing automation, employee workflows, custom business agents, and network operations into one strategic frame. The conversation is only the visible edge.

I keep coming back to the handoff hidden inside that scene. Commercial value appears when a signal changes what happens next. A question becomes a service action. A network event changes a message. A campaign response changes the next offer. Calling this a chatbot would mistake the doorway for the building.

The new product surface is that turn.

The handoff is becoming the product

Workflow placement is the point where a signal becomes an interpretation and that interpretation enters action. The signal might be a customer question, a campaign response, or a network anomaly. The value lies in carrying its meaning into a consequential next step.

Most software presents records and leaves a person to bridge the systems around them. Adding a model to an existing screen can summarize an account or draft a response. Useful. But once the text appears, someone still has to reconnect it to the campaign, the service state, and the operating decision.

At the handoff, AI can be designed around current state, relevant tools, and the destination of its output. This changes the experience from isolated assistance to continuity. The distinction sounds architectural. When it works, a customer feels it as fewer repetitions, more relevant choices, and a service response that remembers the commercial promise around it.

The Verizon announcement matters because it sketches this connected architecture at enterprise scale. Google Cloud describes a common data and AI foundation spanning conversational and multimodal customer experiences, marketing platforms, employee agents, and network intelligence. Breadth is the product claim. Continuity is the more interesting design signal.

“Serving each and every one of our customers by name requires working AI-first at every level.”

Alfonso Villanueva, Verizon chief transformation officer and EVP of Verizon Consumer, in the partnership announcement

A first name in an email is cheap. Continuity is expensive. If the proposed substrate genuinely connects these functions, a service event could shape the next marketing action, a network issue could alter customer communication, and campaign context could help an employee make a more sensible offer. The intelligence matters because of where it arrives and what it can change.

Marketing’s larger opportunity is continuity

Much of marketing’s generative AI adoption has centered on production. Models can create more copy, images, videos, and campaign variants at far greater speed. That capability is real and valuable. A team can also produce dozens of polished assets while losing the customer’s meaning somewhere between discovery, targeting, sales, and service.

The Verizon design points toward a richer move: connecting those stages inside a signal-to-action loop. Discovery and audience response can shape the creative brief. Creative work can reflect current product and customer context. Targeting can account for what happened in service. Follow-through can carry the original promise instead of beginning from a blank screen.

This would change creative work in a useful way. The brief becomes more alive to customer intent. Variants correspond to actual commercial moments rather than abstract personas. Service language and campaign language can respond to the same state. AI stops being confined to the production desk and begins coordinating the customer experience around it.

I previously asked who owns the customer context when agents can orchestrate campaigns. The Verizon signal adds a practical answer: context becomes commercially useful through placement. A customer attribute sitting in a warehouse has potential. Context carried into the next decision has consequence.

Likewise, Letaido placed the marketing dashboard inside a persistent agent loop. This announcement extends the idea beyond the dashboard. When the loop reaches customer care, employee tools, and network operations, marketing becomes part of a wider system that can sense a change and respond coherently.

My read is that AI-native marketing will be defined by continuity of intent across the commercial journey. Generation volume will be abundant. Remembering what the customer meant, and carrying that meaning into the next action, will remain valuable.

Vertical software is following the same design instinct

Telecom provides the primary commercial scene, but two professional software launches reinforce the product pattern.

Thomson Reuters announced Thomson, a proprietary model trained on its domain content and initially placed inside Tabular Analysis for legal professionals. Model ownership attracts attention. Its initial location shows where Thomson Reuters expects the capability to matter: inside a recurring professional task where information must become structured work.

Reveal’s agentic AI suite follows the work from preservation and search through review, fact-finding, and case development. The design moves AI along a sequence of expert decisions rather than confining it to isolated document retrieval.

Neither announcement establishes adoption, autonomy, or economic advantage. Both reveal the same design instinct as the Verizon partnership: place intelligence where information becomes action.

I argued earlier that the agent is not the moat because durable value accumulates in distinctive signal, legitimate access, and a learning loop. What I would add now is a location. Those assets become more defensible when the product occupies the recurring decision boundary where they can shape what happens next.

A launch announcement is a map, not a result

Product diagrams make orchestration look finished. Diagrams have no latency, departmental boundaries, or customers having a strange Tuesday.

The categories need to stay clean. The facts are that Verizon and Google Cloud announced a broad partnership, while Thomson Reuters and Reveal announced products placed inside professional workflows. Their strategic intent is clear. Verizon wants a common AI and data layer across several parts of the enterprise. Thomson Reuters is placing proprietary intelligence inside an existing legal product. Reveal is connecting stages of eDiscovery.

Deployment and outcomes are less uniform. The Verizon release describes an existing contact-center footprint and vendor-reported improvements, then presents the broader connection across marketing, employee work, and network operations as an expansion. The evidence here does not independently establish the resulting economics. The Thomson Reuters and Reveal releases establish launches, not durable customer advantage.

The inference I draw is narrower and still important: product design is moving toward workflow placement. The conditional forecast is that companies able to preserve context across consequential handoffs will gain more from frontier models than companies using the same models as detached assistants.

Placement alone is not a moat. A familiar screen with an AI button can be copied. Placement becomes defensible when recurring use retains context, improves the handoff, and embeds the product in an action customers or professionals already need to complete.

The thing I would watch is whether meaning survives the boundary between functions. If a service event can change the next campaign action, or a network event can change customer communication before frustration compounds, the architecture is doing more than producing output. It is changing the operating moment.

Build around the turn

If I were sketching the product, I would begin one level below the interface, with the repeated sequence of decisions. Find the moment when intent arrives, context changes the interpretation, and the next action carries customer or economic consequence. That transition is where frontier intelligence can become part of the product rather than an ornament attached to it.

For marketing leaders, the implication is focused. The larger opportunity is not a faster content queue. It is the operating loop that connects discovery, creative work, targeting, service, and follow-through without discarding context at every handoff.

A team may not determine which model leads six months from now. It can decide where intelligence enters the work, what context it receives, and whether its output reaches an action. The model may be rented, replaced, or improved. The product surface that teaches intelligence where the work turns is the part worth building.

Which operating loop in your organization, if redesigned around frontier AI, would create the most durable advantage?

Sources

Sources and further reading