Now who owns your brand’s front door? Unpacking the realities of personal consumer AI agents

Ask an AI chatbot to find the coffee pods you like and it will show you where to buy them. Ask a personal AI agent to do the same and it will find them, buy them and arrange delivery for you.

Unlike chatbots that answer questions, personal AI agents carry out actions on their user’s behalf, from making purchases and booking travel to managing subscriptions and chasing refunds.

While AI agents are still in their infancy, adoption is accelerating quickly. Research by Bain & Co estimates that by 2030, fully agentic commerce could reach $300 billion to $500 billion in revenues in the US, representing up to one-quarter of total e-commerce. A Mastercard report predicts that more than 1 in 10 people will routinely use AI agents to shop and pay by 2030.

First, a note of caution. The laws needed to regulate personal consumer AI agents lag behind the technology, especially in sensitive sectors such as finance and healthcare.

Right now, consumers in many jurisdictions are entitled by law to demand a human representative at any stage. Regulation is particularly tight in Europe; in Spain, for instance, the government has expanded its consumer protection laws. In addition, in many jurisdictions and circumstances, only a human can use their own personal data, so for example an AI agent dealing with your bank on your behalf is – for now – simply not realistic.

Permitting a digital agent to act legally for a person would require a clear form of delegated authority: a mandate comparable to a power of attorney, or another viable legal route. 

Meanwhile, let’s go back to that coffee pod purchase.

How a consumer instructs their AI agent can result in two very different outcomes for brands. If they name a brand (“Reorder my Nespresso pods”), that brand still gets the sale. On the other hand, if they ask for the cheapest option (“Get me espresso pods under 2 euros”), the agent could buy an alternative without the consumer ever seeing the supermarket or coffee company.

That’s one of the biggest challenges now facing companies. Personal agents aren’t persuaded by the appeal of a great brand story, landing page or video. What they care about is price, availability, fulfilment, latency and ranking signals.

For commodity items, that may be great. But what happens to a business if the consumer never interacts with its storefront? Big brands have spent millions on distinctive and engaging customer experiences. Agents that buy without seeing these could usher in the success of faceless brands. 

So far, broadly speaking, brands’ responses to the arrival of personal agents fall into three categories.

The digital giants. Large companies have the scale to block external agents from their retail sites and launch their own proprietary shopping assistants, keeping customers in direct contact with their brand. As always, things are evolving. It will be interesting to watch the progress of Amazon’s lawsuit in the US to block a third-party shopping agent, now that the ruling has been appealed.

Large retailers. Retailers such as Walmart have taken a hybrid approach by integrating their own shopping agents into mainstream consumer-facing agentic AI platforms such as OpenAI (ChatGPT agent), Anthropic (Claude), Muse (Meta) and Google (Gemini Spark) as well as new contenders like Instinct, to make sure that consumers find them and are guided to their sites.

Everyone else. Other companies rely on consumers finding them via the mainstream agentic AI platforms. In these cases, brands must make their differentiators legible to the agent. Key attributes such as sustainability, packaging, product quality, provenance and supply-chain credentials need to be structured so that agents can understand, compare and act on them.

The brands we work with at Konecta may fall into any one of these three categories. However, in our experience, every company has the same critical priorities: to stay on top of standards and compliance, to seamlessly manage higher volumes of inflow/outflow, to treat this as a customer experience evolution not just a cost-cutting exercise, and to retain the human touch when it’s important. 

Most of the technology building blocks needed to handle consumer personal AI agents already exist. What’s less clear are the interoperability and governance issues. Industry standards and frameworks are emerging. The Personal Agent Protocol, for example, enables businesses to recognize a customer’s agent and shape the relationship, while giving consumers a faster and safer path to outcomes. The Universal Commerce Protocol replaces multiple, custom one-off connections with a single standardized layer, making life easier for consumers while helping businesses to sell anywhere.

At the same time, it is the responsibility of any good CX operation to make sure that usage of AI agents is legal in all relevant jurisdictions, including limits on certain financial loans or purchases. At Konecta, for example, our strategy has been to prepare backend systems for agent-to-agent e-commerce with the relevant authentications, experiment with orchestrating non-sensitive agent-to-agent communications, and continuously monitor AI regulations to ensure ongoing compliance for clients at all times.

One thing to watch out for is increased traffic, with a growing share of inbound contacts coming from customers' agents, not customers. Agents are patient, persistent and precise, so expect more contacts about refunds, cancellations and billing disputes. When consumers no longer have the tedious job of initiating and tracking returns, they will order and return more freely. And agents will actively hunt for duplicate charges, missed credits and unwanted subscriptions.

A future-proofed infrastructure should already be able to handle increased inbound/outbound traffic securely - with clear escalation paths to humans for exceptions. 

All brands now need to be engaging and memorable, not just to people but also to machines. Yet they should treat this as a customer experience shift, not just a cost-cutting opportunity.

Commodity goods such as soap and milk will increasingly be bought automatically. Luxury is different: not everyone wants an AI agent to choose their Cartier watch or a pair of high-end shoes, and those purchases are likely to remain high-value brand experiences.

What’s more, the complexity of CX infrastructures and the importance of good governance make it very unlikely that hybrid human and AI-powered customer services will be disintermediated in the foreseeable future. 

Personal consumer AI agents are still new, so expect more hype and a few hiccups along the way. Robust oversight is critical. Agent-to-agent disputes, for example, will require new monitoring and controls that should be refined continuously as problems surface.

The good news is that in modern AI-powered operations, the infrastructure for consumer AI is already here. Enduringly, however, the human touch is still important – especially at sensitive or stressful times. Only recently, a SurveyMonkey survey found that 89% of US consumers believe companies should always offer the option to speak with a person.

Successful companies will be those that stay agile and compliant while making human moments matter to their customers and their brand.