Gartner predicts that by 2028, 60% of brands will use agentic AI to deliver streamlined one-to-one customer interactions. The forecast, published on 15 January 2026, points to a shift from broad segmentation to individualised, AI-driven engagement at scale.
For marketing and customer-experience leaders, the temptation will be to move fast: train agents on customer data, deploy them across channels and start optimising response rates. The better move is to build governance first.
What one-to-one agentic engagement means
Agentic AI in customer engagement goes beyond personalising an email subject line. It means systems that can hold extended conversations, make recommendations, resolve issues and adapt to context in real time. The interaction can feel genuinely individual because the agent has access to purchase history, browsing behaviour, preferences and prior service interactions.
The business case is strong: faster resolution, higher conversion, lower cost to serve and more consistent experiences. But the same data that makes one-to-one effective also makes it risky.
Why governance has to come first
When a human agent serves a customer, accountability is straightforward. Training, scripts, supervision and escalation paths are well understood. When an AI agent serves a customer, the same accountability has to be engineered in.
Several failure modes are already visible:
- Over-familiarity. An agent that knows too much can feel intrusive. The line between helpful and creepy is thinner than most firms assume.
- Misrepresentation. An agent may present a recommendation as objective when it is shaped by commercial incentives the customer cannot see.
- Hallucinated offers. Agents trained to be helpful can commit to discounts, refunds or policies that do not exist.
- Data misuse. One-to-one engagement requires data integration across systems. Without clear purpose limitation, the same data can be used in ways the customer did not expect.
Each of these is a trust event. A single badly handled interaction can undo years of brand investment.
A governance-first roadmap
Define the agent’s mandate. Be clear about what the agent is for and, just as importantly, what it is not for. A sales agent should not impersonate a service agent. A retention agent should not make promises the fulfilment team cannot keep.
Publish the disclosure standard. Customers should know when they are interacting with AI. The disclosure should be persistent, not buried in terms of service, and it should offer an easy path to a human.
Separate recommendation from commitment. Let agents suggest, explain and guide. Do not let them bind the firm to terms, pricing or policies without human or rules-based approval.
Build a consent and purpose model. Know which data the agent uses, for what purpose and on what basis. One-to-one engagement often pulls from multiple systems; the legal basis for each use may differ.
Test for adversarial outcomes. Red-team the agent with scenarios designed to elicit over-promising, inappropriate disclosure or manipulative behaviour. Treat the results as seriously as any functional bug.
The competitive angle
Gartner’s prediction should be read as a market signal, not just a technology forecast. Brands that deploy customer-engagement agents without governance will win short-term efficiency and lose long-term trust. Brands that get the governance right will turn it into a differentiator: reliable, transparent AI service that customers actually prefer.
The winners will not be the firms with the most capable agents. They will be the firms whose agents customers can rely on.