Onboarding & Capability Discovery Framework

Agentic Onboarding Is Fundamentally Different

Traditional product onboarding teaches users how to navigate an interface. Agentic onboarding must teach something fundamentally different: how to delegate. The user needs to understand what the agent can do, what it cannot do, and how to specify intent effectively. This is not a tutorial - it is the formation of a working relationship.

Mental Models Must Be Accurate

Users form mental models of agent capability from their first interactions. If those models are too optimistic, users will delegate tasks the agent cannot handle, leading to failure and trust erosion. If too pessimistic, users will under-delegate, missing the value proposition. Onboarding must form accurate mental models.

Capability Disclosure Must Be Honest

The agent must honestly communicate what it can and cannot do. Overpromising capability to drive adoption creates a trust debt that compounds with every failure. Honest capability disclosure - including explicit limitations - builds the foundation for calibrated trust.

First Delegation Is a Critical Moment

The first time a user delegates a real task to the agent is the most consequential moment in the relationship. The outcome of that first delegation sets the trust trajectory for everything that follows. Onboarding must prepare the user for a successful first delegation.

Expectations Must Be Set Before Experience

Users should understand what to expect from the agent before they experience it. This includes expected performance levels, typical response times, common limitations, and how to get help when things go wrong. Setting expectations before experience prevents the disappointment that comes from unmet assumptions.

Agent Capability Calibration

In agentic commerce, the consumer needs to understand what the shopping agent can handle: which product categories, which retailers, what price ranges, what negotiation capabilities. Onboarding calibrates these expectations so the consumer delegates tasks the agent can actually fulfil.

First Purchase Guidance

The first agentic purchase is heavily guided: the agent walks the consumer through intent specification, shows its search process, explains its selection criteria, and presents the purchase decision for explicit approval. This over-transparency builds the foundation for future autonomous operation.

Preference Learning Introduction

During onboarding, the agent explains how it will learn the consumer's preferences over time: what data it will collect, how it will use it, and how the consumer can inspect and correct its understanding. This transparency about the learning process builds trust in the personalisation system.

Limitation Honesty in Commerce

The agent honestly communicates what it cannot do: categories it does not cover, retailers it cannot access, negotiations it cannot perform, and price guarantees it cannot make. This honesty prevents the disappointment that comes from discovering limitations through failure.