Practical notes from the engineers who build agents, copilots and AI products.
Personalisation sequences lessons or onboarding steps per user based on progress and behaviour, raising completion and activation. Personalised onboarding replaces one fixed path with a next-step model that reads what each user has done, chooses what to show next, and is measured against a control.
Self-serve onboarding with KYC, catalogue and payouts is what lets supply grow without operations headcount. Treat the merchant onboarding flow as a product with an owner and a measured funnel: verify identity and bank details through APIs, tolerate imperfect catalogue data, and show partners what is blocking approval.
Hypercare is daily triage, fast fixes, floor-walking and a defined exit when tickets fall below a threshold. The hypercare period is the month in which adoption is won or lost, so it needs named people, a daily rhythm, a same-day fix path and an exit criterion agreed before go-live rather than a date on a calendar.
Field apps need offline routes, proof of delivery, payments, navigation and a fast sync; anything else is secondary. Technicians judge an app on whether it works in a basement, captures a signature in one tap and never loses a job, so design the five core flows first and keep dashboards and chat for later.
Good copilot UX streams responses, shows confidence and citations, proposes actions before executing and hands off to the normal UI when unsure. Those four patterns decide whether users trust the copilot after the first week. This article covers each pattern, the entry points that drive use, and pre-launch testing.
Service CRM links tickets, SLAs, schedules and field apps so a customer issue moves from call to closure without re-keying. The build centres on the installed asset rather than the contact, runs SLA clocks automatically, and gives technicians an offline-capable app so closure and parts data come from the field.
Features die when they lack context, cannot act or are scoped too broadly; measure weekly active use per account from a beta cohort. The launch spike is not adoption. Adoption is the share of accounts still using the copilot in week six, decided by three design choices before launch and one measurement habit after.
Conversational banking agents handle balances, card controls and disputes with identity checks and policy-gated actions. Read-only requests are answered after authentication; card actions run inside limits the bank sets; disputes are captured fully and handed to people with a summary, never decided by the agent.
A Slack or Teams bot answers metric questions with charts, cutting the data team's ad-hoc request queue to the hard questions. It works when it uses the dashboards' governed metric definitions, respects each asker's permissions even in channels, replies in-thread with chart and query, and hands off when stuck.