Skip to content
Ali
← Back to work
AIDesign SystemsDesign Leadership0→1

Building H.AI, Hornbill's AI era

H.AI is Hornbill's move into the AI era: an AI layer built into the platform so users can resolve tickets, add assets and manage articles just by talking to it. I designed it end to end, and it runs on the design system and blueprint I'd built across the products first, which is what makes an AI layer this cohesive possible.

Role
Product Management, Product Design
Team
Engineering, account management, support
Year
2026
Industry
ITSM
Problem
Early AI features, AI insights and AI assist, showed users valued AI, but they lived in scattered places across five products that felt like five different companies. Hornbill needed one AI experience built into the whole platform.
Outcome
H.AI brings AI into the everyday journey so users get work done by talking to the product rather than navigating it. In design and development now.

The outcome and results. Switch to Deep dive for the full process.

What I've accomplished so far

  • Designing H.AI, Hornbill's move into the AI era: an embedded AI layer with insights, an assistant that drafts articles and requests, and a Claude-style conversational interface, so users resolve tickets, add assets and manage articles just by talking to it.
  • Tested with 14 users across 7 high-value clients, including Guy's Hospital, Cisco and Vinci: 12 were really happy, and the other 2 raised critiques that shaped the next phase.
H.AI focused on Knowledge: published, draft, out-of-date and helpfulness stats, suggested questions and suggested actions
H.AI focused on Requests: suggested actions to assign unassigned requests, prevent SLA breaches, group incidents into a problem and close resolved requests
H.AI focused on Assets: suggested actions to reclaim unused licences, plan a laptop refresh, check silent devices and use stock before buying
An H.AI chat preparing two changes to a request, a timeline update and a priority change, each needing approval
An H.AI chat recommending escalation to a major incident, citing the Phishing Response Procedure, with follow-up suggestions
H.AI home in high-contrast mode: a greeting, an Ask H.AI prompt, Focus on Requests, Knowledge and Assets under the prompt, and quick starts like Morning triage and Resolve a request
H.AI home in light mode viewed as an Admin, with quick starts for team workload, SLA risk, knowledge gaps and licence usage

Focused on Knowledge: article health, gaps and quality at a glance, with suggested actions.

1 / 7

That’s the short version. There’s a lot more behind each decision.