Product
Julian Thorne
When chat is the wrong AI interface
Julian ThorneThe industry's obsession with chat interfaces is fading. Learn why structured UI elements often provide a superior user experience for complex AI tasks.
Woodland Hills
Julian Thorne
Tata Consultancy / LLM Applications Lead
Writes logic for document processing pipelines and fine-tunes prompt behaviors for automated search and retrieval solutions.
julian.thorne@asterlane.onlineRecent articles
9 published pieces in the Aster Lane AI archive.
ProductJulian Thorne
The UX cost of magical AI features
Julian ThorneMasking the complexity of AI behind 'magic' buttons creates brittle user experiences. Transparency in AI processes is proving to be a stronger driver of long-term retention.
ResearchJulian Thorne
How to compare open and hosted models pragmatically
Julian ThorneThe debate between open-weights and proprietary API models is often ideological. A pragmatic research approach focuses on total cost of ownership, privacy, and fine-tuning potential.
InfrastructureJulian Thorne
Designing fallbacks for AI service outages
Julian ThorneWhen the API goes down, your product shouldn't. Discover the architectural patterns required to design seamless fallbacks for unpredictable AI service outages.
SecurityJulian Thorne
How to design audit trails for AI decisions
Julian ThorneBlack-box AI decisions are unacceptable in regulated industries. Learn how to design comprehensive audit trails that document exactly how and why a model arrived at an output.
AutomationJulian Thorne
Designing approvals for agentic workflows
Julian ThorneAgentic workflows require unique approval structures. Discover how to design asynchronous approval mechanisms that keep humans in the loop without causing bottlenecks.
DataJulian Thorne
When to delete content from an AI knowledge base
Julian ThorneHoarding old data destroys AI accuracy. Knowing when to aggressively delete or archive stale content is a critical skill for modern data management.
MeasurementJulian Thorne
Why exception rates belong in every AI ROI model
Julian ThorneModels fail, and handling those failures costs money. Incorporating exception rates into your ROI calculations provides a realistic picture of automated workflows.
SecurityJulian Thorne
Mapping tool permissions before an AI launch
Julian ThorneAI agents with unchecked API access are ticking time bombs. Learn how to rigorously map and restrict tool permissions to secure your autonomous workflows.