Contact me
To support complex workflows within a multi-tool asset management platform, I designed an Agentic AI integration as persistent and embedded across all relevant areas of the platform, not as a siloed chatbot. This design was grounded in best practices from AI conversational design, with a strong emphasis on creating a seamless, brand-aligned user experience, and informed by workflows identified in user workshops. The architectural approach enables the AI to maintain fluid, continuous context across sessions, allowing it to surface relevant history, track task progression, and respond with contextual precision.
1. Enhanced operational efficiency
Enable faster, more accurate workflows by surfacing relevant data, task history, and next steps without users needing to search through layers of complexity.
2. Improve data traceability
Automatically log, summarise, and tag conversations against GIS assets and investigations, creating clean audit trails that support compliance in a regulated sector.
3. Promote user adoption
Lower the learning curve with natural language interactions, contextual awareness, and proactive guidance, reducing the need for training and boosting day-to-day usability in a sector used to complex, engineering-heavy Saas.
4. Differentiate the product
Add valuable, future-facing capability as a core platform capability, showcasing product maturity and innovation that influences buying and investment decisions.
Contextual tagging
Every interaction with the AI is automatically summarised, tagged, and pinned to relevant GIS assets, investigations, or planning workbooks. This creates a content-aware log that users can rely on when reviewing asset history or picking up a task in progress. It removes the need to remember what was said or where, making the platform as easy to use as possible. It also enhances operational traceability, which is critical in a heavily regulated sector.
Content aware assistance
When a user selects one or more assets within the GIS map view the AI assistant knows what assets have been selected and gives relevant options and conversation starters, eliminating the need for manual typing, and reducing friction as much as possible.
Natural Language Processing (NLP) enables the assistant to interpret free-form user input, even as the conversation shifts focus mid-stream.
For example, if a user begins discussing 'Asset XYZ' and then mentions 'Asset ABC', the AI detects this shift. It automatically splits and tags the chat accordingly, associating relevant sections with each asset’s metadata.
This allows for:
These small details collectively ensure the AI feels intelligent, helpful, and genuinely aware of the user’s workflow, giving greater confidence in the AI's responses and in the platform as a whole.
A personified name (such as ‘Siri’ or ‘Gemini’) was intentionally avoided in this assistant’s design. Rather than introducing a separate persona, the AI is positioned as a seamless, intelligent aspect of the platform that reinforces the core product’s identity rather than competing with it and confusing potential customers when adoption is the prime focus.
This approach also reflects evolving user expectations, where AI in task-oriented SaaS platforms is increasingly seen as an embedded feature - reliable, integrated, and focused on productivity. Introducing a distinct persona would imply a level of generality or open-ended conversation that isn’t relevant here; this agent is tightly scoped, purpose-built, and designed to deliver value within a specific operational context.
Visually, the embedded, 'omnipresent' nature presented an opportunity to develop a subtle ripple effect identity that was used as pre-loaders while the AI was generating information.
