This release introduces a fully upgraded conversational Agent that guides you through gathering master plan inputs, completes the Majlis spatial map animation series, and delivers meaningful stability and accuracy improvements across the platform.
A Reimagined Conversational Agent
We've upgraded the Agent that powers project intake from the ground up. Instead of a static questionnaire, the Agent now works conversationally — walking through each input the master plan requires and engaging with you only where it matters.
A clearer four-step rhythm:
Every part of the conversation now follows a consistent Searching → Analyzing → Finding → Confirming pattern. You can see what the Agent is doing at each moment, follow along with its reasoning, and step in to confirm or adjust when it matters most.
What this means for you:
Less manual data entry — the Agent gathers the inputs your master plan needs through natural conversation, rather than asking you to fill out long forms.
More transparency — at every stage you see what's being searched, what's being analyzed, what's been found, and what's awaiting your confirmation.
Stay in control — confirmation moments are clearly surfaced so you can keep the conversation moving or refine the direction at will.
Majlis Spatial Map Animations — Series Complete
The Majlis spatial map animation series is now fully built out, giving every stage of the conversation a purpose-built visual moment on the map.
Highlights:
A complete set of animation stages that pair each conversational round with a corresponding map narrative — from terrain reveals and regional framing to local-context analysis and the climate envelope.
Public facilities, expanded — education, community, and industry variants are now distinguished on the map with their own visual treatments, alongside a refreshed utility-infrastructure icon set.
New Terrain, Military Legacy, WindRose, and WindSpeed legends surface context-specific information at the right moments.
The Flood Risk legend has been refactored to follow the same animation-directive pattern as the rest of the system — predictable timing, cleaner state.
Legend positioning now adapts to Majlis mode, keeping the layout balanced across viewer modes.
Feasibility & Plan Comparison
Feasibility numbers and the plan-comparison surface both got a meaningful round of accuracy and consistency work.
Water and power calculations in the feasibility view now compute correctly — fixing math errors that previously skewed comparison.
Shared unit-calculation logic between the frontend and the data pipeline ensures the numbers you see on screen exactly match the ones used during plan generation.
The plan comparison table and Feasibility tab have been refactored for clearer rendering and more reliable updates.
Map Stability
A focused round of resilience work makes the map noticeably less prone to edge-case crashes.
The white boundary layer no longer obscures custom layers and 3D buildings — a long-standing z-order issue is now resolved across all map hooks.
Map style loading errors are handled gracefully via a shared safety check, so unmounting a map mid-load no longer throws.
The population layer is wrapped in protective error handling to prevent crashes during style transitions.
A persistent infinite-render issue caused by legacy transport overlay markers has been fixed, and the unused legacy transport layers have been retired.
Other Improvements & Bug Fixes
The deal search dropdown has been tidied up for a cleaner experience, tag styling in the conversation cards has been refined for better visibility, and several smaller spacing and copy fixes are sprinkled throughout the project intake flow.