Objective Develop an AI-driven real estate translation analysis tool to identify UK-wide office properties suitable for residential conversion. The tool will use a multi-agent architecture, where a central executor agent coordinates specialized sub-agents for different tasks. The solution must be developed in or be compatible with no-code or low-code platforms for easy modification. The architecture should allow sub-agents to be replaced or additional agents to be added. While initial datasets are provided, future iterations should enable automated data searches and scraping. Further detail is below on the subagents and features required. --- Research Subagents 1. Transaction Data Enricher (Base Dataset Provided) A dataset of property transactions will be given in CSV format. The sub-agent must research and enrich the dataset with additional details: • Location – Address, nearest tube/train station, Google Maps link • Listing Type – Sale or Rent • Sale Price – Blank for rental properties or if unavailable • Rent – Absolute (£) and relative (£/psf) • Size – Area (psf) • Floors – Number of floors • Agent Details – Name, Email, Contact Number, Website • Sales Details – Link to sales website and brochure • Photos – Extract images from sales details Current Data Format: • CSV 1 – Properties for Sale • CSV 2 – Properties for Lease Data Handling Expectations: • If data is missing or inconsistent, cross-check alternative sources where available. • Flag incomplete entries for manual review. • Provide confidence scores for enriched data points. --- 2. Residential Market Value (Base Dataset Provided) The sub-agent retrieves the latest transaction data for residential properties in the immediate vicinity, prioritizing: • Minimum 25 transactions collected • Average property values calculated by type (Flat, Terraced House, Detached House, etc.) • Outputs in both map and list format with full UK Land Registry data Current Data Format: CSV from UK Land Registry --- 3. Constraints Analysis (Base Datasets Provided) The sub-agent checks if a property is affected by UK-specific constraints. If triggered, it provides details; otherwise, it marks "No" for each category: • Article 4 Area – Format: JSON, GPKG • Flood Risk – Format: GeoJSON • Conservation Area – Format: GeoJSON • Listed Building – Format: GeoJSON --- 4. Planning Analysis (Optional but Beneficial) The sub-agent researches the planning history of the subject property and surrounding properties (approved or rejected). Data Points Collected: • Planning History • Title of Application • Application Number • Link to Application • Approval Status • Decision Date • Nearby Major Developments • Local council development plans, regional planning updates News Articles - Identify relevant news about the property or surrounding area --- Analysis & Reporting Subagents 5. Property Ranking • Rank properties using a profitability score, factoring in projected residential value vs. existing commercial value. • Consider binary constraints (e.g., flood risk, conservation area) in ranking. 6. Data Output & Visualization • Database & Interactive Dashboard – A structured database with an interactive dashboard displaying findings. • Filters – Users can filter by profitability score, constraints, and location. • Export Options – Allow exporting data in CSV and PDF formats. • Automated Alerts – Notify users of new high-potential properties. --- Additional Features • Discounted Properties View – Highlight properties with price reductions and reasons for discounts. • Feedback Loop – Confirm data accuracy through user feedback, enabling continuous improvement. • Error Handling – Handle site changes, rate limiting, and data inconsistencies. Please revert with expected timings for the first iterations and final output.