Mapflow.ai

@mapflow.ai check my account balance and show which AI models I can run for building detection in satellite imagery

Your Mapflow account has 2,450 credits remaining. Available AI models include: building-detection (v2.1, consumes ~0.8 credits/km²), road-extraction (v1.9, ~1.2 credits/km²), and forest-change-detection (v2.0, ~1.5 credits/km²). Your plan supports up to 5 concurrent processings and 10 GB of custom imagery storage. All models accept Maxar, Planet, and Sentinel-2 imagery sources. This illustrative response shows the kind of account-capability summary the MCP returns before you commit credits to a processing run.

This example demonstrates the get-capabilities tool, which you should call before estimating or starting any processing. The MCP returns your current credit balance and model catalog in one shot, so you can verify affordability and choose the right workflow. API key must have read access to account metadata.

@mapflow.ai estimate the cost to detect buildings across downtown Portland using Maxar imagery, then start the processing if it's under 500 credits

Cost estimate for building-detection over 12.3 km² of downtown Portland: 9.84 credits (0.8 credits/km²). Since this is well under your 500-credit threshold, I've created processing job proc_7x9k2m in your default project. Status: RUNNING. The job will analyze Maxar imagery from the past 30 days and output building footprints as GeoJSON. Estimated completion: 8–12 minutes. I'll poll status every 60 seconds and notify you when results are ready for download. This illustrative example shows the estimate-then-create workflow.

This example chains estimate and create-processing tools, demonstrating credit-aware automation. The create call consumes credits immediately and starts an async job—no undo. You must poll get-processing every 30–60 seconds until status is OK, then call download-result. API key needs processing:write scope.

@mapflow.ai list all my recent forest-change processings, download the results for any that finished successfully, and summarize which regions showed the most deforestation

Found 3 forest-change processings from the past two weeks. Two completed successfully: proc_4a8n (Willamette Valley, 45 km²) and proc_6c1p (Mount Hood foothills, 78 km²). I've downloaded both GeoJSON result files. Analyzing the vector data: Willamette Valley shows 2.1 km² of forest loss (4.7% of AOI), concentrated in the southwest quadrant. Mount Hood foothills show 0.8 km² of loss (1.0% of AOI), scattered across logging roads. The third processing (proc_5d9m) is still running. This illustrative response shows how the MCP pairs list, download, and AI reasoning to surface insights from completed jobs.

This example combines list-processings, download-result, and the AI's spatial reasoning to synthesize findings across multiple runs. The MCP returns GeoJSON vector layers; the AI can parse geometry and attributes to answer comparative questions. Requires API key with processing:read scope and sufficient credit history to have completed jobs.

When satellite imagery analysis beats manual labeling for emergency ops

A 6-person humanitarian team needs building footprints and road networks mapped across 200 square kilometers after a flood, and they need it in 48 hours. Mapflow wins here because the Create Processing tool runs AI models on satellite imagery without human labeling—estimate the credit cost first with Get Account Capabilities, confirm your balance covers it, then fire the processing and poll Get Processing every minute. The GeoJSON output from Download Processing Result drops straight into QGIS or a web map. The threshold: if your AOI is under 50 sq km and you have GIS staff on hand, manual digitizing might be faster than waiting for satellite tasking and processing. Above 100 sq km, Mapflow's batch approach is the only realistic play for small teams without a full geospatial pipeline.

Track neighborhood development without a dedicated GIS analyst

A 3-person city planning consultancy monitors construction permits against actual building growth in a 15 sq km district, running quarterly reports for the client. Mapflow's building-detection models let you compare imagery from different dates without hiring a remote sensing specialist—create a project with Create Project to organize the quarterly runs, estimate costs upfront, then schedule processings when new imagery drops. The catch: Mapflow consumes credits per processing, so if you're running weekly updates or covering 500+ sq km, the cost stacks up fast and you need a standing budget line. For quarterly snapshots under 50 sq km, the credit model is cheaper than a fractional GIS hire. Use Get Storage Overview to reuse uploaded basemaps across quarters and keep costs predictable.

When you need vector data from imagery but lack in-house spatial skills

A 4-person agtech startup needs field polygons for 80 farms (300 sq km total) to power a crop-monitoring dashboard, but no one on the team has used PostGIS or trained a segmentation model. Mapflow's AI workflows handle the imagery-to-vector pipeline in one API call—List Processings shows you what's already running, Estimate Processing Cost prevents budget surprises, and the GeoJSON output from Download Processing Result plugs directly into your app's map layer. The trade-off: if your farm boundaries change every month (new leases, subdivisions), the per-processing credit burn adds up and you might need a cheaper solution like manual digitizing or a one-time data purchase. For annual or seasonal updates across 100+ fields, Mapflow's automation beats hiring a contractor to trace polygons by hand.