FieldMate

FieldMate is a ChatGPT-guided capture workflow that pays households energy credits for safe, useful site data that APIs, smart meters, and satellite imagery do not reliably capture.

Instead of sending a truck roll just to learn where a meter box is, whether a roof photo is usable, or how a service pole is obstructed, FieldMate turns a smartphone owner into a guided field observer. The participant completes three short capture sections, receives an immediate reward estimate, and creates a structured data sample that can help solar installers, EV charger installers, retailers, field technicians, networks, and government resilience teams plan work faster.

Try The Demo

TRY THE DEMO

FieldMate mockup, integrated into Amber Electric

Prototype Preview

The Problem

The energy transition depends on local, physical details that are still missing from digital systems:

• Smart meters know usage, not whether the meter box is accessible, damaged, blocked, or safe to inspect.
• Google Maps and aerial imagery miss recent roof changes, shading, access constraints, storm damage, and meter/pole conditions.
• Installers and network teams often discover basic site constraints only after scheduling a visit.
• Households are asked to participate in electrification, but rarely get paid for the useful data they can safely provide.

FieldMate closes that gap with guided, validated, participant-owned capture.

Solution

FieldMate asks for a small set of high-value observations:

• Meter Box: Exterior video or photos of the meter box location, access path, obstructions, and visible condition.
• Roof Setup: Ground-level roof photos and roof-edge outlines to support solar, battery, insulation, and electrification planning.
• Power Pole: Two safe-distance pole angles with notes about trees, sagging lines, storm changes, or access constraints.

The workflow is designed for normal phone users. It uses browser autofill when available, avoids forced manual demographic entry, gives concrete examples, validates blurry or incomplete roof submissions, and warns users not to climb onto roofs or approach unsafe infrastructure.

Why It Wins

Category	FieldMate case
Innovation	Converts household smartphone capture into structured energy-transition data that existing API, meter, and map sources miss. The reward loop turns data collection from an unpaid burden into a household incentive.
Usefulness	Serves both sides of the market: households earn energy credits, while installers, networks, retailers, field technicians, and government teams get timely site intelligence before dispatch.
Viability	Uses phones, browser autofill, single-file MCP app delivery, and reusable validation flows. It can scale suburb-by-suburb without deploying new hardware.
Technical Feasibility	The repo includes a working MCP app artifact, React capture UI, screenshots, and a server flow that can be run over Streamable HTTP or stdio. Future work can add computer-vision scoring, address verification, and buyer APIs.
Business Readiness	Clear value proposition: pay households a small reward for data that can reduce failed visits, shorten quoting cycles, improve storm response, and enrich electrification planning.
Sustainability	Better pre-site data can reduce unnecessary truck rolls, accelerate solar and EV readiness, and improve infrastructure resilience after extreme weather.

Business Model

FieldMate can sell verified capture packages to:

• Solar panel and battery companies assessing roof readiness.
• EV charger installers checking meter and service conditions.
• Electricity retailers targeting electrification offers.
• Distribution networks triaging pole, wire, vegetation, and storm-related risk.
• Government teams mapping resilience gaps after heatwaves, floods, fires, or storms.

Participants receive energy credits for approved submissions. Buyers pay for the data because it can reduce wasted visits, improve quote accuracy, and speed up deployment decisions.

Pitch to Companies

Solar Company Data Product

FieldMate's sample Mildura report found 1,000 resident-submitted properties,
with 460 showing strong roof suitability, 390 showing clear meter-box access,
and 210 ready for immediate solar-plus-battery follow-up.

We recommend prioritising these 210 high-readiness homes first, then
retargeting the 140 homes needing shade, switchboard, or access review.

Solar companies can buy this data to reduce wasted site visits, avoid weak
leads, and focus sales teams on homes with clearer installation potential.

Download Example Home Data Report
Buy Full Mildura Solar Lead Dataset

Distribution Network Data Product

FieldMate's sample Mildura report found 1,000 resident-submitted properties,
with 260 showing tree, pole, access, or service-line visibility issues, and
90 requiring higher-priority field review.

We recommend using these resident-confirmed captures to prioritise inspection
routes and update asset visibility records.

Distribution networks can buy this data to reduce blind spots, improve crew
planning, and identify access or safety issues before field workers arrive.

Download Example Property Visibility Report
Buy Full Grid-Edge Risk Dataset

Technical Architecture

The copied artifact contains a React MCP app packaged as a single HTML resource. The MCP server registers a tool with linked UI metadata, then serves the FieldMate capture UI to the host. The current prototype demonstrates the interaction model; future production work would replace the demo get-time tool with capture submission, validation, buyer export, and reward redemption tools.

Prototype artifacts are organized in MCP-Server-and-Assets/:

• server-source/ contains the compressed MCP server and React source bundle.
• single-file-app/ contains the renderable mcp-app.html resource.
• mockup-examples/ contains the Instagram acquisition ad.
• pitch-assets/ contains presentation assets for explaining the data gap.
• screenshots/ contains the seven-step participant flow screenshots.

Capture Flow

flowchart TD
    Start([Start FieldMate capture])
    Intro["Explain the job: 3 sections, roughly 10 minutes per section. At the end FieldMate calculates the value of energy credits and lets the participant redeem right away."]
    Demographics{"Browser has saved contact details?"}
    Autofill["Use saved browser details for name, phone, address, and email."]
    SkipDemographics["Do not force manual demographic entry."]

    MeterBrief["Section 1: Meter Box. Stand at the front of the house, record a video walking to the power meter, then hold the camera still in front of it. Show existing power meter examples."]
    MeterCapture["Participant records meter box video."]
    MeterThanks["Great, thanks. Move to the next step."]

    RoofBrief["Section 2: Roof Setup. Stand in front of the house and take a ground-level roof photo. Do not climb onto the roof."]
    RoofPhoto["Participant takes roof photo."]
    RoofPhotoCheck{"Is the roof photo clear?"}
    BlurryPhoto["Your picture was too blurry. Take the photo again holding it still."]
    OutlineBrief["Draw around the roof edge with your finger. Show a completed outline example."]
    RoofOutline["Participant outlines the roof edge."]
    OutlineCheck{"Is the roof outline complete?"}
    IncompleteOutline["You didn't complete the outline. Try to draw a complete shape around the edges of the roof with your finger, like this."]
    RoofSides["Repeat the roof photo and outline flow for the other 3 sides of the roof."]
    RoofDone["Success. Next section."]

    PoleBrief["Section 3: Power Pole. Take photos of the power pole from 2 safe angles."]
    PoleCapture["Participant captures 2 power pole angles from a safe distance."]
    PoleNotes["Capture useful notes: trees, sagging lines, storm changes, obstructions, and access constraints."]

    Rewards["Calculate rewards."]
    Wheel["Show spinning reward wheel."]
    Amount["Reveal $82 AUD in energy credits."]
    Redeem["Click here to redeem."]
    FollowUp["We will reach out when we need another data sample for another reward, likely at 6 months or after a natural disaster, hopefully just 6 months."]
    End([Submission complete])

    Start --> Intro --> Demographics
    Demographics -- Yes --> Autofill --> MeterBrief
    Demographics -- No --> SkipDemographics --> MeterBrief
    MeterBrief --> MeterCapture --> MeterThanks --> RoofBrief
    RoofBrief --> RoofPhoto --> RoofPhotoCheck
    RoofPhotoCheck -- No --> BlurryPhoto --> RoofPhoto
    RoofPhotoCheck -- Yes --> OutlineBrief --> RoofOutline --> OutlineCheck
    OutlineCheck -- No --> IncompleteOutline --> RoofOutline
    OutlineCheck -- Yes --> RoofSides --> RoofDone --> PoleBrief
    PoleBrief --> PoleCapture --> PoleNotes --> Rewards --> Wheel --> Amount --> Redeem --> FollowUp --> End

Loading

Close Data Gaps

MCP Server Flow

This is the MCP app/server structure copied into MCP-Server-and-Assets/server-source/fieldmate-mcp-app.tar.gz. The bundled server is based on the @modelcontextprotocol/server-basic-react example and serves the FieldMate React capture UI as a single MCP app resource.

flowchart TD
    Client["MCP host or ChatGPT client"]
    Transport{"Transport mode"}
    HTTP["Streamable HTTP server\nExpress app listens on /mcp"]
    Stdio["Stdio server\nUsed by local MCP client config"]
    Factory["createServer()\nCreates a fresh McpServer instance"]
    Server["McpServer\nBasic MCP App Server (React)"]
    Tool["registerAppTool: get-time\nReturns current server time"]
    Meta["_meta.ui.resourceUri\nui://get-time/mcp-app.html"]
    Resource["registerAppResource\nLoads dist/mcp-app.html"]
    CSP["Resource metadata\nMCP app MIME type, no border, font CSP"]
    UI["FieldMate React app\nDemographics, meter, roof, pole, reward"]
    HostContext["Host context bridge\nsafe-area insets and app callbacks"]
    Result["Tool result or rendered app response"]

    Client --> Transport
    Transport -- HTTP POST/GET /mcp --> HTTP
    Transport -- stdio mode --> Stdio
    HTTP --> Factory
    Stdio --> Factory
    Factory --> Server
    Server --> Tool
    Tool --> Meta
    Meta --> Resource
    Resource --> CSP
    Resource --> UI
    UI --> HostContext
    Tool --> Result
    UI --> Result

Loading

Roadmap

• Replace demo get-time tool with submit-capture, validate-capture, calculate-reward, and redeem-credit.
• Add computer-vision checks for blur, roof outline completeness, meter visibility, and unsafe pole proximity.
• Add address-level consent, expiry, and buyer-specific data packages.
• Add network and installer dashboards for filtering verified submissions by geography, asset type, urgency, and recency.
• Add disaster-response campaigns that request new samples only where conditions have changed.

MCP Server and Assets

These files were copied from https://fieldmate-mcp-app.b-cdn.net/ into MCP-Server-and-Assets/ and organized by purpose:

• server-source/fieldmate-mcp-app.tar.gz
• single-file-app/mcp-app.html
• single-file-app/mcp-app.html.gz
• mockup-examples/00-instagram-ad.png
• pitch-assets/close-data-gaps.png
• screenshots/01-demographics.png
• screenshots/02-intro.png
• screenshots/03-meter-section1.png
• screenshots/04-roof-blur-error.png
• screenshots/05-roof-outline-error.png
• screenshots/06-calculating.png
• screenshots/07-reward-82-redeem.png

Team

Team FieldMate: Jasper Fyfv, Ishaan Kataria, and Kasey Robinson.