Skirrus

The description states that Skirrus is a chat-first drone mission planner. Users describe a location, desired outcome, aircraft, and optional capture settings in natural language.

Key technical components include:

• A React + TypeScript frontend
• Node.js + Express backend
• Integration with OpenAI's GPT-5.6 model via typed planning tools
• Geographic services such as Google Places, Nominatim, OpenStreetMap Overpass, and Open-Topo-Data
• Deterministic code for route geometry, camera math, validation, and export

The system separates AI interpretation from deterministic execution:

• GPT interprets intent, extracts constraints, chooses mission strategy, and explains results.
• Deterministic code owns coordinates, route geometry, validation, and export.

It supports exporting missions in formats like DJI Fly KMZ, Skirrus JSON, or GeoJSON.

Inference The product is a prototype built for a hackathon; no evidence of production deployment or customer usage exists.

The author positions Skirrus as:

• A tool that bridges the gap between human descriptions of drone jobs and detailed flight plans.
• An exploration into how AI can make complex workflows approachable without hiding uncertainty or taking responsibility for safety-critical decisions.

Claims made:

• AI makes a complex technical workflow approachable without hiding uncertainty or taking ownership of safety-critical calculations.
• The project explores how AI can interpret intent while deterministic code handles mathematical and flight-critical claims.
• It demonstrates a pattern for operational AI: let the model interpret intent and coordinate tools, but require sourced evidence, deterministic calculations, visible uncertainty, and human approval before action.

Inference These are self-stated claims about design philosophy and potential impact. No evidence of market validation or product-market fit is provided.

The description states that Skirrus targets:

• Drone pilots
• Photographers
• Inspectors
• Surveyors
• Small teams

These users understand the job they need to perform but may not have specialized GIS or mission-planning expertise.

Inference The target customer profile is inferred from the author's stated use case. No evidence of actual customers, personas, or segmentation data exists.

There is no mention of pricing, monetization strategy, or business model in the description.

Not evidenced

Key technical signals:

• Built with React, TypeScript, Vite, MapLibre GL JS, Node.js, Express
• Uses GPT-5.6 via OpenAI API with typed planning tools
• Integrates multiple geographic data sources (Google Places, Nominatim, OpenStreetMap Overpass, etc.)
• Implements deterministic code for geospatial calculations and export generation
• Includes a Plan Inspector showing AI decisions, evidence, and operator actions
• Supports DJI Fly KMZ export verified using native RC 2 mission data

The architecture separates AI interpretation from deterministic execution:

• GPT selects and revises mission intent.
• Deterministic code owns coordinates, route geometry, validation, and export.

Inference The technical stack and architecture suggest a well-thought-out hybrid approach. However, no evidence of scalability, performance metrics, or production deployment is provided.

The project was submitted to the OpenAI 2026 hackathon on Devpost.

No evidence of:

• Revenue
• Customers
• Adoption
• Usage data
• Product-market fit
• Market traction

Not evidenced

There is no mention of competitors or competitive landscape in the description.

Not evidenced

Key risks and red flags:

• The product is described as a hackathon prototype with no evidence of real-world validation.
• No revenue, customer, or traction data is available.
• The author notes that DJI export support remained blocked until native mission files confirmed required aircraft values — suggesting incomplete functionality.
• The system relies heavily on external geographic services that can be slow, ambiguous, unavailable, or rate-limited.
• There is no indication of how the product will scale beyond a single developer's implementation.

Inference The lack of real-world testing and validation raises concerns about readiness for commercial deployment.

1. What specific feedback have you received from drone operators or industry professionals outside of the hackathon?
2. How do you plan to validate the accuracy of AI-generated flight plans in real-world conditions?
3. Have you conducted any field tests using imported RC 2 missions and post-flight telemetry?
4. What is your roadmap for integrating airspace awareness, terrain data, and obstacle detection?
5. Are there any regulatory or safety compliance considerations that need to be addressed before commercial use?