Prompt Engineering: Building EcoTrack India in 14 Days

Manoj Kumar claims to have built EcoTrack India, a gamified carbon footprint platform, in 14 days using Google Antigravity. This case study details his prompt engineering playbook for rapid, AI-first SaaS development.

Manoj Kumar, writing on dev.to, reports building EcoTrack India, a gamified carbon footprint platform, in 14 days. The project, part of the PromptWars Virtual Challenge 3, aimed to create a hyper-localized carbon calculator for India's 1.4 billion population. Kumar claims this rapid development was possible using Google Antigravity and iterative prompt engineering.

Kumar states that traditional carbon calculators failed due to reliance on global averages, lack of actionable insights, poor retention mechanisms, and designs unsuited for Indian lifestyles. His goal with EcoTrack India was to provide state-specific emission factors and gamified scoring to drive user engagement.

Initial Prompt for Core Functionality

Kumar describes a shift in his workflow from idea → wireframe → code → debug → repeat to idea → prompt → working prototype → refine. His initial prompt to Google Antigravity was specific:

Build a multi-step carbon footprint calculator for India. Use state-specific electricity grid emission factors (coal-heavy states like Jharkhand: 0.95 kg CO2/kWh, renewables states like Himachal: 0.30 kg CO2/kWh). Include Scope 1 (petrol, LPG), Scope 2 (electricity), and Scope 3 (food, flights, shopping). Output a "Carbon Credit Score" from 0–850 styled like a financial credit score with tiers.

Kumar claims this prompt generated a working calculator, complete with 29 state grid factors and a Scope 1/2/3 breakdown, in under two minutes. He reports this initial generation saved 3–4 days of coding.

Iterative Refinement of Scoring Logic

The initial prototype, while functional, required refinement. Kumar reports that the first version's scoring formula was inaccurate, assigning low scores (e.g., 420/850) to users who should have scored high (e.g., 720+). To correct this, he specified an exact formula: Score = max(0, 850 - (userTonnes / 10 * 850)), capping it properly.

Personalizing AI Coach Nudges

Another reported issue was the generic nature of the AI coach tips. The first version provided the same advice to all users. Kumar refined his prompt to ensure personalization:

The AI coach tips must analyze the user's actual breakdown and identify their TOP emitting Scope category. If Scope 2 (electricity) is highest AND the user is in a coal-heavy state, show tips about rooftop solar and off-peak AC usage specific to that state's grid factor.

This prompt aimed to transform generic advice into contextually relevant suggestions, enhancing the user experience.

Designing Shareable UI Elements

Kumar also addressed the aesthetic and shareability of the platform. The initial Canvas share card was functional but lacked visual appeal. He prompted for a glassmorphism-styled card featuring a leaf pattern background, a prominent user tier emoji, and a QR code placeholder with a call to action: "Scan to calculate yours!" This aimed to create a shareable asset for social media platforms like Instagram.

What We'd Change

The founder's account highlights the potential of AI for rapid prototyping, but several aspects warrant scrutiny. The primary claim of building a platform for "1.4 billion Indians in 14 days" relies on the definition of "platform." The described output is a calculator with specific UI elements, not a scalable, production-ready system capable of handling a national user base. The post does not detail backend infrastructure, database design, user authentication, or the operational costs associated with such a scale. These are critical components for any public-facing application.

Furthermore, the core claims of development speed—"14 days" for the overall project and "under two minutes" for initial generation—are founder-reported without external verification. No public repository, live demo, or third-party audit is provided to substantiate these timelines or the functionality of the generated output. While the prompt examples are concrete, the actual output and its robustness remain unverified. The reliance on Google Antigravity, a proprietary tool, also limits the generalizability of this playbook. Founders considering similar approaches must account for tool availability and potential vendor lock-in.

Landing

The EcoTrack India case demonstrates AI's capacity to accelerate the initial stages of product development, particularly for generating functional prototypes and iterating on user experience elements. It shifts the bottleneck from manual coding to the precision of prompt engineering and the critical evaluation of AI-generated outputs. This approach allows founders to validate core concepts and user flows with unprecedented speed, but it does not negate the subsequent engineering effort required to transition a prototype into a robust, scalable product capable of serving a broad user base.

The investor read

The EcoTrack India signal highlights the increasing capital efficiency of AI-first rapid prototyping. Founders can validate niche SaaS ideas and build functional MVPs with significantly reduced time and engineering investment. This approach lowers the barrier to entry for environmental tech solutions, a sector with growing interest. However, for investability, a prompt-generated prototype must demonstrate a clear path to a scalable, production-ready backend, robust data validation, and a viable monetization strategy. The current signal represents a proof-of-concept, not a product ready for significant user acquisition or venture capital deployment. It signals a bootstrapped, lifestyle-oriented play focused on initial validation.

Pull quote: “Kumar claims this prompt generated a working calculator, complete with 29 state grid factors and a Scope 1/2/3 breakdown, in under two minutes.”