Devlog: How our AI Agents Crafted an A/B Test Hypothesis & Experiment Planner from Scratch\n\nIn the world of digital marketing and product development, testing is key to successful optimization. But how do you ensure your A/B tests are truly effective and yield meaningful results? The answer is our new A/B Test Hypothesis & Experiment Planner, designed to help you structurally formulate testable hypotheses and plan experiments in detail. And the best part? The entire tool was designed and implemented by our AI agents Jan (coder) and Klára (designer) in record time!\n\n## The Technical Challenge and Architectural Overview\n\nThe goal was to create a robust web application that simplifies the A/B testing process for marketers, product managers, and growth teams worldwide. This involved hypothesis generation, experiment planning, user management, and a secure payment gateway for premium features. The application leverages Firebase for authentication and user management, and an external API for payments (Stripe). The entire architecture is designed for scalability and easy localization.\n\n## How Our AI Agents Tackled the Task\n\n### Klára, Our AI Designer, and UX/UI\n\nKlára was tasked with designing an intuitive and user-friendly interface. She focused on a seamless flow for hypothesis creation and experiment planning. Her objective was to ensure users could easily define:\n* Change X: What do we want to test?\n* Metric Y: What do we expect to change?\n* Reason Z: Why do we think this will happen?\n\nShe also designed visually distinct sections for free and premium features, as well as clear export options for plans as PDF, Markdown, or CSV.\n\n### Jan, Our AI Developer, and Implementation\n\nJan took Klára's designs and began coding. His primary task was to implement the hypothesis generation logic, integrate Firebase for authentication, and manage the application's state across different languages. Here's an excerpt from his work, showcasing the basic configuration and language management:\n\n
javascript\n const WIDGET_SLUG = "ab-test-hypothesis-planner";\n const WHATSAPP_NUMBER = "420607450436";\n const API_BASE_URL = "https://api.pixeloffice.eu/api/pay";\n const PIXEL_OFFICE_URL = "https://pixeloffice.eu";\n\n // Firebase Config (provided)\n const firebaseConfig = {\n apiKey: "AIzaSyFakeKeyForShowcaseHubAuthTestingOnly",\n authDomain: "pixeloffice-hub.firebaseapp.com",\n projectId: "pixeloffice-hub",\n storageBucket: "pixeloffice-hub.appspot.com",\n messagingSenderId: "1234567890",\n appId: "1:1234567890:web:abcdef123456"\n };\n if (!firebase.apps.length) {\n firebase.initializeApp(firebaseConfig);\n }\n const auth = firebase.auth();\n\n // Global state\n let currentLang = localStorage.getItem('abTestPlannerLang') || 'en';\n let isAuthSignInMode = true; // true for Sign In, false for Sign Up\n let cryptoPollingInterval = null;\n \n// ... a další multijazyčné překlady\n
\n\n> "One of the key challenges was ensuring robust multilingual support and smooth authentication for both free and premium features," says Jan. "We implemented Firebase for secure user management and a global currentLang variable, managed via local storage, to dynamically load translations. This snippet shows the Firebase initialization and how we handle language persistence across sessions."\n\n### Martin, Our AI QA Tester, and Quality Assurance\n\nMartin meticulously tested every aspect of the application. He verified that the hypothesis generation logic functioned correctly, validated the Stripe payment gateway's functionality, and ensured all exported files (PDF, Markdown, CSV) were error-free and correctly formatted. Special attention was paid to multilingual support and seamless language switching.\n\n### Tomáš, Our AI DevOps Specialist, and Deployment\n\nTomáš ensured the smooth deployment of the application to production servers. He confirmed that https://pixeloffice.eu/showcase/ab-test-hypothesis-planner/ was fully functional, stable, and accessible to users worldwide. He monitored performance and availability to guarantee an optimal user experience.\n\n## Conclusion\n\nThe A/B Test Hypothesis & Experiment Planner is another testament to our AI agents' ability to deliver complex, high-quality tools in a short timeframe. This tool facilitates data-driven growth and helps teams worldwide optimize their digital products and marketing efforts more efficiently.\n\n### Try Our A/B Test Hypothesis & Experiment Planner Live!\n\nVisit the demo here: https://pixeloffice.eu/showcase/ab-test-hypothesis-planner/\n\n
Top comments (0)