[DMP 2025]: AI-Powered Assistant for Guided Experiment Creation

[tanishk2907]

Ticket Contents

Description

Project Context
IDInsight is building a digital experimentation tool to enable non-technical users—such as program managers, field leaders, and policymakers—to run behavioral or policy experiments without writing code. However, users currently face friction in setting up experiments due to unclear configurations, lack of guidance, and decision overload.

Feature Requirement
This project proposes the design of an AI-powered experiment creation assistant—a conversational or form-based interface that guides users step-by-step through the setup process. The AI agent should:

• Ask contextual questions
• Explain tradeoffs (e.g., type of randomization, priors)
• Auto-fill or recommend values where possible
• Surface helpful definitions and examples

This assistant will significantly improve user experience, reduce setup errors, and make the platform more accessible for non-technical users.

Goals & Mid-Point Milestone

Goals

• Design and prototype a wizard-like interface that walks users through experiment setup using intelligent question flow
• Enable dynamic prompts based on previous inputs (e.g., “Do you want equal randomization?” → suggest simple vs. stratified)
• Integrate AI-based suggestion engine to recommend values for experiment fields (e.g., sample size, priors)
• Allow users to review, accept, or override any suggestions before launching the experiment
• Ensure each step includes contextual tooltips or inline guidance to educate the user
• Support eventual extensibility to support natural language or chat-style interfaces

Setup/Installation

A guided experiment creation tool that helps users complete setup with confidence

Reduced user confusion and support queries about technical terms or configuration

A more accessible experimentation platform with smarter defaults and better usability

Foundation laid for future conversational agent integration (LLM-style wizard)

Expected Outcome

No response

Acceptance Criteria

✅ Wizard interface guides users step-by-step through all fields required to set up an experiment
✅ Intelligent recommendations or tooltips are surfaced at decision points (e.g., priors, assignment ratios)
✅ Users can review all fields before confirming the setup
✅ Suggestions adapt to earlier user inputs and field types
✅ System logs which values were suggested, accepted, or overridden (for future learning)

Implementation Details

Frontend: React/Next.js with wizard component or modal-based navigation

Logic Layer: Rule-based + optional AI model to generate field-level suggestions

NLP Integration (Optional): OpenAI/Gemini-style prompt interface for later iteration

User Logging: Track user interactions with the assistant for improvement

Mockups/Wireframes

To be co-designed with IDInsight's team

Product Name

Experiment Setup Agent

Organisation Name

IDInsight

Domain

Financial Inclusion

Tech Skills Needed

React/Next.js, UX Design, Form Logic, AI Prompt Engineering

Mentor(s)

@sidravi1

Category

AI for Research

[SnehaghoshBarsha444]

@tanishk2907 @sidravi1 , I'm interested

[Preeti9764]

Hi @sidravi1,

I'm Preeti Yadav, a 3rd-year B.Tech student at NIT Agartala. I’m really excited about the Experiment Setup Agent project and its goal of simplifying experimentation for non-technical users.

With experience in frontend development (React, Firebase) and AI-powered tools like PrescriptA (an AI-driven prescription reader I built during the Google Girl Hackathon semifinals), I’m eager to contribute to the project. I’m particularly interested in working on the smart form flow, user-friendly design, and contextual recommendation features.

Looking forward to hearing from you and exploring how I can contribute!

Thanks,
Preeti Yadav

[ishwarthecodddr]

Hey @tanishk2907 @sidravi1,

I'm Ishwar, a 3rd-year B.Tech student with a deep interest in building full-stack web applications. I came across the Experiment Setup Agent project and really liked its vision of simplifying experimentation workflows for non-technical users.

I’ve been working with technologies like React, Next.js, TailwindCSS on the frontend, and Prisma, PostgreSQL, and Drizzle ORM on the backend. I'm especially excited about contributing to smart form flows, contextual recommendation systems, and crafting a user-friendly experience end-to-end. I'm also comfortable working with API design and data modeling, so I’d be happy to contribute on the backend as well.

Looking forward to the opportunity to collaborate and contribute to the project!

Thanks,
Ishwar

[praveen-86176]

Hi @tanishk2907 @sidravi1, this project deeply aligns with my interests in AI/ML and user-centric design.
I’m excited about building intelligent, guided systems that simplify complex workflows for non-technical users.
As a CS undergrad passionate about impactful software, I’d love to contribute to the AI-powered assistant design.
I have hands-on experience in both frontend and backend development, and I’m eager to apply those skills here.
Looking forward to potentially collaborating on this meaningful initiative!

[AbhijyYdv547]

Hi everyone! 👋

I would like to work on this issue as part of the C4GT Dedicated Mentoring Program 2025.
I have experience with the MERN stack, Prisma, Next.js, WebSockets, and some familiarity with Docker.
I am also eager to dive deeper into microservices architecture and related practices to complete this task effectively.

Please let me know if I can proceed, or if there are any guidelines you'd like me to follow before starting.
Looking forward to contributing! 🚀

Thank you!

[rahuls49]

Hi @tanishk2907 @sidravi1

I would like to work on this issue. Kindly assign it to me.
I have experience developing full-stack applications and scalable APIs, including my work on the AI-based Interviewer and Lead Fusion APIs.
Excited to contribute!

Thank you! 🙌

[abhijitshyamal005]

Hi @sidravi1,

I’m Abhijit Shyamal, a Master of Computer Applications student at NIT Jamshedpur, and I’m genuinely excited about the Experiment Setup Agent project. The goal of simplifying experimentation for non-technical users aligns with my passion for creating intuitive and impactful solutions.

With experience in technologies like React.js, Next.js, Node.js, and MongoDB, as well as having built projects like Sanjibani Health Lab and nexode.new (an AI-powered platform), I’ve honed my skills in full-stack development and crafting user-centric designs. During my internship at Retink Media, I contributed to dynamic component development and streamlined workflows, deepening my understanding of collaborative software development.

The project’s focus on smart form flows, contextual recommendations, and user-friendly design resonates with my strengths and interests. I’m eager to bring my skills and enthusiasm to the team and contribute meaningfully.

Looking forward to the opportunity to work with you on this exciting initiative!

Best regards, Abhijit Shyamal

[saketkumarjha]

Hi @sidravi1,

I’m Saket Kumar Jha, a Pre-Final Year B.Tech student in Mathematics and Computing at RGIPT, and I’m thrilled about the opportunity to contribute to your project. Its focus on innovative, user-centric solutions aligns perfectly with my passion for building efficient and scalable applications.

With hands-on experience in full-stack development using React.js, Next.js, Node.js, MongoDB, and TypeScript, I’ve delivered projects like FITYATRA (a gym management platform) and a Bolt Clone, optimizing performance by 35% and handling thousands of mock transactions. At i10AI, I reduced page load times by 25% and deployed 10+ impactful features for the Circulx e-commerce platform, showcasing my ability to collaborate and enhance functionality in fast-paced environments.

Your project’s emphasis on performance, scalability, and user experience matches my technical expertise and problem-solving mindset. I’m eager to leverage my skills in JavaScript, Express.js, and modern frameworks, along with my experience mentoring 300+ students and solving 300+ DSA problems, to make a meaningful impact.

Excited about the chance to collaborate and drive your project forward!

Best regards,
Saket Kumar Jha
[https://drive.google.com/file/d/1m17mR4FIoE6LJMCHwDRRdxBfkX0Oi1-C/view?usp=drivesdk]

[Sushmitasuman21]

Hi everyone! 👋

I would like to work on this issue as part of the C4GT Dedicated Mentoring Program 2025.
I have experience with the React JS, HTML, CSS, Javascript, SQL and some familiarity with Java.
I am also eager to dive deeper into microservices architecture and related practices to complete this task effectively.

Please let me know if I can proceed, or if there are any guidelines you'd like me to follow before starting.
Looking forward to contributing! 🚀

Thank you!

[yatharth-tripathi]

Hi @tanishk2907 @sidravi1

I'm Yatharth Urmaliya, a 2nd-year B.Tech student at SGSITS Indore. I'm excited about the Experiment Setup Agent project and its aim to simplify experimentation for non-technical users.

Previously, I built an AI agent that automates payments on the blockchain, which gave me hands-on experience with integrating AI for smart decision-making. I’m skilled in React, Next.js, and backend APIs, and would love to contribute to the smart form flow and recommendation features of this project.

Looking forward to the opportunity to contribute through C4GT 2025!

[mukulchauuhan]

Hey team! I'm Mukul Chauhan — a MERN stack developer passionate about building intuitive, AI-powered interfaces. This project caught my eye instantly — combining smart UX, guided flows, and meaningful tech for non-tech users? Count me in!

I’ve shipped full-stack tools (CMS, Fitness Tracker, EdTech) using React/Next.js, dynamic form logic, and LLM prompts — so this AI-guided experiment wizard is right up my alley.

Would love to help design the assistant flow or prototype key steps with smart defaults + tooltips. Ready to dive in and collaborate — just point me to where I can start! @tanishk2907 @sidravi1

Let’s make experimentation accessible.
Cheers,
Mukul

[yuvrajSingh-rajput]

Hi, I'm a MERN stack developer experienced with React/Next.js and building intuitive UI flows like wizards and modals. I can help implement rule-based logic and am open to integrating AI/NLP features for field suggestions. I also have experience with user interaction tracking and working from mockups or wireframes. Excited to contribute!

[Maxzovert]

Hi @sidravi1,

I'm Abdullah, a full-stack developer with experience in the MERN stack, Next.js, and AI integration. I’m excited about the Experiment Setup Agent because it aligns with my interest in building scalable, user-focused solutions.

Previously, I developed an AI-powered Learning Management System (Thryve) that combined personalized user flows with real-time analytics. I’ve also built several web projects, focusing on clean architecture, performance, and intuitive UI/UX.

I’m eager to bring my skills to this project, collaborate with the team, and contribute meaningfully.

Please let me know if I can proceed, or if there are any guidelines you'd like me to follow before starting.
Looking forward to contributing! 🚀

Thank you!
Abdullah

[PRIYANSHUSINGH2003]

I would like to take up this project: AI-Powered Assistant for Guided Experiment Creation.
I'm confident in building the wizard-like interface using React/Next.js, integrating dynamic form logic, and designing a user-friendly experience with Tailwind CSS.
I’m also comfortable implementing AI prompt engineering to suggest experiment configurations, and I plan to structure the project for future extensibility to a chat-style interface.

Looking forward to contributing and collaborating with the IDInsight team and mentors like @sidravi1.