CloseIQ
Designed, developed and shipped 0→1 AI assistant in 5 days
Focus Areas:
Design
Vibe Coding
AI Software
AI Software
Front-end Dev
Back-end Dev
Testing
Back-end Dev
Testing
Software:
Project Overview
Building an end-to-end AI assistant for internal search
Problem
Sales reps often waste time searching for relevant answers hidden in messy internal documentation. Many companies lack a unified system for turning scattered content into clear, actionable insights. This is especially true for teams without robust CRMs or knowledge bases. The result is delays, duplicated work, and missed opportunities.
Solution
CloseIQ transforms messy internal content into a smart, conversational search experience powered by OpenAI’s API. Reps can ask natural-language questions like “How do we handle renewals?” or “What’s our cold call script?” and instantly get focused, on-brand answers sourced from company PDFs, notes, and sales materials.
My Role / Responsibilities
I led the project end to end, starting with defining the core use case and designing the UX. From there, I developed the front-end, integrated OpenAI, and deployed the back-end. Throughout the process, I used ChatGPT as a thinking and coding partner to help architect the system, debug issues, and quickly iterate on features.
Research & Planning
Using LLMs as both a development partner and a technical foundation
1 / Market Need
Many companies operate without a robust CRM or a centralized knowledge base. Instead, critical information lives in scattered docs, siloed team notes, or outdated wikis. CloseIQ addresses this gap by turning disorganized internal content into a searchable, conversational interface. It's especially valuable for fast-moving teams that need quick answers but lack a clean way to access their own knowledge.
2 / Product Concept
CloseIQ is an AI-powered assistant designed to answer internal questions using a company’s own documentation. Users can upload PDFs, notes, or sales materials, and the system turns that content into a lightweight, chat-based knowledge layer. The experience is fast, familiar, and designed to feel like chatting with a well-informed teammate.
3 / Technical Knowledge Gap
In addition to data from third-party platforms, user interviews were conducted with both external users and internal stakeholders to evaluate the existing onboarding screens across desktop and mobile. Feedback was aggregated and analyzed to identify key bottlenecks, providing valuable insights into UX improvements aimed at increasing conversion rates.
1 / Market Need
2 / Product Concept
3 / Technical Knowledge Gap
Production
Designing, building, and shipping a working AI product from scratch
4 / System Architecture
I used ChatGPT to help map out the full system architecture, from document ingestion to vector-based search and GPT response generation. I designed the pipeline to be modular and scalable, allowing new datasets to be added with minimal setup. Supabase handled vector storage and querying, while Vercel hosted the front and back ends. This foundation made it easy to test, iterate, and deploy quickly.
5 / Interface Design
The interface was built to feel familiar and frictionless, more like a chat experience than a traditional dashboard. I designed it to reveal complexity only when needed, letting users ask questions and get instant answers without setup or training. All components were built in Framer and React, with Tailwind used for precise styling and layout.
6 / Development
I handled both the front-end and back-end development, using ChatGPT as a thinking partner and code assistant. On the front end, I implemented responsive layouts, typing animations, and a persistent chat interface. On the back end, I set up document parsing, embedding, Supabase integration, and GPT prompt tuning. Every part of the system was designed to be fast, flexible, and easy to extend.
4 / System Architecture
5 / Interface Design
6 / Development
Project Overview
Building an end-to-end AI assistant for internal search
Problem
Sales reps often waste time searching for relevant answers hidden in messy internal documentation. Many companies lack a unified system for turning scattered content into clear, actionable insights. This is especially true for teams without robust CRMs or knowledge bases. The result is delays, duplicated work, and missed opportunities.
Solution
CloseIQ transforms messy internal content into a smart, conversational search experience powered by OpenAI’s API. Reps can ask natural-language questions like “How do we handle renewals?” or “What’s our cold call script?” and instantly get focused, on-brand answers sourced from company PDFs, notes, and sales materials.
My Role / Responsibilities
I led the project end to end, starting with defining the core use case and designing the UX. From there, I developed the front-end, integrated OpenAI, and deployed the back-end. Throughout the process, I used ChatGPT as a thinking and coding partner to help architect the system, debug issues, and quickly iterate on features.
Project Overview
Building an end-to-end AI assistant for internal search
Problem
Sales reps often waste time searching for relevant answers hidden in messy internal documentation. Many companies lack a unified system for turning scattered content into clear, actionable insights. This is especially true for teams without robust CRMs or knowledge bases. The result is delays, duplicated work, and missed opportunities.
Solution
CloseIQ transforms messy internal content into a smart, conversational search experience powered by OpenAI’s API. Reps can ask natural-language questions like “How do we handle renewals?” or “What’s our cold call script?” and instantly get focused, on-brand answers sourced from company PDFs, notes, and sales materials.
My Role / Responsibilities
I led the project end to end, starting with defining the core use case and designing the UX. From there, I developed the front-end, integrated OpenAI, and deployed the back-end. Throughout the process, I used ChatGPT as a thinking and coding partner to help architect the system, debug issues, and quickly iterate on features.
2 / Product Concept
CloseIQ is an AI-powered assistant designed to answer internal questions using a company’s own documentation. Users can upload PDFs, notes, or sales materials, and the system turns that content into a lightweight, chat-based knowledge layer. The experience is fast, familiar, and designed to feel like chatting with a well-informed teammate.
2 / Product Concept
CloseIQ is an AI-powered assistant designed to answer internal questions using a company’s own documentation. Users can upload PDFs, notes, or sales materials, and the system turns that content into a lightweight, chat-based knowledge layer. The experience is fast, familiar, and designed to feel like chatting with a well-informed teammate.
3 / Technical Knowledge Gap
As a designer without a traditional engineering background, I used ChatGPT as a hands-on coding partner throughout development. From setting up the API architecture to writing search logic and embedding pipelines, ChatGPT helped me overcome technical hurdles and build a fully functional product from the ground up.
3 / Technical Knowledge Gap
As a designer without a traditional engineering background, I used ChatGPT as a hands-on coding partner throughout development. From setting up the API architecture to writing search logic and embedding pipelines, ChatGPT helped me overcome technical hurdles and build a fully functional product from the ground up.
Launch & Refinement
Testing in the real world and refining the product through live feedback
7 / Launch
After building a working MVP, I deployed CloseIQ to a live environment and tested it across multiple branded datasets. I focused on making the deployment process as seamless as possible, ensuring that each instance loaded the correct visuals, content, and organizational context. This allowed me to simulate real-world usage and evaluate how the system performed under actual conditions.
8 / Product Refinement
With the system live, I iterated quickly based on how the agent responded to real queries. I adjusted prompts to reduce fluff, improved how sources were cited, and made the responses feel more direct and on-brand. ChatGPT was instrumental in helping me identify weak spots, rewrite prompt logic, and make fast improvements to both the frontend and backend experience.
4 / System Architecture
I used ChatGPT to help map out the full system architecture, from document ingestion to vector-based search and GPT response generation. I designed the pipeline to be modular and scalable, allowing new datasets to be added with minimal setup. Supabase handled vector storage and querying, while Vercel hosted the front and back ends. This foundation made it easy to test, iterate, and deploy quickly.
4 / System Architecture
I used ChatGPT to help map out the full system architecture, from document ingestion to vector-based search and GPT response generation. I designed the pipeline to be modular and scalable, allowing new datasets to be added with minimal setup. Supabase handled vector storage and querying, while Vercel hosted the front and back ends. This foundation made it easy to test, iterate, and deploy quickly.
5 / Interface Design
The interface was built to feel familiar and frictionless, more like a chat experience than a traditional dashboard. I designed it to reveal complexity only when needed, letting users ask questions and get instant answers without setup or training. All components were built in Framer and React, with Tailwind used for precise styling and layout.
5 / Interface Design
The interface was built to feel familiar and frictionless, more like a chat experience than a traditional dashboard. I designed it to reveal complexity only when needed, letting users ask questions and get instant answers without setup or training. All components were built in Framer and React, with Tailwind used for precise styling and layout.
7 / Launch
8 / Product Refinement
6 / Development
I handled both the front-end and back-end development, using ChatGPT as a thinking partner and code assistant. On the front end, I implemented responsive layouts, typing animations, and a persistent chat interface. On the back end, I set up document parsing, embedding, Supabase integration, and GPT prompt tuning. Every part of the system was designed to be fast, flexible, and easy to extend.
6 / Development
I handled both the front-end and back-end development, using ChatGPT as a thinking partner and code assistant. On the front end, I implemented responsive layouts, typing animations, and a persistent chat interface. On the back end, I set up document parsing, embedding, Supabase integration, and GPT prompt tuning. Every part of the system was designed to be fast, flexible, and easy to extend.
7 / Launch
After building a working MVP, I deployed CloseIQ to a live environment and tested it across multiple branded datasets. I focused on making the deployment process as seamless as possible, ensuring that each instance loaded the correct visuals, content, and organizational context. This allowed me to simulate real-world usage and evaluate how the system performed under actual conditions.
7 / Launch
After building a working MVP, I deployed CloseIQ to a live environment and tested it across multiple branded datasets. I focused on making the deployment process as seamless as possible, ensuring that each instance loaded the correct visuals, content, and organizational context. This allowed me to simulate real-world usage and evaluate how the system performed under actual conditions.
8 / Product Refinement
With the system live, I iterated quickly based on how the agent responded to real queries. I adjusted prompts to reduce fluff, improved how sources were cited, and made the responses feel more direct and on-brand. ChatGPT was instrumental in helping me identify weak spots, rewrite prompt logic, and make fast improvements to both the frontend and backend experience.
8 / Product Refinement
With the system live, I iterated quickly based on how the agent responded to real queries. I adjusted prompts to reduce fluff, improved how sources were cited, and made the responses feel more direct and on-brand. ChatGPT was instrumental in helping me identify weak spots, rewrite prompt logic, and make fast improvements to both the frontend and backend experience.
1 / Market Need
Many companies operate without a robust CRM or a centralized knowledge base. Instead, critical information lives in scattered docs, siloed team notes, or outdated wikis. CloseIQ addresses this gap by turning disorganized internal content into a searchable, conversational interface. It's especially valuable for fast-moving teams that need quick answers but lack a clean way to access their own knowledge.
1 / Market Need
Many companies operate without a robust CRM or a centralized knowledge base. Instead, critical information lives in scattered docs, siloed team notes, or outdated wikis. CloseIQ addresses this gap by turning disorganized internal content into a searchable, conversational interface. It's especially valuable for fast-moving teams that need quick answers but lack a clean way to access their own knowledge.
Research & Planning
Using LLMs as both a development partner and a technical foundation
Research & Planning
Using LLMs as both a development partner and a technical foundation
Production
Designing, building, and shipping a working AI product from scratch
Production
Designing, building, and shipping a working AI product from scratch
Launch & Refinement
Testing in the real world and refining the product through live feedback
Launch & Refinement