AI agents & chatbots for business
I build AI assistants that work with your own data: answering customers from your documents, qualifying leads and taking repetitive work off your team.
What I build
Chatbots on your documents
Answers from your PDFs, pages and data, with links back to the source.
Lead-qualifying bots
Ask the right questions, collect details and hand qualified leads to sales.
AI inside your product
Summaries, recommendations and assistants built into your app.
Document reading & extraction
Pull structured data out of emails, PDFs and forms automatically.
Custom chatbot platforms
Multi-tenant chatbot SaaS with onboarding, billing and embeddable widgets.
Built for
- Businesses with lots of support questions
- Content-heavy websites
- Products that need AI features
Related work
Projects where I built this for real clients.
LiveChatAI
Create Custom Chatbot SaaS Platform
A modern SaaS solution developed with Next.js, Supabase, TypeScript, LangChain, and OpenAI API, empowering users to create fully personalized AI chatbots. Users can design chatbots by customizing names, avatar, chat starters, personas, and instructions during a seamless onboarding process. The platform allows content uploads via PDF, enabling chatbots to deliver responses based on user-provided data. With built-in authentication (signup/login), Stripe payment integration, and a user-friendly dashboard for managing chatbot settings, users can easily fine-tune their chatbots. Each chatbot comes with a unique URL for effortless integration into any website.
LivePod Transcripts AI Chatbot Application
Podcast Transcript AI Chatbot using Vercel AI SDK, Langchain & Supabase. I built a smart AI chatbot for a podcast website featuring popular episodes with detailed transcripts. The chatbot uses Vercel AI SDK, Langchain, and Supabase vector database for efficient and accurate responses. Key Features: Natural Language Querying: Users can ask questions in natural language, and the chatbot responds using data stored from podcast transcript PDFs. Contextual Answers: Uses Langchain and embeddings to understand and respond with accurate information sourced from stored vectors in Supabase. Related Podcast Recommendations: In addition to answering questions, the chatbot suggests relevant podcast episodes with clickable links that navigate directly to the episode’s transcript page. This solution is ideal for content-rich platforms like podcasts, video courses, or blogs that want to provide an interactive experience through AI-driven search and discovery.
LiveFlowtopia
Options Flow Analytics & Trader Intelligence Platform
Purpose: Flowtopia is built to help traders analyze options flow, monitor market sentiment, and learn data-driven trading strategies through one interactive platform. It is an ongoing web application focused on delivering a modern trading dashboard experience with features such as ticker insights, training content, notifications, and AI-assisted tools. In this project, I have been contributing to the frontend development and API integrations, with a strong focus on the ticker dashboard, notifications feature, training module, and AI bot feature. My work involves building responsive and user-friendly interfaces, integrating APIs and live data flows, and improving the overall usability of the platform as it continues to evolve.
From manual process to automated system
A clear, milestone-based process, so you always know what's being built and when.
- 01
Discover
We walk through your current workflow, tools and bottlenecks, including every manual step.
- 02
Blueprint
I map the automated flow, data model and scope, split into clear milestones.
- 03
Build
Weekly demos you can click through, integrated with the tools you already use.
- 04
Launch & support
Deploy, onboard your team and keep improving as the business grows.
Common questions
Can the chatbot answer from our own documents?
Yes. Your PDFs, pages and data are embedded into a vector database, and the assistant answers from them with links back to the source.
Which AI models do you use?
I work with OpenAI models, LangChain and the Vercel AI SDK, and choose the model based on quality, speed and cost for your use case.
Can it be added to our existing website or app?
Yes, as an embedded chat widget, a dedicated page or an AI feature inside your product.
Still running on spreadsheets, emails and copy-paste?
Tell me how your team works today and I'll show you what can be automated.