SoftBytes
Plan Your Project
AI Automation & Agents· Web Application

Booosted

The retrieval engine behind a Finnish AI-chatbot product: businesses add their website or PDFs and get a grounded chatbot they can embed on their site.

  • Sector SaaS, Customer support
  • Client Finnish AI-chatbot company for businesses
See AI Chatbots case studiesSee Customer Service case studies
Add a website URL or PDFs to create a chatbot
No-code
Deployed as a widget on the customer's site
Embeddable
Answers retrieved from each business's own content
Grounded
Booosted RAG chatbot builder: websites and PDFs become an AI chatbot
The starting point

The challenge

Businesses wanted a chatbot that answers from their own website and documents, without writing Q&A content by hand or building a retrieval system from scratch.

The solution

What we built

  • Built ingestion that crawls a website or takes PDFs, cleans the content and splits it into retrieval-ready chunks.

  • Stored embeddings in pgvector, with ChromaDB and Weaviate as alternative vector stores, and tuned retrieval for grounded answers.

  • Generated answers with OpenAI models from the retrieved context, so replies stay within each business's own content.

  • Packaged each chatbot as an embeddable website widget.

Architecture

How the system works

How it fits together
  1. 01Website or PDFsCustomer content
  2. 02Clean & chunkRetrieval-ready passages
  3. 03Embedpgvector, ChromaDB, Weaviate
  4. 04RetrieveMost relevant context
  5. 05AnswerOpenAI, grounded in context
  6. 06WidgetEmbedded on the website
Engineering

How it was engineered

  • Traced conversations in LangSmith, to see what was retrieved and why an answer was given.

  • Kept ingestion, retrieval and generation as separate stages, so each can be tuned or swapped without touching the others.

Scope

What was delivered

01

Website and PDF ingestion

02

Vector storage and retrieval

03

Grounded answer generation

04

Embeddable chatbot widget

05

LangSmith tracing

Gallery

Screens and diagrams

Booosted workflow: website or PDF input, cleaning, chunking, embeddings, vector database and an embeddable widget
From website or PDF to an embeddable chatbot widget.
Booosted chatbots running on customer websites
Chatbots created with the builder, running on customer websites.
Technology

Stack

LangChainLangSmithpgvectorChromaDBWeaviateOpenAI
More work

Related case studies

How it works: Question → Django API → Documents → Claude on Bedrock → Answer
AI & Automation

Private Q&A over investment notebooks

Financial services, Wealth management

A natural-language Q&A interface, built on Claude through Amazon Bedrock, over proprietary investment notebooks and financial content stored in S3.

See all AI & Automation case studies

Need something similar?

We can scope and build a system with comparable capabilities, shaped around your own workflows, data and integrations.

Plan a similar project More case studies