Skip to content
Building AI-powered tools and teaching teams how to work with AI

AI & LLM Expertise

Hire us

AI in Practice

Visuality builds AI-powered internal tools for mid-sized companies using Ruby on Rails. We don't just integrate APIs — we build structured workflows for AI coding, implement MCP servers and clients, and ship vector search systems to production.

We developed and open-sourced our own AI-assisted development process used daily across the team.

AI Integrations

Speech & Image Recognition

Speech recognition, image recognition systems, voice-command interfaces

Tool Calling & LLM Integration

RubyLLM, dynamic tool calling, conversational interfaces, multi-step reasoning

Model Context Protocol

MCP Server & Client

ActionMCP, FastMCP, MCP Client integration, legacy system integration

MCP Template for Rails

Open-source template — generates 5 MCP tools per CRUD action. View on GitHub.

Vector Search & Embeddings

pgvector & OpenAI Embeddings

PostgreSQL pgvector, semantic similarity search, high-dimensional vectors (256–3,072 dimensions)

Multiple Similarity Measures

Cosine similarity, Euclidean distance, Manhattan distance

AI-Assisted Development

Structured Workflow

Research → plan → implement for production codebases. Adapted from HumanLayer's context engineering. Used daily with Claude Code.

Open Source

Open-sourced as visuality-humanlayer for Ruby on Rails codebases.

View on GitHub Read the article

Delivered Solutions

We've shipped AI-powered tools for real clients in production. These aren't demos — they're solutions that handle complex legacy systems, serve thousands of users, and replace manual processes with intelligent automation.

Visuality team collaborating

MCP Chat Assistant for Legacy System Navigation

Challenge: Complex legacy internal business system difficult to navigate

Solution: MCP Server + MCP Client chat assistant

  • Create, update, and search core domain entities
  • Intelligent navigation through legacy system
  • Natural language interface for system interaction

Impact: Simplified access to complex business logic

MCP Server for Legacy Application Admin

Challenge: Complex admin functionalities requiring deep system knowledge

Solution: MCP Server integrated with Claude Desktop

  • LLM-powered chat interface for administrative tasks
  • Easy access to complex functionalities without manual navigation

Impact: Reduced admin training time, increased efficiency

Event Creator with Tool Calling

Challenge: Event creation requiring multiple steps and data points

Solution: Conversational interface using Tool Calling and RubyLLM

Scale: Serving thousands of customers

  • Interactive question-based event creation
  • Intelligent data gathering through conversation
  • Guided user experience

Impact: Simplified event creation process for thousands of users

PDF Parsing for Multipage Reports

Challenge: Analyzing completion level of complex multipage PDF reports

Solution: Multi-step extraction with preprocessing and evaluations

  • Field completion assessment across multiple pages
  • Intelligent extraction with preprocessing
  • Quality evaluation and validation

Impact: Automated report analysis replacing manual review

How We Share Knowledge

We contribute to the Ruby and AI community through articles, workshops, and open-source tools.

Article Series

MCP Series — 5 articles

Ruby on Rails and Model Context Protocol

Introduction to MCP as the "USB-C port for AI applications"

MCP Server with Rails and ActionMCP

Production-ready MCP server with PostgreSQL-based session handling

MCP Server with Rails and FastMCP

Framework-agnostic MCP server with flexible argument definition

MCP Client in Rails using ruby-mcp-client gem

Dynamic tool loading from MCP servers and LLM integration patterns

MCP Template for Rails Applications

Open-source template for automatic CRUD tool generation

Vector Search Series — 2 articles

Vector Search in Ruby

PostgreSQL pgvector implementation for semantic similarity search

LLM Embeddings in Ruby

Semantic vectors, generation methods, and applications in search and recommendations

AI Development Process

From Vibes to Process: AI Coding in Production Codebases

How we built a structured AI coding workflow for complex Ruby on Rails projects

Open Source

MCP Template for Rails

Enhances Rails scaffolding to automatically generate 5 MCP tools per CRUD action. Built by Paweł Strzałkowski.

View on GitHub

AI Coding Workflow

Structured research → plan → implement workflow adapted for Ruby on Rails codebases. Used daily by the whole team.

View on GitHub Watch the talk

Building Mastery

AI isn't a side project at Visuality — it's a skill the whole team is building. We run regular internal hackathons where small groups tackle real problems with LLMs, MCP, and vector search.

Visuality AI hackathon dashboard — 5 projects, 132 commits, 20 contributors

Hackathon Projects

5 projects, 20 contributors

Snitcher

Work report generator — aggregates JIRA, Git, ClickUp, Linear via MCP, then generates reports using LLM.

MCP LLM

Notes Maker

Meeting recording processor with AI transcription, summarization, pgvector search, and RAG indexing.

LLM pgvector RAG

Falconbot

AI assistant with personality-driven chatbot, RubyLLM, Notion MCP integration, multi-provider LLM support.

RubyLLM MCP

RetroKraken

Team retrospective app with AI-powered note grouping and automatic mood analysis via LLM.

LLM

Conference & Community

We're active in the Ruby and AI community — organizing events, running workshops, and speaking at conferences worldwide. Paweł Strzałkowski (CTO) is one of the top 3 most active speakers in the Ruby community over the past 12 months.

Key Moments

2025

Rails World — "Make Rails AI-Ready by Design with MCP"

The talk that put Ruby + MCP on the map. Watch on YouTube.

2026

Ruby Community Conference, Cracow

Visuality as organizing partner. Lightning talk + 3-4 hour hands-on MCP workshop (20 seats). 3 of 7 workshops were AI-focused.

2026 — upcoming

Tropical on Rails, São Paulo

"MCP & OAuth on Rails: Building a Production-Ready AI App"

Paweł Strzałkowski presenting at a Ruby conference
Ruby Community Conference audience

Talks & Presentations

International Conferences

EuRuKo — Make Rails AI-Ready by Design with MCP

Enhanced version with additional content on MCP implementation

Balkan Ruby — Creativity: The only skill you need in the long term

Vector search, embeddings, speech recognition, image recognition. Watch on YouTube.

Ruby Community Conference — The joy of creativity in the age of AI

WebGL animated dog reacting to voice commands. Watch on YouTube.

Ruby Community Conference — Make a game with Ruby and MCP

Lightning talk demonstrating creative applications of Model Context Protocol

Meetup Presentations

Introduction to MCP in Ruby on Rails

Fundamentals of MCP and its implementation in Rails. Watch on YouTube.

Adding MCP to a legacy web application

Practical approach to integrating MCP into existing legacy applications. Watch on YouTube.

Madrid.rb — My LLM is smarter than yours

Deep dive into RAG and vector search in Ruby. Watch on YouTube.

Berlin Ruby + AI — Joy of creativity in the age of AI

Creative approaches to AI development in modern Ruby projects. Watch on YouTube.