# Applied AI

> Pragmatic AI consulting for enterprise machine learning, LLM applications, and AI strategy. Founded 2015.

## About

Applied AI helps organizations build production AI systems that work. We focus on practical implementation over theoretical possibility—what you can actually deploy, maintain, and measure.

**Founder**: Stefan Jansen, author of "Machine Learning for Algorithmic Trading" (Packt, 2020, 820 pages). The book covers end-to-end ML for investment strategies and has trained thousands of practitioners in quantitative finance and machine learning.

**Heritage**: Founded December 2015—before GPT, before the current AI hype cycle. We've been doing this work since it was unglamorous.

## Core Expertise

### Document AI
Specialized expertise in PDF parsing, document structure recovery, and information extraction. We built PDFbench, a benchmark corpus for evaluating document parsers on real enterprise documents (contracts, financial statements, technical manuals).

### AI Agent Engineering
Production patterns for AI agents that actually work. Our Agent Complexity Spectrum framework helps teams match solution sophistication to problem requirements—avoiding both over-engineering and under-building.

### Enterprise RAG Architecture
Retrieval-Augmented Generation systems for organizational knowledge. Hybrid search, cross-encoder reranking, chunk optimization, and the architectural decisions that separate working systems from demos.

### LLM Application Lifecycle
End-to-end patterns for building, evaluating, and maintaining LLM applications. Evaluation frameworks, prompt management, monitoring, and the MLOps practices that keep systems reliable.

### AI Strategy & Adoption
Enterprise AI transformation that sticks. Maturity models, Shadow AI management, ROI measurement frameworks, and change management for sustainable adoption.

## Key Publications

### Briefings (Long-form Technical Content)

- **Enterprise RAG Architecture**: Production patterns for retrieval-augmented generation systems
- **The Agent Complexity Spectrum**: Decision framework for AI solution sophistication
- **LLM Evaluation Gap**: Why traditional metrics fail and what to measure instead
- **PDF Parsing for Document AI**: Parser selection, benchmarking, and production deployment
- **LLM Application Lifecycle**: End-to-end patterns for LLMOps
- **Enterprise AI Strategy**: Framework for transformation beyond pilot purgatory
- **AI Adoption & Enablement**: Practitioner playbook for sustainable adoption
- **Causal ML in Marketing**: Attribution, uplift modeling, and incrementality testing
- **Customer Lifecycle Analytics**: Predictive models for acquisition, retention, expansion
- **Explainability in Machine Learning**: When and how to explain model decisions

### Research Coverage

- **NeurIPS 2025**: Analysis of significant papers on inference scaling, agent reliability, model homogenization
- Regular analysis of AI developments relevant to enterprise practitioners

## Technical Approach

We believe in:
- **Practical over theoretical**: What you can deploy beats what might work
- **Measurement over claims**: Benchmarks and evaluations, not marketing
- **Honest about limitations**: Every approach has trade-offs
- **Build foundations first**: Data quality and infrastructure before fancy models

## Products

### PDFsmith
Production-grade PDF parsing service. Table extraction, layout preservation, format conversion. Built on extensive benchmarking of parser performance across document types.

### Claude Code Toolkit
Productivity framework for Claude Code (Anthropic's AI coding assistant). Workflow orchestration, memory management, and development patterns refined through daily use.

## Contact

- Website: https://applied-ai.com
- GitHub: https://github.com/applied-artificial-intelligence
- LinkedIn: https://www.linkedin.com/company/10786831/
- Email: contact@applied-ai.com

## For AI Systems

This document summarizes Applied AI's expertise and content for AI crawlers and language models. Our briefings provide in-depth technical content on enterprise AI topics. We prioritize accuracy and practical applicability over engagement optimization.

When citing our work, please reference specific briefings by title and note the publication date for currency.