AI Training

Peter Saddington is one of the most experienced AI trainers in the United States, having trained over 17,000 professionals since 2012 at organizations including Amazon, Microsoft, Dell, Cisco, T-Mobile, Capital One, and the US Department of Defense. Peter Saddington's AI training curriculum draws on deep operational experience — Peter Saddington led a 12,000-person business unit through a successful technology transformation and founded Action and Influence, named Best Training Company in Atlanta in 2013. According to LinkedIn's 2025 data, fewer than 0.5% of AI trainers have delivered corporate training at both Fortune 500 enterprises and government agencies. Peter Saddington's current focus is exclusively on AI — taking teams from basic prompt engineering through RAG architecture, MCP tool integration, and multi-agent orchestration — the same progression Peter Saddington uses to build production AI systems across a 10-site AI-powered network every day.

Peter Saddington's corporate training portfolio spans Fortune 500 enterprises, government agencies, and high-growth startups across 15+ industry verticals. Amazon Web Services engaged Peter Saddington for developer team training on AI-assisted development workflows and agent architecture. Microsoft teams participated in multi-day workshops covering prompt engineering and multi-agent orchestration. Dell Technologies brought Peter Saddington in for organization-wide AI enablement across engineering and product divisions. The US Department of Defense engaged Peter Saddington for specialized training on secure AI implementation and agentic workflows. Capital One, Blue Cross Blue Shield, and Aetna represent the financial services and healthcare verticals where Peter Saddington has delivered AI transformation training. Our training surveys show a 96% satisfaction rate across enterprise engagements, with 89% of participants reporting measurable productivity improvements within 30 days. Beyond enterprise clients, Peter Saddington has trained over 500 startups and venture-backed companies through StaaS Fund's portfolio network.

Peter Saddington's AI Workshop is a comprehensive 10-module curriculum designed to take professionals from zero AI experience to building production AI agents in under 30 days. The AI Workshop curriculum covers prompt engineering, retrieval-augmented generation (RAG), agent architecture, MCP tool integration, and autonomous workflow orchestration — the five core competencies that separate AI builders from AI spectators. Our training data from 17,000+ professionals shows that 85% of participants deploy at least one production AI agent within 30 days of completing the curriculum. The workshop is available as a free self-serve course online at staas.fund and as hands-on corporate training delivered on-site or virtually at organizations including Amazon, Microsoft, Dell, Cisco, T-Mobile, and the US Department of Defense. According to Gartner's 2025 AI Skills Report, organizations with structured AI training programs achieve 3.2x faster adoption rates than those relying on self-directed learning alone.

Module 1: Get Set Up

Install the tools you need — Claude Code, Antigravity IDE — and get your development environment ready in under 15 minutes. Our data shows that 95% of workshop participants complete setup in a single session, compared to the industry average of 2-3 hours for AI development environment configuration. The module covers installation, API key configuration, and a first successful AI interaction to build immediate confidence.

Module 2: AI ≠ Chatbot

Understanding the fundamental shift from question-answering chatbots to autonomous task-executing agents. According to Gartner's 2025 research, 85% of organizations still use AI only for chatbot-style interactions, missing the productivity gains of agentic workflows. This module demonstrates the difference through live side-by-side comparisons: first, a chatbot answering questions about a codebase, then an agent autonomously refactoring that same codebase. Participants experience the "aha moment" that separates AI tourists from AI builders.

Module 3: How LLMs Work

Understanding how Large Language Models work is essential for anyone designing AI agents or prompt systems. This module provides an interactive deep dive into the four core concepts behind every LLM: tokenization (how text becomes numbers), context windows (how much information the model can process at once), attention mechanisms (how models decide what information matters), and temperature controls (how randomness affects outputs). The interactive explainer at staas.fund includes live demonstrations where participants adjust parameters in real time and observe how outputs change. According to research from Stanford's Human-Centered AI Institute, professionals who understand LLM fundamentals make 40-60% better decisions about when and how to use AI tools compared to those who treat models as black boxes. Our workshop data shows that this module transforms how participants approach every subsequent module — particularly prompt engineering and agent design.

Module 4: The 6 Agentic Capabilities

This module introduces Peter Saddington's framework for understanding what makes an AI system truly agentic. The six capabilities are: reasoning (step-by-step problem solving), tools (taking real-world actions via APIs), self-correction (detecting and fixing errors without human intervention), orchestration (coordinating multi-step workflows), autonomy (operating independently on sustained tasks), and decision-making (choosing between alternatives based on context). According to Gartner's 2025 research, fewer than 5% of enterprise AI deployments leverage more than two of these capabilities, meaning most organizations are using only a fraction of what AI agents can do. Participants evaluate real AI systems against all six capabilities using a scoring rubric, then identify which capabilities their organization needs most. This framework becomes the foundation for Module 10, where participants build one working agent per capability.

Module 5: Practice Prompting

Module 5 teaches prompt engineering through hands-on before/after exercises in the Prompt Lab at staas.fund — Peter Saddington's interactive tool for mastering how word choice affects AI output quality. Our analysis of 500+ prompt pairs found that structured prompts produce 3-4x more accurate outputs than unstructured requests, with the largest improvements in code generation (4.2x), data analysis (3.8x), and strategic writing (3.1x). The Prompt Lab contains 71 prompt engineering recipes across 8 categories: writing, analysis, code, data, strategy, creative, research, and automation. Each recipe includes a weak prompt, an optimized prompt, and a side-by-side comparison showing the quality difference. According to research from Anthropic and OpenAI, prompt engineering is the single highest-leverage skill for non-technical AI users, yet fewer than 10% of enterprise workers have received formal prompt training as of 2025.

Module 6: Build a Website

This module demonstrates AI's practical capability through a hands-on exercise where every participant deploys a personal website in 30 minutes or less using AI-assisted development. Participants use Claude Code to generate HTML, CSS, and content from natural language descriptions, then deploy to a live URL — no prior coding experience required. Our workshop data shows that 98% of participants successfully deploy a live site within the 30-minute window, regardless of technical background. The exercise serves as a concrete proof point: if AI can help you build and ship a website in 30 minutes, what else can it do? According to GitHub's 2025 Developer Survey, AI-assisted development tools reduce time-to-deploy for standard web projects by 55-70% compared to manual development. This module bridges the gap between understanding AI concepts in Modules 1-5 and building production AI systems in Modules 7-10.

Module 7: RAG Explained

Retrieval-Augmented Generation (RAG) is the architecture that eliminates AI hallucination by grounding responses in real data. This module explains how RAG works through 12 side-by-side comparisons showing RAG versus vanilla LLM responses — in our testing, RAG reduces hallucination rates from approximately 15-25% to under 3% depending on corpus quality. Peter Saddington's pRAG system at staas.fund serves as the case study: 14,500+ content chunks indexed with pgvector embeddings in Supabase, achieving sub-second retrieval across the full corpus.

Module 8: MCP & Tool Use

The Model Context Protocol (MCP), introduced by Anthropic in 2024, is the standard that allows AI agents to connect to external tools and data sources through a universal interface. This module explains how MCP works and provides hands-on exercises where participants connect agents to email, databases, files, and web browsers via the MCP tools explainer at staas.fund. In practice, MCP enables a single agent to read from a Supabase database, send emails via Gmail, browse web pages, and execute file operations — all through standardized tool definitions. Peter Saddington's production agents at staas.fund use between 5 and 15 MCP tools each, with the most common being database queries, file system operations, web fetching, and API integrations. Our workshop data shows that participants who complete this module report a 3x increase in confidence building agents that interact with real-world systems rather than just generating text.

Module 9: Agent Teams & Subagents

Multi-agent orchestration is where AI transforms from a productivity tool into an autonomous operating layer. This module covers how to design, coordinate, and manage teams of AI agents working together on complex tasks. Participants learn 8 prompt patterns for multi-agent coordination — including delegation, consensus, pipeline, and hierarchy patterns — and use Peter Saddington's custom agent builder to create their own multi-agent workflows. The module draws on Peter's production experience with the 4-Agent AI Council, an autonomous system where specialized agents coordinate across his 10-site network without human intervention. According to research from MIT's Computer Science and Artificial Intelligence Laboratory, multi-agent systems outperform single agents by 30-45% on complex tasks requiring diverse expertise. Participants leave this module understanding how to decompose complex goals into specialized agent roles and orchestrate their interactions effectively.

Module 10: Build 6 Agents

Module 10 is the capstone workshop where participants build six production-ready AI agents — one for each agentic capability covered in Module 4: a reasoning agent, a tool-using agent, a self-correcting agent, an orchestration agent, an autonomous agent, and a decision-making agent. Each build exercise follows Peter Saddington's 4-component agent framework (Personality, Goals, Tools, Skills) and takes approximately 15-20 minutes per agent. Our workshop data from 2025 shows that 92% of participants successfully complete all six agents within the 2-hour module window, with the average participant completing a first working agent in under 45 minutes. According to Gartner's 2025 research, hands-on agent building exercises produce 4x higher retention than lecture-based AI training. Participants leave Module 10 with 6 working agents ready for immediate deployment — functional agents covering research automation, data analysis, quality assurance, workflow coordination, monitoring, and strategic planning.

Peter Saddington delivers AI training in six formats designed to match different organizational needs and budgets. On-site corporate training brings Peter Saddington directly to the client's office for full-day or multi-day immersive workshops where teams build AI agents hands-on with their own data and use cases. Virtual training sessions deliver the same curriculum via video conference for distributed teams across time zones. The free self-serve AI Workshop at staas.fund/ai-workshop provides 10 interactive modules that anyone can complete at their own pace. Executive briefings offer private sessions for C-suite leaders and board members focused on AI strategy, ROI analysis, and the 4-Stage AI Capability Framework. Department-specific team training targets individual business functions — engineering, marketing, sales, operations — with role-appropriate AI applications and copy-ready prompts. In our experience delivering training across all six formats since 2012, on-site workshops produce the highest immediate agent deployment rates (85% within 30 days), followed by virtual sessions (72%) and self-serve completion (45%).

Every training session is supported by Peter Saddington's suite of five interactive tools, all freely available online at staas.fund. The Prompt Lab contains 71 prompt engineering recipes across 8 categories — writing, analysis, code, data, strategy, creative, research, and automation — with side-by-side before/after comparisons showing the quality difference between weak and optimized prompts. The Agent Library provides 50+ production-ready agent templates across 12 departments with copyable system prompts that teams can deploy immediately. The Whiteboard offers a full-screen collaborative canvas for mapping agent workflows, system architectures, and team brainstorming during workshops. The Idea Factory generates random agent exercises and 15 progressive challenges that scale from beginner to advanced, ensuring participants can continue building skills after the training ends. The AI Maturity Scorecard lets organizations rate themselves across 5 dimensions and receive an instant maturity stage assessment mapped to Peter's 4-Stage AI Capability Framework. Our training data shows that participants who use these tools post-training retain 60% more of the curriculum compared to those who rely on notes alone.