Brightpool
Prompt engineering with production rigor
We’re an AI engineering studio working on our own products, sharing what we’re learning in educational courses and training. We’ve been prompt engineering since the GPT-3 beta in 2020, and we spend all our time experimenting with the latest AI tools and techniques.
Purpose-built AI partners
Strategy, research, and production build-outs for ambitious teams that want reliable AI systems—not just clever demos.
Translating experiments into durable products & education
We publish what we learn, ship our own tools, and document the prompts, evals, and systems so others can reproduce them.
Also True for Humans
Mike’s monthly column at Every unpacks how AI shifts business and marketing strategy with real operators.
Prompt Engineering for Generative AI
Published by O’Reilly in June 2024, the definitive playbook for building production-grade AI workflows.
AI Vision for UGC Game Ads
We watch hundreds of hours of gameplay footage to auto-select the perfect, high-performing ad clips.
Prompt Engineering Principles
A non-technical, free Udemy course that teaches prompt fundamentals for teams shipping content daily.
Complete Prompt Engineering Bootcamp
100,000+ students trained on hands-on generative AI workflows with Mike & James guiding the builds.
Vexpower AI Growth Marketing
Subscription courses turning marketing teams into AI-first operators with reusable playbooks.
Hands-on operators building the future of prompt engineering
We combine a decade of growth marketing with deep experimentation in generative AI to help teams ship more reliable automation.
Mike Taylor
Co-founder, AI strategy & growth
Co-founded Ladder, a 50-person marketing agency serving Booking.com, Monzo, and Time Out. Over 400,000 learners across LinkedIn Learning, Udemy, and Vexpower, co-author of the O’Reilly prompt engineering book, and contributor to Reforge, Every, and Lenny’s Newsletter.
James Phoenix
Co-founder, AI engineering & data systems
Builds reliable data pipelines that automate thousands of growth tasks. Delivered 50+ General Assembly data science bootcamps and partners on Udemy, Vexpower, and the O’Reilly prompt engineering book.
Built with deep prompt engineering experience and production rigor
Each engagement blends experimentation, measurement, and documentation so teams inherit working systems and the knowledge to evolve them.
Non-Technical Training
Workshops that give your team prompt patterns and workflows so they automate confidently without needing code.
Tooling we use
Prompt Optimization
Rapid experimentation loops to improve eval scores, latency, and cost across your AI application.
Tooling we use
AI Applications
Design and delivery of production-ready AI tooling, from prototypes to cloud deployments with observability built in.
Tooling we use
We love difficult problems and ambitious briefs
Share the context, the blockers, or the hypotheses you want to explore. We’ll help validate what’s possible and design the experiments to prove it.
Business enquiries
Start a project or request training via LinkedIn.
Talk AI with us
Ideas, experiments, and hot takes welcomed on X.
How we respond
Bring us your thorniest workflows, evaluation puzzles, or prototype ideas. A short note, Loom, or doc with your goals is all we need to share a recommended approach, next steps, and what to tackle first.
Every message gets a thoughtful reply from Mike or James—usually within a day—so you can stress-test plans before investing the build time.
“Prompt Engineering for Generative AI” — published by O’Reilly, June 25th 2024
A practical guide to designing prompts, evaluations, and agent workflows that survive rapid platform changes. Covering text, image, and Python automation, it teaches how to build context-rich systems that scale.
Everything changes quickly in AI, so we distilled the principles that transfer across models and modalities. They worked with GPT-3, still work with GPT-4, and prepare teams for GPT-5 and beyond.
The Five Principles of Prompting
- •Give Direction: Describe the desired style in detail or reference a persona.
- •Specify Format: Define the structure, rules, or schemas the model must follow.
- •Provide Examples: Seed the prompt with correct test cases to anchor behavior.
- •Evaluate Quality: Score outputs and surface failure modes to improve reliably.
- •Divide Labor: Chain steps and roles so complex goals stay manageable.
Who the book is for
- •You value your time above $40/hour and want a distilled field manual instead of scattered tips.
- •You’re building AI products or internal templates used by teams hundreds of times a day.
- •You need strategies for reducing hallucinations and raising reliability with real-world case studies.
- •You want a comparison of OpenAI against emerging models, frameworks, and vector databases.
- •You’d like an end-to-end walkthrough from a naive prompt to a shipped AI agent with UI prototypes.
“The absolute best book-length resource I’ve read on prompt engineering. Mike and James are masters of their craft.”
Dan Shipper, Cofounder & CEO, Every
“If you’re looking to improve the accuracy and reliability of your AI systems, this book should be on your shelf.”
Mayo Oshin, Founder & CEO, Siennai Analytics
The book spans ten chapters, covering both text and image prompting alongside Python automation. It’s not a collection of prompt “hacks”, but a systems-first approach to giving models the right context, testing the results, and scaling what works.