News 26 April 2026 3 min read A book for the people who have to decide By Old Forge Technologies 83 reads Contents <p class="lede">Your roadmap now says <em>AI agents</em>. Engineering is quoting months. A vendor on LinkedIn is quoting weeks. Both can't be right - and somebody on your side of the table has to work out who is.</p> <p>Today I'm publishing <em>AI Agents for Everyone: From ChatGPT to Autonomous Systems That Actually Work</em> on Kindle. It's the book I wish had existed two years ago, when the first wave of "let's just bolt an agent onto it" requests started landing on product managers who'd been given roadmaps but no map.</p> <h2>The gap I wrote it into</h2> <p>Most AI books teach engineers how to build. Strategy books tell executives why AI matters. Neither helps the product manager in the middle - the one who already owns the craft (MVPs, stakeholder management, business cases) but now has to make credible calls on questions like:</p> <ul> <li>Which model tier actually justifies its cost on this workload?</li> <li>When does memory add genuine value, and when is it a liability dressed up as a feature?</li> <li>Is this a job for N8N or Make, or does it warrant custom development - and where exactly is the line?</li> <li>What level of safety investment does the actual risk warrant? (The same compliance theatre applied to every agent <em>is itself</em> a form of risk.)</li> </ul> <p>You can answer these without learning transformer maths. You cannot answer them by reading marketing copy. The book sits in that gap.</p> <h2>What's inside</h2> <ul> <li><strong>Agent architecture for decisions, not implementation.</strong> Memory, tools, reasoning, orchestration - the trade-offs that matter and the ones that don't.</li> <li><strong>ROI frameworks with realistic numbers.</strong> Break-even ranges, conservative volumes, no marketing projections.</li> <li><strong>Low-code vs. custom.</strong> Specific thresholds, not vibes.</li> <li><strong>MCP, in plain English.</strong> Why integration timelines have collapsed from weeks to days for mainstream tools - and when you should still build custom anyway.</li> <li><strong>Production operations.</strong> Monitoring, cost control, quality degradation, recovery patterns.</li> <li><strong>Legal, ethical, privacy and security</strong> mapped to current UK and EU regulation.</li> <li><strong>A clear-eyed chapter on AI slop</strong>, and why quality is a moat rather than a cost.</li> </ul> <h2>What it isn't</h2> <p>It isn't a tutorial for engineers writing MCP servers from scratch. It isn't a one-hour executive briefing. And it is decidedly not breathless hype about AGI in eighteen months. The tone is British, direct, occasionally witty, never patronising. Pricing is in pounds. Examples are honest about whether they're documented cases, illustrative scenarios, or composites - no manufactured citations, no generic "Fortune 500 retailer" waffle.</p> <h2>Staying current</h2> <p>Agent technology moves in quarterly cycles. Printed books don't. The contents are current as of April 2026, and the <a href="/book-updates">companion page</a> is maintained quarterly with updated model lineups, pricing, and ecosystem changes. Your purchase includes lifetime access to that page.</p> <h2>Where to buy</h2> <p><a href="https://www.amazon.co.uk/dp/B0GYM7MX72/" rel="noopener" target="_blank">AI Agents for Everyone - Kindle Edition on Amazon UK</a>. If you read it and disagree with something in it, I genuinely want to hear about it - there's a contact form at the bottom of every page on this site.</p> <p class="signoff"> - Andrei</p>
