There is far more technical material available today than any person could consume in a lifetime: for almost any subject you’ll find dozens of YouTube videos, multiple Substack posts, GitHub repositories and Reddit threads, many published within the last six months and often technically accurate. Yet most practitioners I speak with say they don’t know whom to trust. They can’t tell what to read first, or which of ten plausible answers truly holds up. That uncertainty existed before AI, and AI has amplified it.
Editing, verification and the old model
For most of technical publishing’s history, editing and verification were a single, slow, and costly process. Publishing a book could take years: finding an author, vetting them, working with an editor, and having technical reviewers check claims. Much of that time was spent separating what was correct and useful from what was confusing or merely plausible. The work was laborious, but it meant readers could rely on claims on the page — the credibility of the book and the publisher mattered as much as the content.
When production costs fall toward zero, that credibility becomes more valuable, not less. Content is easier to produce than ever, but without a transparent process behind it readers have no idea where the knowledge came from or whether it holds up. As Jasmine Sun writes in "The Independent Writer’s Advantage in the Age of AI," trust isn’t only about information quality — it’s about the messenger: who says it, their track record, and what they’ve told you before. A practitioner trusts a source because someone she respects has staked reputation on it; because a publisher has a history of being right and of correcting itself; and because the work is attributable and verifiable.
The corpus matters, but so do the assurances
The body of content matters, but the assurances around it are difficult to replicate. Those assurances come not just from creators but from people whose judgment vouches for them. Sometimes creators bring credibility with them; other times a publisher lends reputation to an unknown author. Dave Hickey made this point about gallery owners in Air Guitar: they gain status from the famous artists they represent and share it with emerging talent. That is what O’Reilly has done for nearly half a century — building a network of experts who vouch for what’s worth knowing.
Expertise is alive and compounding
Expertise is a living thing. Content begins to decay the moment it’s published: frameworks change, libraries deprecate, and yesterday’s best practice can become today’s security incident. Keeping expertise current requires a pipeline of practitioners who stay up to date and an editorial layer that notices when something is stale and either retires it or calls for fixes.
That pipeline isn’t something you flip on when an author has a book to ship. O’Reilly says it has long sought what Tim O’Reilly calls the “alpha geeks” and spread their knowledge. They think about content in pace layers: some advice is timeless; some shifts slowly (some O’Reilly books are still in print after nearly 50 years); and some topics change weekly. They maintain relationships with hundreds of top practitioners and keep them engaged continuously with quick takes when something breaks, structured responses to major research, and live sessions on emerging topics.
Trust is bidirectional
An institution doesn’t unilaterally stamp trust onto content. In a technical community, trust is conferred both ways: a practitioner earns standing because others who already have standing engage with, cite, argue with, and build on their work. That was the insight behind PageRank: a page mattered because other important pages linked to it. Reputation works similarly.
Audiences don’t just consume reputation signals; they generate them. When a senior engineer whose judgment others respect publicly says something is worth reading, she spends some of her own credibility; the author gains a bit; and observers recalibrate whom to trust next time. When O’Reilly puts its mark on a work, it amplifies a community judgment and adds its own track record. A reader who finds the work reliable hands status back to the source.
When the readers are machines
It’s not only human practitioners who need trusted engineering knowledge. AI systems embedded in workflows — coding agents, debugging assistants, architecture advisors — need it just as much, because many are trained on scraped web data and documentation that was stale before it was indexed. These models are fluent but wrong often enough that you can’t accept their output uncritically.
The stakes rise as AI is used to generate not only provably correct outputs like code (which works or doesn’t), but persuasive documents in fuzzier domains like hiring or strategy. O’Reilly is confronting the consequences of being able to ask a model and get back something that looks smart at a glance: after a few rounds, the slop is still there. In the last few months, maybe ten times as many documents have crossed their desks — from new product ideas to strategic plans and proposals — but the ease of generating text masks the fact that either the model or the prompter doesn’t actually understand the subject.
Knowledge workers need ways to ground their output in human expert insight, particularly when AI assists the work. To that end, O’Reilly is building tools that let agents draw on its repository of expertise to support proposed decisions.
Credible sources matter for decision-making
Credible sources are especially important when thinking through and justifying significant choices. Andrew Odewahn, O’Reilly’s CTO, describes the shift: “Eighteen months ago, it was all about how to get engineers to be more productive; now it’s about how to get organizations to make better decisions. Engineering tasks are moving away from coding output to planning.” For planning tasks such as comparing implementation approaches, you need expert-over-your-shoulder guidance for contextual decision-making. You can’t rely on an LLM’s best guess alone, which is why there’s perceived opportunity for products like O’Reilly’s Expert Intelligence that deliver grounded knowledge embedded in AI tools and workflows. Trust is foundational because the expertise behind it remains human, practical, and verifiable.
