A field guide is defined as much by what's excluded as what's in it. Five things we left out on purpose.
The most requested thing we don't print. Comp data for AI roles has a short half-life, and a printed number is wrong within a quarter in a way that damages the reader rather than helping them. Someone budgeting a 2027 role against a figure published in mid-2026 will lose candidates and not know why.
We have this data because we run these searches. It just belongs in a conversation with a date attached, not in a PDF that circulates for two years.
AI in hiring is regulated differently across jurisdictions, and the rules are moving. Bias auditing requirements, disclosure obligations and restrictions on automated decision-making vary by state and by country, and anything we wrote would be partially stale on publication and dangerous if treated as advice.
What we will say: the four-level automation framework in Chapter 4 maps roughly onto increasing regulatory attention. If you're at assisted evaluation or above, that's the point to involve counsel rather than the point to check whether you should have.
This is the largest omission and it deserves its own guide.
The Enhanced tier described in 2.3 is mostly an internal development problem rather than a recruiting one. Most of the AI capability a mid-size organization needs already sits in the building, in people who could fold these tools into their work if anyone showed them how. We touched on the training side in 3.3 and left the harder question, how to move people between roles as work changes, largely alone.
We left it out because it's a different discipline with different owners, and treating it as an appendix to a hiring guide would have done it badly.
Adjacent, deep, and usually owned by people who weren't the intended readers. Data readiness appears repeatedly in the guide as a failure mode, and we describe it as a staffing gap rather than working through the architecture. That's a real limitation of the document.
The honest answer is that we don't know yet.
There's a plausible argument that AI assistance compresses the junior tier, since a lot of entry-level work in technology and analysis is exactly what these tools do adequately. There's an equally plausible argument that it expands, because a junior person with good tooling now produces at a level that used to require more experience.
We see evidence for both in the searches we run, and not enough of either to call it. Writing a confident chapter on this would have been the least defensible thing in the guide.
Anything here you'd want covered properly? Tell us, it shapes the next one.