Skip to main content
HHOLI

Field Notes · Issue 023 · 30 June 2026 · 6 min read

Feedback loops that make the AI sharper (not louder)

Not all feedback is training signal. Which corrections Holi weights heavily, which it ignores by design, and why the difference matters.

The trap most AI in recruitment falls into is treating every user action as equal training signal. That produces a model that gets louder every week, not sharper. Holi's learning loop is deliberately picky about what it learns from.

The signals Holi weights heavily

Rewrites. If you rewrite a draft substantially before sending, that is a strong signal about where the current model is wrong. Rewrites feed straight back into next week's drafts.

"Not this" rejections in the digest. When you tell Holi an observation is wrong, the underlying pattern gets marked down immediately. Three rejections on the same pattern kill it.

Rejected outreach recipients. If Holi suggested a person to reach out to and you said no, the reasoning behind that suggestion gets weighted down for similar profiles.

The signals Holi weights lightly

Opens. A good subject line does not mean a good message. Opens are useful in aggregate for subject-line patterns, worthless in isolation.

Accepts. A draft accepted unchanged might mean it was perfect. It might mean you were rushed, or the recipient was low-stakes, or you did not care enough to edit. Accepts are ambiguous; Holi treats them as such.

Time on page. Session-length metrics tell you nothing about whether the work was any good. Holi ignores them.

Why picky feedback matters

Because a model that treats everything as signal ends up drifting in whichever direction the noise points. Specialist recruiters need a model that changes when there is a real reason to change, and holds still otherwise. The learning loop is where the picky filtering happens, and the weekly digest is where you see what it decided.

The best feedback you can give Holi is a rewritten draft with a one-line note about why you rewrote it. That is worth more than a thousand opens. And Holi will show you exactly how that feedback moved the following week's defaults, section by section, in Monday's digest.

Sharper, not louder. That is what a working feedback loop looks like.

Enjoyed this? Get the next one on Monday.

Subscribe to Field Notes

One lead piece, every Friday. Specialist recruiting, intelligence, and the work itself. One-click unsubscribe in every issue.

Also building a desk in 2026? Join the Cohort 01 waitlist , 33 companies, founding pricing locked for 12 months.

Asked, and answered

Why not weight opens and accepts heavily?
Because they are noise. Opening a message means the subject worked, not that the message did. Accepting a draft unchanged might mean it was perfect or might mean you were rushed. Rewrites and rejections carry real signal.
Can I add my own feedback signal?
Yes. Any note you add to a rejected draft or a corrected observation is weighted alongside the rejection itself, and shows up in the following Monday's digest.

About the author

James Doyle

Founder, HOLI

James founded HOLI to build a partner for specialist recruiters — the kind that captures the intelligence you gather and turns it into the action that wins the role. Twenty years in specialist search across healthcare, behavioural health, and creative.

LinkedIn ↗

Share

Cite this piece

James Doyle (2026). Feedback loops that make the AI sharper (not louder). Field Notes, HOLI. https://hey-holi.com/field-notes/feedback-loops-that-make-ai-sharper

Published 30 June 2026.

More from Field Notes

Low commitment

Get Field Notes weekly

One signal, one draft, one thing Holi learned. No spam.

Subscribe to Field Notes

One lead piece, every Friday. Specialist recruiting, intelligence, and the work itself. One-click unsubscribe in every issue.

Also building a desk in 2026? Join the Cohort 01 waitlist , 33 companies, founding pricing locked for 12 months.

Ready to build with us

Apply for Cohort 01

33 companies. Founding pricing locked for 12 months. 60-second application.

Start application