--- title: 'The decision journal: how every reply you reject trains your AI team' description: 'An edit shows where the restaurant''s real boundary sits. See how a readable, portable decision record improves the next version.' url: 'https://www.semperi.com/blog/decision-journal-train-your-ai' --- # The decision journal: how every reply you reject trains your AI team > An edit shows where the restaurant's real boundary sits. See how a readable, portable decision record improves the next version. *May 21, 2026 · 8 min · AI Operations · Semperi Team, Restaurant growth research* A restaurant's menu, sales history, and reviews describe what happened. The owner's edits and rejections describe judgment. When an owner says "no, not like that," the useful signal is the boundary a general-purpose model could not know on its own. At Semperi we call the system that captures this the decision journal. Every action your AI employees take or propose — and every keep, edit, or reject decision — is written to a permanent, readable, exportable log. This post explains how that journal works, why specific edits beat silent acceptance, and what it means in practice over time. ## What is a decision journal in an AI system? Mechanically, it needs three things. First, a readable log of each version, its available sources, and the restaurant's decision. Second, usage controls that keep cost and volume bounded. Third, a feedback record the next version can use instead of asking the owner to repeat the same correction. The first two are about safety; you can read how they fit into Semperi's broader trust architecture at https://semperi.com/trust. The third is about something else entirely: compounding. ## Why are edits and rejections so valuable? Keeping a version tells the system "this was acceptable." Useful, but weak. A rejection or an edit shows exactly where the boundary of your taste sits, and boundaries are where learning happens. - You reject a reply for being too apologetic → Grace learns your house style is warm but unapologetic, and stops opening with "We're so sorry." - You edit "customers" to "guests" three times → it stops happening a fourth. That's not a model retrain; it's your journal becoming part of the writing context. - You reject a comp suggestion on a borderline complaint → the system learns where your generosity line sits, and stops proposing comps below it. > A generic AI gets the same answer for every restaurant. A journaled AI gets your answer — because it has read every decision you've ever made inside it. ## How is this different from "the AI learns your brand voice"? Every AI tool claims it learns your voice. Usually that means you filled in a settings form once — "casual, friendly, no emojis" — and a prompt was generated. That's a snapshot. The decision journal is a stream. Your voice in March after a rough health-inspection scare is not your voice in July when the patio is full. A settings form can't track that. A journal of two hundred real decisions can. There's also a trust difference. Because every Semperi version cites its sources (the cite-or-die rule — no fact without a citation to your real data), the journal isn't just "what the AI said." It's "what the AI said, what evidence it said it from, and what you decided." When you reject a version, the system can tell whether you rejected the facts, the tone, or the judgment — three very different lessons. ## How should the journal improve over time? - At the start, expect to correct tone, wording, and assumptions that only the owner knows. - Repeated corrections should become less common only when the system can read and apply the decisions already recorded. - Over time, the journal should make strategic boundaries clearer, but no fixed timeline or automation rate is guaranteed. Switching tools should not erase that learning. The journal belongs to the restaurant, exports as readable structured data, and should move with the restaurant. Portability keeps the accumulated record useful without turning it into vendor lock-in. ## Where do you actually see all this? Two places. The audit log itself is always available — every action, every citation, every decision, exportable. And the digest version arrives every morning in your Daily Brief (https://semperi.com/daily-brief): what your AI team did, what needs a decision from you, and what it learned from yesterday's decisions. You answer each request from your phone in seconds, because the whole system only compounds if deciding stays cheap. For the highest-stakes example — public review replies — start with our source-backed reply breakdown (https://semperi.com/blog/ai-review-replies-with-human-approval). The decision journal is the learning record underneath it. --- Semperi gives your restaurant a team of AI employees working 24/7 to grow your brand, sales, and profit. - Full site map for agents: https://www.semperi.com/llms.txt - Free restaurant visibility scan: https://app.semperi.com/scan - This page for humans: https://www.semperi.com/blog/decision-journal-train-your-ai