> For the complete documentation index, see [llms.txt](https://themochi-app.gitbook.io/documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://themochi-app.gitbook.io/documentation/module-4-ai-setup-and-intelligence/training-center.md).

# Training Center

## Training Center: Review Your AI's Annotations

**What it is** — Beyond drafting replies, Mochi's AI continuously annotates every conversation in the background — flagging links, objections, sentiment, scheduling signals, and engagement level. The Training Center is where you review those annotations and correct the ones it gets wrong, which is how the AI keeps getting sharper for your organization specifically.

**Who can use it** — Expected to be a **Creator/Manager** area, consistent with the rest of Settings and Intelligence.

**Before you start** — Enough conversation history to have generated annotations in the first place. See the data benchmark in Set Up Your AI.

### What the AI annotates

Mochi's AI tags conversations behind the scenes across five categories:

* **Link Detection** — classifies a link sent in conversation as Booking, Payment, Qualifier, Nurture, VSL, or Other.
* **Objection Detection** — classifies an objection as financial, timing, authority, self-confidence, trust, competitor, or other. This only applies **after a CTA has been sent** — a question asked before any offer is on the table isn't counted as an objection.
* **Sentiment Detection** — reads the lead's tone as Positive, Neutral, Negative, or Frustrated.
* **Scheduling Detection** — flags whether a call time was Proposed or Confirmed.
* **Engagement Scoring** — a 0–100% score based on message length, questions asked, voice notes sent, and sentiment.

### Review an annotation

1. Open a conversation that has AI annotations.
2. Read the AI's tag or classification.
3. Mark it:
   * **Correct** — the AI read the conversation accurately.
   * **Incorrect** — provide the correct answer yourself.
   * **Partially correct** — the AI was on the right track but not fully accurate.

### Golden Scripts

A related view surfaces your best-performing openers, qualifiers, and qualification questions, along with an explanation of why they worked — useful for spotting what to standardize across your team.

### Why this matters

Reviewing annotations isn't busywork — it's the training signal that improves the AI's judgment for your specific business over time. The more consistently your team corrects mislabeled objections or sentiment, the more accurate future AI Tags, Lead Summaries, and reply suggestions become.

### Tips

* Review annotations regularly rather than in one big batch — little and often keeps the AI's understanding current with how your conversations actually go.
* Pay close attention to Objection Detection accuracy — it directly affects how well the AI Setter handles pushback in Away Mode and AI Mode.
* Use Golden Scripts as a training resource for new setters, not just an AI-feedback tool.

### Common mistakes & fixes

| Mistake                                                          | Fix                                                                                         |
| ---------------------------------------------------------------- | ------------------------------------------------------------------------------------------- |
| Ignoring the Training Center entirely                            | Set aside regular time to review — it's how the AI improves specifically for your business. |
| Marking something "Incorrect" without providing the right answer | Always fill in the correct answer — that's the actual training signal.                      |
| Assuming objection detection applies to any pushback             | It only counts after a CTA has been sent — earlier questions aren't objections.             |

### Limits & good to know

* Gated behind the same conversation-data benchmark as the rest of the AI Setter (see Set Up Your AI).

**Related:** Set Up Your AI: Setter Playbook & Creator DNA, AI Setter: Suggestion Pills, Away Mode & AI Mode, AI Tags.