> For the complete documentation index, see [llms.txt](https://docs.voicemoat.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.voicemoat.com/voice-and-audenai/claim-bank-and-learning-from-edits.md).

# Claim bank and learning from edits

**In short:** your claim bank, now called **Your stories and views**, is what you actually say, pulled from your own posts: the views you repeat, the stories you reuse, and the framings you avoid. You review it in Voice Lab, and audenAI writes with it. audenAI also learns from the changes you make to its drafts, and from the posts you mark as most like you. All of it is per profile, on every paid plan and the trial.

#### Where to find it

Voice Lab → **Teach audenAI** → **Your stories and views**. The rules audenAI learns from your edits are next to it, in **Your rules**. Each profile has its own, because you may say different things on Twitter and on LinkedIn.

#### What is in it

Your stories and views is one list. Each item carries a tag:

| Tag       | What it holds                                                                       |
| --------- | ----------------------------------------------------------------------------------- |
| **View**  | Opinions you come back to.                                                          |
| **Story** | Examples and numbers. audenAI only tells these as yours, never new ones.            |
| **Avoid** | Ideas, not words, that you never write like. They only count once you confirm them. |

Each item audenAI found shows the exact words from your post. Tap the source chip (the post and its date) to see the quote, with a link to open the post on Twitter or LinkedIn. Items you add by hand have no source post, so they show **Added by you** instead, and an item you edit uses your new wording. Views and stories that several of your posts back are marked **Strong**.

<figure><img src="https://708994200-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FpIMRJgVWZK5TkQX6SclK%2Fuploads%2Fgit-blob-83a07384d3d17bfc675d86d39b31244159748401%2Fstories-and-views.png?alt=media" alt="Your stories and views in Voice Lab: one list with items tagged Story, View, and Avoid, each with the source post and its date, the All, To review, and Pinned filter, and the Sensitive topics switch"><figcaption><p>Your stories and views: one list, each item tagged Story, View, or Avoid and linked to the post it came from.</p></figcaption></figure>

#### How it is built

When you build or rebuild your Profile DNA, audenAI reads your own posts and pulls out views and stories. It keeps one only if its quote is found word for word in your post. New items wait in **To review**. Building the list costs no extra credits.

Politics, religion, and health stay out of your list and out of audenAI's drafts unless you turn **Sensitive topics** on, at the bottom of the card.

#### Reviewing it

Filter the list with **All**, **To review**, or **Pinned**. It shows 8 items first, then **Show more**. For each item you can:

* **Keep** it (or **Confirm** a framing you avoid),
* **Pin** it, so audenAI puts it first,
* **Edit** it into your own words (audenAI uses your wording exactly),
* or choose **Not me** (**I say this** on a framing you avoid): it is removed and never suggested again.

**Keep all strong** keeps every Strong item waiting for review in one go. **Add a story, a view, or a topic to avoid** adds one by hand, and audenAI uses it right away. On a computer, the keys **K**, **P**, **E**, and **N** keep, pin, edit, or remove the item you are on.

Pinned, edited, and removed items survive retraining. After a retrain, a short "Updated after your last re-train" note shows what moved.

If your list is still thin, give it a head start: add a view you always share, or use **Pick 3 posts that read most like you** at the bottom of the card.

#### Mark the posts that read like you

On Voice Lab's **Your voice** tab, open a post from the **Your posts** graph, or from **See all posts**. Each post has a **Mark this post** menu:

| Mark                  | What it does                                                                                               |
| --------------------- | ---------------------------------------------------------------------------------------------------------- |
| **This is the voice** | audenAI leans on this post from your next draft, and reads it first when it builds your stories and views. |
| **Bad day**           | Left out of training and out of your stories and views.                                                    |

#### Learning from your edits

When you change a draft audenAI wrote and then post, schedule, or save it in Studio, the before and after are kept so audenAI learns what you change. Nothing pops up for this.

After a big edit, at most once a day, the confirmation asks one question: "You changed a lot of audenAI's draft."

* **This is me**: your version says what you think. audenAI may suggest a view or story from it, which waits in To review.
* **This is not**: audenAI learns from what you changed.

#### Your rules

Once your edits point the same way, a plain rule shows up in **Your rules**, under **audenAI noticed**, such as "Cut the closing question." For each one you can:

* **Keep** it, and it joins your rules tagged **From your edits**,
* edit it first,
* **Dismiss** it: it is never suggested again,
* or open **See the edits** to see the changes it came from.

<figure><img src="https://708994200-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FpIMRJgVWZK5TkQX6SclK%2Fuploads%2Fgit-blob-1b436aef971d03c323cc24b9c64ec89c1c5f59a0%2Fyour-rules.png?alt=media" alt="Your rules in Voice Lab: never-say words, rules tagged You set and From your edits, an audenAI noticed suggestion with Keep, Dismiss, and See the edits, and the Learn from my edits switch"><figcaption><p>Your rules: everything audenAI follows, with a rule it noticed in your edits waiting for Keep or Dismiss.</p></figcaption></figure>

The **Learn from my edits** switch at the top of Your rules pauses learning. While it is paused, edits are not kept, and the rules you kept still apply. Learning from edits costs no credits.

{% hint style="success" %}
Your stories and views and your edits belong to that one profile. They are never shared, and never used to train shared models. See Security and your data.
{% endhint %}

#### Who can use it

|                          | Free | Trial | Creator | Pro | Enterprise |
| ------------------------ | ---- | ----- | ------- | --- | ---------- |
| Your stories and views   | no   | yes   | yes     | yes | yes        |
| Marking posts            | no   | yes   | yes     | yes | yes        |
| Learning from your edits | no   | yes   | yes     | yes | yes        |

#### Common questions

**Will audenAI put words in my mouth?** It writes with your stories and views, and personal stories and numbers come only from that list or from what you tell it. A draft that goes against one of your views shows as **Off voice** in Stance. See Style and Stance.

**It pulled a view I no longer hold. What do I do?** Tap **Not me**. It is removed and never suggested again.

**Does retraining wipe my choices?** No. Pinned, edited, and removed items stay as you left them.

**Where did the claim bank go?** It is the same list, renamed **Your stories and views**, in the Teach audenAI tab. Claims are now called views.

**How is this different from Memory?** Memory holds facts from your chats with audenAI (goals, projects, preferences). Your stories and views holds what you say in your posts.
