How Auctore's Speech Wizard Writes Speeches That Actually Sound Like You

The problem with using AI to write a speech isn't that the AI can't produce competent sentences. It's that a speech generated without enough personal input sounds exactly like what it is: a competent, generic, slightly lifeless piece of text that could have been written for anyone. You stand up at your best friend's wedding and read something that could have been written by a stranger who met them twice. Everyone can feel it.

Auctore's Speech Wizard is built around a different premise: that the quality of the speech is determined entirely by the quality of the input. The wizard collects enough personal, specific, contextual information that what comes out the other end sounds like it came from a person who actually knows the subject — because the material it's working with does.

What the Speech Wizard Collects

The wizard runs through a structured set of questions before generating anything. Each one is pulling material that will end up directly in your speech:

Who the speech is for. Name and basic context. Are they a colleague being honored at retirement, a sibling getting married, a friend receiving an award? The wizard needs to know who's at the center of this moment.

Your relationship to them. How you know this person, how long you've known them, and what the texture of that relationship is. "My college roommate" and "my business partner of twelve years" will produce speeches with very different emotional registers, even if the occasion is the same.

Key stories and memories. This is where most of the raw material comes from. The wizard asks you to describe specific incidents, shared experiences, or moments that capture something true about the person. "The time she drove four hours at midnight because I called her crying" is more useful than "she's always been there for me." The more concrete your stories, the better the speech.

The occasion. Wedding, retirement, memorial, graduation, roast, keynote, award ceremony? The occasion sets the emotional framework and the appropriate range of tone.

Tone. Formal, warm, or humorous — or a blend. A roast calls for something different than a farewell tribute. If you want 70% warmth and a few well-placed laughs, you can say that.

Rhetorical devices. This is where you can shape the craft of the speech itself. Do you want a recurring metaphor that runs through the whole thing? A callback to your opening at the close? A moment of self-deprecating humor? You can request specific devices, and the wizard will build them in rather than sprinkle them in accidentally.

Speech length. Two minutes at a podium, five minutes as best man, fifteen minutes as keynote speaker? Length affects structure, how many stories fit, and how much breathing room the speech has.

Opening and closing style. Some speeches open with a question; others open mid-story, already in the moment. Some close with a toast, others with a call back, others with a direct address to the honoree. Telling the wizard what you want keeps you from ending up with an opening you don't like.

The "Anything Else?" Field

At the very end of the wizard, there's a freetext field. This is where you add everything the structured questions couldn't capture — and it's where the speech stops sounding like AI and starts sounding like you.

Some things writers have added here:

• "Avoid sports metaphors — she hates them and it'll get a laugh if I visibly don't use one."
• "Include a callback to the Paris trip story I mentioned. That's the one people at this wedding will recognize."
• "My delivery style is dry. Don't write jokes that need big energy — write jokes for someone who delivers them flatly."
• "She's gone through a lot this year. I want the speech to acknowledge that without dwelling on it."
• "I cry easily. Build in a natural pause around the second-to-last story so I can collect myself."

Every one of those details changes the speech. The wizard reads the freetext field with the same weight as the structured questions — it doesn't treat it as an afterthought. If you want the speech to honor something specific, or avoid something specific, or land in a particular way for a particular audience, put it here.

After Generation: Open in Editor

Once the Speech Wizard generates your speech, you'll see an "Open in Editor" button. This saves the speech directly to your Auctore project as an editable document — formatted, structured, and ready for revision.

The generated speech is a strong first draft. It has the stories you gave it, the structure you asked for, and the tone you specified. But it's not finished. Your job now is to read it out loud, mark the places where it doesn't sound like you, and rewrite those lines in your own voice. You're not starting from scratch anymore — you're editing something that already has shape. That's a fundamentally easier task.

The editor supports everything you'd expect: full text editing, version history so you can return to the original, and access to the Speech Suite tools if you want AI feedback on pacing, delivery rhythm, or rhetorical effectiveness.

The Standard It's Trying to Meet

The test for a good speech isn't whether the sentences are grammatically correct. It's whether, when you're done, someone in the audience who knows you turns to the person next to them and says "that was so him." That recognition — of a specific human voice, a specific set of relationships, a specific way of seeing the world — is what the Speech Wizard is trying to make possible.

It can only get you there if you give it enough to work with. The more you put in, the more recognizably yours the output will be. Don't summarize. Tell the stories. Use the freetext field. Give it the real stuff, and the speech will return the favor.