Your speech is AI-generated
The text ToastPetal produces is generated by a large language model from the answers you provide. It is not written by a human speechwriter, and we do not claim otherwise anywhere on this site.
What is human is the design: the interview questions, the structure rules, the etiquette for each role, the banned clichés and the prohibited topics were all written by people who have thought hard about why wedding speeches fail.
What the model is given
Only what you typed into the intake form, plus our instructions. It has no access to your social media, the couple's names beyond what you provided, or any other source of information about you.
What can go wrong
- It can misread a memory and place the emphasis in the wrong spot.
- It can smooth a detail in a way that makes it subtly inaccurate.
- It can produce a sentence that is not how you talk.
- Despite explicit instructions not to, it can occasionally generalise in a way that reads as a claim you did not make.
This is why manual editing is unlimited and free, and why we tell everyone to read the speech out loud three times before delivering it.
Is using AI for a wedding speech acceptable?
We think so, with one condition: the material has to be yours. A speech built from your own memories, shaped by a tool and then edited in your own words, is not meaningfully different from one written with a friend's help. A speech generated from nothing, delivered unread, is a different thing — and it shows from the microphone.
Whether to tell people is entirely your call. Most speakers do not, for the same reason they do not mention which book they read about public speaking.
Our own content
The guides, role pages and example speeches published on this site are written and reviewed by us. The example speeches are original and were written for this site; they describe invented people and are provided as models of structure, not as real speeches from real weddings.