Event Speech Writer is a Claude AI skill — finds the real story in every occasion, every relationship, every room.
Imagine you're the maid of honour. You've known the bride for twelve years. You were there when she got the text from him after their first date — the screenshot is still in your camera roll. You know exactly why they work, and you know the specific, embarrassing, wonderful story that proves it. You open an AI tool and type: "Write me a wedding toast for my best friend." What comes back is technically a wedding toast. It's warm. It's structured. It thanks everyone for coming. It says the couple complement each other perfectly. It could have been written for anyone, by anyone, about anyone. And you know it.
You delete it and start again. You try different phrasing. You get a different speech that has the same problem. Eventually you spend two hours writing something yourself that's better than either AI draft — which was the whole reason you tried AI in the first place.
This failure has a name. It's not a quality problem. It's a context problem.
The Structural Failure Behind Every Flat Speech
When you ask a general-purpose AI to write an event speech, you typically give it four pieces of information: the occasion, the person's name, your relationship, and a vague adjective or two. "Write a retirement speech for my manager David, who is kind and has worked here for 20 years." That's it. The AI has to build something on almost nothing.
What it builds is structurally correct and emotionally inert. It knows the formula — open warmly, anchor with a memory, close with a wish. The problem is it has no real memory to anchor to, so it invents a placeholder: "I remember when David first joined the team..." and then says something so generic that David himself might not recognise it.
Generic AI doesn't fail at structure. It fails at specificity — because specificity requires context it was never given. You can't extract a real story from a name and a job title.
The people in the room listening to your speech already know David. They know he's kind. They know he's been there for twenty years. Telling them that adds nothing. What they don't know is the moment you're thinking of — the specific afternoon he stayed late to walk you through the system when you'd just started, and then made you both tea, and then told you the thing he told you that you've thought about ever since. That's the speech. That's what makes people look up from their plates.
Why the Same Failure Happens Every Time
It doesn't matter which AI tool you use or how many times you rephrase the request. The problem isn't in the generation — it's in the input. You're asking the AI to write something personal with impersonal ingredients. Warm + 20 years + kind = a Hallmark card, not a speech.
The material for a great speech already exists — it's in your memory. The problem is that no one asked you the right question to get it out.
This is the specific gap that most people don't diagnose. They assume the AI draft is bad because AI is bad at writing speeches. But when you read their draft, you can usually see the speech that could have been — it just needed one real detail to come alive. The AI wasn't missing talent. It was missing information. And it never asked for it.
The result is a pattern that repeats across every kind of occasion speech. Wedding toasts that describe the couple's "beautiful love story" without naming a single moment of it. Retirement speeches that enumerate a colleague's qualities without showing any of them in action. Birthday tributes that could apply to every forty-year-old who has ever existed. The structure is there. The person isn't.
What Happens When the Right Questions Get Asked
The Event Speech Writer skill approaches the problem from the opposite direction. Before generating anything, it interviews you. Three questions, targeted to your specific occasion and your specific relationship — because what a best man needs to surface is completely different from what a daughter needs to find for her father's retirement.
The questions are designed to get past adjectives and into scenes. Not "describe what makes them special" — that produces adjectives. But "tell me about the moment you knew they were going to be okay" — that produces a scene. Scenes are what speeches are made of. Adjectives are what speeches fall back on when scenes aren't available.
| What generic AI asks for | What the Event Speech Writer asks for |
|---|---|
| "Describe the person in a few words" | A specific moment that shows that quality in action |
| "What's your relationship to them?" | The context of how you know each other — including the awkward or surprising parts |
| "What do you want to say about them?" | The thing you've never said out loud that the occasion finally gives you permission to say |
| "What tone do you want?" (formal/funny/heartfelt) | What the room is like — size, formality, relationship to the subject — so the tone is calibrated, not chosen from a dropdown |
The difference in output is significant. A speech built from a scene — the moment they chose to stay, the call they made at midnight, the thing they did that you didn't expect — has something to hold onto. An audience remembers it because it gave them something specific to picture. A speech built from adjectives gives them nothing to hold onto, so they hold onto nothing.
A Retirement Speech, Written Both Ways
Same occasion. Same person. The speech is for Margaret, who is retiring after 27 years as the operations lead at a logistics company. She's known for being calm under pressure and for remembering everyone's birthdays.
The first speech says Margaret is steady and warm. The second one shows it happening — a specific afternoon, a specific crisis, a specific detail that only someone who was there could know. The first speech could be read at any retirement by anyone. The second one is about Margaret, and Margaret alone, and the room will feel that.
Who Actually Needs This Kind of Help
Anyone who knows what they want to say but can't find the shape to say it in. Best men and maids of honour who've tried AI and felt the flatness. Colleagues with a genuine story who need help getting it onto the page in a form they can actually stand up and deliver.
The skill works best when there's real material to work with — a relationship, a memory, something the person being honoured actually did. It can't manufacture specificity from nothing, and it won't try to. But if you have a genuine connection to the occasion and at least one real moment to build from, the skill will find it and build the speech around it.
The underlying premise of the Event Speech Writer skill is that the right speech already exists in your experience — it just needs the right question to pull it out. Most people have the material. They just haven't been asked for it in the right way.
The speeches people remember from weddings and retirements and send-offs are almost never the most eloquent ones. They're the ones that contained a detail so specific, so clearly true, that it made the room feel like it was witnessing something private. That's not a writing skill. That's a listening skill — knowing which question to ask to get to the moment that matters. That's what the skill is designed to do. The rest follows naturally.
The next piece most people tackle from here is vows that sound like you wrote them, not an AI. If you're working across the full Wedding workflow, the Wedding bundle covers everything in one place.