Warm
Earnest, family-toned. Best for when the room is sentimental and you want to honour the bromance without the jokes.
Best man speech generator
Claude Sonnet powered. Best-man specific. Five tones — warm, funny, heartfelt, formal, irreverent. Calibrated to 4-6 minute runtime. Avoids AI clichés. Produces the 5-act structure professional speechwriters use. Free.
How it works
01
Warm, funny, heartfelt, formal, irreverent. Pick one — the AI calibrates the whole draft to that single register.
02
One specific story (60-90 seconds when read aloud). One trait the story reveals. Your relationship to the groom. The bride's name. AI uses these specifics to write — not generic adjectives.
03
Hook (15-30s) → History (60-90s) → Heart (60-90s) → Humour (30-60s) → Hand-off (15-30s). Total runtime calibrated to 4-6 minutes. The structure professional speechwriters use.
04
AI gives you the scaffolding in 60 seconds. You spend an hour rewriting it in your voice — the best speeches are AI-drafted, human-rewritten. Read out loud. Time it. Test the jokes on three people. Print in 14pt.
Tones
Don't mix tones — the AI calibrates the whole draft to a single register. Match the wedding's vibe.
Earnest, family-toned. Best for when the room is sentimental and you want to honour the bromance without the jokes.
Self-deprecating + groom-roasting, but never mean. Calibrated to the line where laughs land without anyone wincing.
Maximum emotional weight. Best when you have a strong story about who the groom is, and you want to make the room cry without trying to.
Traditional structure, restrained language. Best for black-tie weddings, religious ceremonies, or family contexts where humour would feel inappropriate.
Sharper edges, modern phrasing, more confidence. Best when the bride/groom asked you specifically because you're the funny friend.
What makes it different
We use Anthropic's Claude Sonnet — the model that handles tone-shifting and structural calibration better than alternatives. Result: drafts that hold together end-to-end, not just sentence-by-sentence.
The prompt is calibrated for best man speeches specifically — not generic wedding speeches. Knows the difference between a best man's role and a maid of honour's role.
AI targets 4-6 minute runtime (the room loses you past 7). Reads at 130 wpm. The draft is the right length when you read it aloud — not when it's printed.
Hook → History → Heart → Humour → Hand-off. The exact structure professional speechwriters use across every wedding speech worth giving.
We explicitly block: "in conclusion", "as we celebrate", "on this special day", "love is a beautiful thing", "may your future be bright". The phrases that make AI-written speeches sound like AI-written speeches.
Warm, funny, heartfelt, formal, irreverent. Pick one — the whole draft commits to that register. Don't try to mix tones in the same draft.
Hate the hook but love the rest? Regenerate the hook only. Keep what works, replace what doesn't. Granular regeneration, not full restarts.
Your input is sent to Anthropic for the generation, then discarded. We don't store the speech text on our servers. Nothing trains future AI models on your draft.
Frequently asked
A best man speech generator is an AI tool that produces a structured first draft of a best man's wedding speech, given your input about the groom, your relationship, and a chosen tone. The best ones (ours, for instance) use Claude Sonnet, target the proven 5-act structure, calibrate to 4-6 minute runtime, and explicitly avoid AI clichés.
AI can write a structurally solid draft in 60 seconds. It cannot write the specific stories that make a speech land — that's still you. The best speeches are AI-drafted (for scaffolding) and human-rewritten (for voice + specifics). Use AI to escape blank-page paralysis, then make it sound like you.
Match the wedding. Black-tie religious wedding → formal. Casual outdoor wedding → warm or funny. Your best friend's wedding where you're chosen specifically because you're hilarious → irreverent. Don't pick "funny" if you're not naturally funny in person — the speech will fall flat. Earnest is always safe.
4-6 minutes. Over 7 minutes, you lose the room — even great speeches. The AI generator targets 4-6 minutes by default. Read your final draft aloud and time it; if you're at 7 minutes after editing, cut sections until you're at 6.
Yes — it's part of our free plan. Generate as many drafts as you need; regenerate individual sections as many times as you need. The free tier also includes the website builder, RSVP, seating chart, budget tracker and 30+ other planning tools.
(1) Hook — open with a specific moment, not thanks. (2) History — one story (60-90s) that reveals a quality of the groom. (3) Heart — pivot to earnest, name the quality, tie it to his partner. (4) Humour — one callback, one self-deprecating beat. (5) Hand-off — raise the glass, sit down. Total 4-6 minutes. We have a full guide on this structure if you want to write your speech from scratch.
Only what you tell it. You provide the groom's name, the bride's name, one story about the groom (the more specific the better), one trait the story reveals, and your relationship to him. The AI uses those specifics. Generic input → generic speech. Specific input → specific speech.
Less than you'd think. Two things prevent this: (1) our prompt explicitly blocks the AI clichés that make speeches sound generic ("in conclusion", "on this special day", "love is a beautiful thing"), and (2) the draft asks for your specific stories, which means the unique content comes from you, not from the model. That said: always edit. The speeches that sound the most generic are the ones the speaker didn't rewrite.
Yes — the AI speech writer ships with all 5 role types: best man, maid of honour, parent of the bride, parent of the groom, the couple themselves. Each role has its own calibrated prompt because the structure differs (parent speeches have different opening conventions, couple speeches have different runtime targets).
AI generates in any language Claude Sonnet supports — natively French, English, Spanish, Italian, Arabic, and many more. Pick your language at the start of the flow; the entire draft generates in that language. Tone calibration works across languages.