The Best Claude Prompts for Writing Speeches

Claude distinguishes itself from other language models through its ability to maintain rhetorical coherence across long texts, calibrate stylistic register with precision, and reason about argumentative structure before generating text. These characteristics make it particularly suited to speech writing, a genre that simultaneously requires narrative logic, prosodic rhythm, and audience adaptation. In practice, I've verified that Claude excels at identifying the kairos — the appropriate rhetorical moment — when provided with sufficient contextual details about the audience and occasion.

Speech writing presents specific challenges that generic prompts don't adequately address: the text must function both as a written document and as oral performance, it must respect precise time constraints, and it must project an authentic voice that doesn't sound artificial. In this guide you'll find engineered prompts to solve exactly these problems, with explanations of underlying principles that will allow you to adapt them and build new ones independently.

Each prompt presented has been tested on Claude 3 Opus and Claude 3.5 Sonnet, with multiple iterations to optimize output consistency. You'll learn how to structure contextual briefings, specify oratorical parameters, manage emotional tone, and request targeted technical revisions.

Ready-to-Use Prompts for Writing Speeches with Claude

1. Opening Speech for Corporate Event

You are a ghostwriter specializing in corporate communication.

CONTEXT:
- Speaker: CEO of an Italian manufacturing company, 52 years old, direct and pragmatic style
- Event: annual company kick-off, 300 employees in attendance
- Duration: 8-10 minutes (approximately 1,200-1,500 words)
- Primary objective: motivate the team after a difficult year without minimizing challenges

TONE: authoritative but accessible, sincere, with a personal anecdote in the opening section

REQUIRED STRUCTURE:
1. Opening with personal anecdote (90 seconds)
2. Honest acknowledgment of 2024 challenges (2 minutes)
3. Pivot toward 2025 opportunities with concrete data [insert real data here] (3 minutes)
4. Collective call-to-action (2 minutes)
5. Memorable closing (30 seconds)

Include pause and emphasis cues in brackets. Avoid motivational clichés like "together we'll make it."

Why it works: This prompt applies the principle of the constrained persona — defining the speaker with specific demographic, professional, and stylistic attributes dramatically reduces output variance. The explicit request to avoid specific clichés activates Claude's negative steering mechanism, which in repeated tests has proven more effective than generic instructions like "avoid platitudes."

Expected output: A structured speech with markup for pauses, varied rhythm between short sentences and longer periods, and a consistent voice from opening to closing.

Customization: Replace the placeholder [insert real data here] with actual company metrics. The more specific the data, the more credible and non-generic the speech will feel.

2. Best Man Wedding Toast

I need to give a best man speech at my best friend Marco's wedding three weeks from now. Here's the relevant information:

ABOUT MARCO:
- Known him for 18 years, met at engineering university
- Methodical, passionate about mountaineering, has dry humor
- Anecdote to include: the 2015 Scotland vacation where he got lost in the mountains and met his wife Giulia by chance at a mountain hut

ABOUT GIULIA:
- Architect, witty, perfectly balances Marco's seriousness
- When I first met her, I immediately knew she was the right one

CONSTRAINTS:
- Duration: 4-5 minutes maximum
- Audience: 80 people, mixed generations (grandparents in their 80s + friends in their 30s)
- Tone: warm, with appropriate humor, not crude, emotional in the ending
- I need to avoid crying so the ending shouldn't be too melodramatic

Write the speech in first person, with a voice that sounds natural for someone who speaks Northern Italian.

Why it works: Personal narration requires episodic anchoring — providing specific, verifiable events that Claude can integrate into the speech structure. The detail "voice natural for someone who speaks Northern Italian" is a sociolinguistic specification that influences lexical choice and syntactic construction measurably.

Expected output: A speech with clear narrative arc: establishing the friendship → story of meeting Giulia → reflection on how love changed him → final toast. The comedic register will be calibrated to not alienate older generations.

Customization: Add a "phrases to avoid" section if you know expressions the speaker finds embarrassing or out of character.

3. Political/Institutional Speech on Specific Topic

You are an expert in political rhetoric and institutional communication.

Write a speech for the mayor of a municipality of 45,000 residents inaugurating a new public park built with European PNRR funds.

TECHNICAL PARAMETERS:
- Duration: 6 minutes (approximately 900 words)
- Primary audience: resident citizens, families with children
- Secondary audience: local journalists, regional officials present
- Register: formal but not bureaucratic

COMMUNICATION OBJECTIVES (in priority order):
1. Celebrate the result without seeming self-serving
2. Connect this project to a broader urban vision
3. Thank entities and collaborators without turning the speech into a list
4. Create a shared emotional moment for neighborhood residents

RHETORICAL CONSTRAINTS:
- Avoid unexplained acronyms (PNRR should be explained at first mention)
- Maximum 2 citations, relevant to the theme of community or public space
- Use the inclusive "we" technique to create a sense of collective belonging

Include an alternative opening proposal that the mayor can choose based on crowd mood.

Why it works: The stratification of audience (primary vs. secondary audience) is an advanced technique that instructs Claude to balance messages with different technical depth in the same text. The request for alternative opening introduces a branching element that increases the practical utility of the output.

Expected output: A speech with two opening options, systematic use of "we," organically integrated citations (not tacked on), and a structure that respects the hierarchy of communication objectives.

Customization: Provide the real name of the neighborhood, specific project data (hectares, investment, included services), and names of entities to thank to make the speech immediately usable.

4. Academic Speech for Master's Degree Ceremony

I'm concluding my Master's degree in Communication Sciences and need to give a brief speech on behalf of the graduating class during the ceremony.

CONTEXT:
- University: public, formal ceremony in the great hall
- Audience: families, professors, rector, other graduates
- Duration: 3-4 minutes
- Role of speech: represent the collective experience of the university journey, not just mine

INFORMATION I WANT TO INCLUDE:
- The pandemic years that marked part of our journey (isolation, remote learning)
- The resilience demonstrated by our generation
- Gratitude toward professors without becoming generic
- A reflection on professional future in a rapidly transforming sector (AI, digital media)
- A moment of light irony to ease the institutional tone

TONE: institutional but personal, optimistic without being naïve, with a cultural reference relevant to today's 25-year-olds

Write in first person plural where appropriate, first person singular in opening and closing.

Why it works: The specific instruction about grammatical person shifts (singular → plural → singular) demonstrates awareness of the rhetorical structure of representative speeches. Claude responds well to this type of technical instruction about narrative voice, producing smoother transitions than a prompt that doesn't make it explicit.

Expected output: A speech with personal opening that captures attention, body in collective voice that creates identification, and closing that restores individual agency to the speaker. The cultural reference will be contextualized to the specified generation.

Customization: Add 2-3 specific episodes from your real university experience — even just brief details — to anchor the speech to lived truth rather than generic reconstruction.

5. Persuasive Speech for Business Pitch or Presentation

I need to pitch a startup to a panel of investors. I have exactly 7 minutes.

COMPANY DATA:
- Sector: agritech, IoT sensor-based irrigation management software
- Traction: 12 active B2B clients, €180,000 ARR, 8% MoM growth
- Team: 3 founders, engineering and agricultural background
- Target market: farms 50-500 hectares in Southern Europe
- Round sought: €800,000 seed

INVESTOR PROFILE:
- Generalist VCs with some agritech in portfolio
- ROI-focused, skeptical of claims unsupported by data
- Already informed on European agritech market

PITCH STRUCTURE (to be respected):
1. Hook (30 seconds): real problem, made vivid
2. Solution (60 seconds): what we do, key differentiator
3. Market (60 seconds): TAM/SAM/SOM with source
4. Traction (90 seconds): metrics, proof of demand
5. Business model (45 seconds): how we generate revenue
6. Ask (30 seconds): amount, for what, expected milestones
7. Closing statement memorable (15 seconds)

CONSTRAINTS:
- No claims unsupported by data
- Avoid "revolutionary," "disruptive," "game-changer"
- Hook must include data on water inefficiency in agriculture
- Tone: confident, not arrogant; acknowledges competition without downplaying it

After the speech include a list of 5 difficult questions investors might ask, with suggested 2-3 sentence answers each.

Why it works: This prompt uses output chaining — by requesting both the speech and Q&A, it leverages Claude's ability to maintain coherence between main text and supplementary material. The explicit constraint on words to avoid is calibrated to terms Claude recognizes as signals of low credibility in investor pitch contexts.

Expected output: A speech with precise temporal structure, each section clearly delineated, followed by 5 question-answer pairs that anticipate the most probable objections. The data inserted in the prompt will be coherently integrated into the narrative flow.

Customization: Replace the metric placeholders with your actual data. If you have a particularly relevant founder story (e.g., background as a farmer), include it in the briefing — Claude will integrate it into the hook more effectively than any generic "I saw this problem with my own eyes" story.

6. Commemorative or Funeral Speech (Eulogy)

I need help writing the funeral eulogy for my grandfather, who passed away at 87.

INFORMATION ABOUT HIM:
- His name was Roberto, he was a master carpenter from Bergamo
- He lived through post-war reconstruction, built his workshop from nothing
- Values that defined him: manual work as dignity, family as absolute priority, quiet generosity (he helped without saying it)
- A phrase he repeated often: "Wood doesn't lie"
- He had a special relationship with me: taught me to use tools when I was 8 years old

CEREMONY CONTEXT:
- Catholic religious ceremony, parish church
- Audience: family members, old friends, neighbors — about 60 people
- Speaker: me, his 31-year-old grandson

EMOTIONAL CONSTRAINTS:
- The speech must honor him, not use him as an outlet for my grief
- I want those who knew him to recognize Roberto in every sentence
- A moment of gentle humor is appropriate (he was a man with subtle irony)
- Final: directed at those present as an invitation to remember him in his way of doing things, not in his absence

Duration: 5 minutes (750 words).
Tone: dignified, authentic, with human warmth. Avoid obituary language or excessive rhetorical flourishes.

Why it works: The constraint "the speech must honor him, not use him as an outlet for my grief" is an instruction that demonstrates emotional intelligence. Claude responds well to this type of guidance, producing text that channels emotion into form rather than raw expression. The detail "he was a man with subtle irony" gives Claude permission to use humor without trivializing the occasion.

Expected output: A eulogy structured around the values and repeated phrases that defined Roberto, with specific anecdotes (the workshop, the tools, the phrase "wood doesn't lie") woven throughout. The humor will surface organically from his character, not forced. The closing will reframe memory as an active practice — what we do in his way, rather than what we've lost.

Customization: Provide 3-4 specific anecdotes or habits that only someone close to him would know. These become the emotional anchors that transform the eulogy from formal tribute into genuine remembrance. The more specific, the more it will resonate with anyone who knew him.

Advanced Techniques for Engineering Effective Speech Prompts

Persona Constraint

Instead of asking Claude "write in a confident tone," specify: "Write as if you were a software engineer with 12 years of experience, comfortable with technical audiences but impatient with vague abstractions." The granularity of persona definition directly correlates with output quality.

Negative Specification

Don't just say what you want; specify what you explicitly don't want. Claude's negative steering is more precise than positive direction. "Avoid motivational clichés like X, Y, Z" is more effective than "use original language."

Temporal Scaffolding

Breaking the speech into minute-by-minute or second-by-second segments — as in the investor pitch example — forces Claude to respect pacing and prevents it from producing bloated sections that exceed time constraints.

Audience Stratification

When speaking to mixed audiences, explicitly name primary, secondary, and even tertiary audiences. Claude will automatically calibrate language depth and references accordingly.

Episodic Anchoring

Provide specific, true stories or details rather than generic character traits. "He was generous" produces generic output; "He helped my mother pay for medical treatment without telling anyone until years later" produces authentic emotional resonance.

Constraint Stacking

Combine multiple constraints — time, tone, audience, structure, specific phrases to avoid — for maximum precision. The more constraints, the more focused the output (up to a point before overconstraint becomes paralyzing).