How to Set Up Your Pitch Context Before You Start Speaking
Two founders use the same voice-to-slides tool on the same pitch. One spends 3 minutes on context setup before starting. The other hits the mic immediately.
The first founder's deck comes out with real company names, accurate team members, and the right metrics on the right slides. The second founder's deck has placeholder text, generic labels, and a team slide with "Co-Founder" instead of anyone's actual name.
The context layer is the difference.
What the Context Layer Is
Before you start speaking in a voice-to-slides session, good tools give you a set of input fields to fill in. This is the context layer -- structured information about your company and your pitch that the AI uses to generate more accurate, more specific slides.
Think of it as briefing your AI before the session. Without the brief, the AI has to guess. With the brief, it works from facts.
The context layer is separate from the speech session. You fill it in once per company (and update it when things change), not once per session.
The Fields That Matter Most
Company Name
This is the simplest field and the one with the most obvious impact. When the AI knows your company is called "Talkpitch," every slide that references your company uses that name instead of a placeholder.
Without this: "We started [Company Name] in 2024." With this: "We started Talkpitch in 2024."
The STT engine doesn't always catch proper nouns accurately on the first pass. Setting the company name in context means the AI uses the correct spelling even if the transcription was slightly off.
Pitch Description
A short description of what you're pitching -- typically 1-3 sentences. This shapes the overall framing of your deck.
Write it as you'd write the opening paragraph of your executive summary. Specific and factual: "Talkpitch is a browser-based SaaS tool that generates presentation slides from your spoken pitch in real time. Founders speak through their pitch once and get a draft deck in under 30 minutes."
This description helps the AI understand the thesis of the pitch, which improves its categorization decisions throughout the session.
Team Members
Enter each person with their name, role, and one key credential. When you speak a team section in your session ("Our team is..."), the AI uses these records to populate the team slide with the right names and titles.
Without this: "Team: [Name] - [Role]" (placeholder, filled from garbled transcript) With this: "Alice Smith - Lead Engineer (ex-Stripe)"
Format matters. "Name, Role, Key Credential" per entry is the cleanest structure.
Key Metrics
Enter the numbers you plan to reference in your pitch: ARR, user count, growth rate, retention, market size, whatever is relevant. When you say "we're at $50K ARR growing 30% month-over-month" in the session, the AI has already verified this against your context and can format it accurately on a metrics slide.
The most common transcription error in voice-to-slides sessions is numeric data. "30%" becomes "3%" becomes a wrong metric on a traction slide. Context-layer metrics serve as a reference the AI can cross-check against.
Context Setup in Practice: A 3-Minute Checklist
Before every session:
- Company name: Written exactly as it appears in your legal name or brand (capitalization matters for some tools)
- Pitch description: 2-3 sentences. What you're building, for whom, and what makes it worth funding.
- Team: Each co-founder and key hire. Name, role, one previous credential.
- Key numbers: Your top 3-5 metrics. ARR, users, growth rate, market size -- whatever you'll say out loud.
- Pitch purpose (if available): Is this an investor pitch? A product demo? A sales presentation? Setting the purpose helps the AI select the right default slide structure.
Total time: 3 minutes for a new context. 30 seconds for updating an existing context when your metrics change.
When to Update Your Context
Your context isn't static. Update it when:
- Your metrics change meaningfully (monthly is a reasonable cadence)
- You add a key team member
- Your pitch description changes (pivots, new positioning)
- You're running a session for a different purpose (investor pitch vs sales demo)
Some founders keep multiple context profiles: one for the investor pitch, one for the sales demo, one for the partnership presentation. Each has the same company fundamentals but different metrics emphasis and pitch description.
What Context Can't Do
Context improves output quality significantly. It doesn't make the session foolproof.
The AI still gets things wrong sometimes. Transcription errors happen even with context (especially for unusual proper nouns). Layout choices are occasionally off. Some slides need editing regardless.
Think of context as the quality floor for your session. Without it, the floor is low. With it, the floor is high enough that editing takes 15 minutes instead of 45.
Context also doesn't replace a good speaking technique. Clear segment structure, deliberate pausing, and organized delivery are still what determine whether the session produces clean output. Context is a multiplier on good technique, not a substitute for it.
For a full walkthrough of how to structure a session once your context is set, see how to create slides by speaking out loud.
And for practical tips on speaking clearly enough to get professional output, how to speak naturally and still get professional slides covers the delivery techniques that work.
Start a session on Talkpitch -- set your context, hit the mic, and see what three minutes of setup does to your output quality.