AI-Powered Pitch Deck Workflow — Claude + Gamma + HubSpot

Overview

Mark Hope developed a repeatable workflow for producing personalized, data-driven pitch decks in under one hour — before ever speaking to a prospect. The process chains competitive research tools (Ahrefs, SpyFu) into an iterative Claude analysis session, then uses Claude's output as a direct script for Gamma to auto-generate slides. A PDF feedback loop between Gamma and Claude refines the deck before delivery.

This workflow was demonstrated live during a weekly call with Ben San Fratello in the context of a same-day discovery call with Ascend Analytics (a $30k–$78k/mo prospect).

"I got in here at 7:30. By 8:30 I had it done." — Mark Hope

The Workflow

Step 1 — Data Gathering

Pull raw competitive and SEO data for the prospect from two sources:

• Ahrefs — domain-level traffic, keyword rankings, trend data
• SpyFu — SEO overview, top organic competitors, shared keywords, most valuable keywords

Export or copy the reports; these become the raw input for Claude.

Step 2 — Iterative Analysis in Claude

This is the core of the workflow. The goal is to let Claude reason progressively rather than front-loading all data at once.

1. Open with context — Tell Claude who you are (Asymmetric), who the prospect is, and that you have a discovery call today.
2. Let Claude think first — Ask for an initial assessment of the prospect's website and market position before providing data. This surfaces Claude's priors and avoids anchoring its analysis.
3. Feed data incrementally — Paste in the Ahrefs data, then the SpyFu report. After each addition, read Claude's updated analysis and note what resonates.
4. Steer the strategy — Prompt Claude to address gaps it hasn't covered on its own. Examples from practice:
   - "What role should paid advertising play? We want to offer a comprehensive solution, not just SEO."
   - "Everyone I talk to thinks SEO is being killed by LLMs. How do we address that?"
5. Request the deck script — Once the strategy feels right, instruct Claude to write a 10-slide Gamma script. Key framing guidance to include in the prompt:
   - Don't sound presumptuous — highlight what research shows, not what you assume
   - Frame it as a draft plan built before speaking to them
   - Specify layout hints (e.g., three-column tables, header text, section structure)

Prompt pattern for the script request:

"Write me a 10-slide introductory pitch deck script for Gamma. Keep in mind we should not sound like we know all their business. We should highlight what we can see and what research tells us, and say that before even speaking to you, we've crafted a draft plan for how we might go about this."

Step 3 — Deck Creation in Gamma

1. Paste Claude's full slide script into Gamma.
2. Gamma auto-generates the presentation — no manual template work required. Claude's script includes layout instructions that Gamma follows directly.
3. Review the generated deck.

Step 4 — PDF Feedback Loop

1. Export the Gamma deck as a PDF.
2. Paste the PDF back into Claude.
3. Ask: "Here's what we got. What do you think?"
4. Claude reviews the visual output and suggests additions or changes it couldn't anticipate from the script alone.
5. Return to Gamma, apply the updates.
6. Repeat 1–2 times until the deck is solid.

Step 5 — Pre-Call Delivery

• Create two versions of the deck:
• Send in advance: Slides 1–10 (no pricing). Emailed to the prospect before the call so they can review and arrive with questions.
• Hold back: Pricing slides (typically 4 additional slides). Only shown if the prospect asks about cost during the call.
• Rationale: A 30-minute discovery call cannot cover 10 slides cold. Sending in advance maximizes the conversation.

Step 6 — Post-Call Follow-Up via Claude

After the call:

1. Upload the call recording to Claude.
2. Provide the deck used on the call.
3. Ask Claude to summarize what was covered, what was missed, and draft a follow-up email.
4. The email thanks the prospect, recaps key points, and surfaces anything that didn't get addressed on the call.

"It said, you did a great job, you covered this and that, but you forgot to talk about this. And then I said, what should I do for next steps? And it said, send them an email." — Mark Hope

HubSpot Integration

All prospect activity is tracked in HubSpot throughout this process:

• Contact record — Prospect info, page views (e.g., scheduling page visits), email activity
• Deal record — Created after a positive discovery call. Includes deal stage, probability, and deal value
• Attachments — The pitch deck PDF is attached directly to the deal record for future reference

Deal valuation convention: Deals are logged at one month's MRR, not annualized value. This prevents false confidence from inflated pipeline numbers.

"Otherwise I just convince myself I'm rich and I'm not." — Mark Hope

Deck Structure (10 Slides + Pricing Appendix)

Based on the Ascend Analytics example:

Key Principles

• Don't guide the witness — Let Claude form an initial view before feeding it data. The analysis is richer when it builds incrementally.
• You are the editor, not the author — Read Claude's output critically. Take what's useful, push back on what isn't, and steer toward what matters for this specific prospect.
• AI speeds you up; it doesn't replace judgment — The workflow only works because a human is making calls about what's relevant, what's presumptuous, and what the prospect actually needs to hear.
• Withhold pricing until asked — Sending pricing in advance risks cancellations before the relationship is established.

Tools Required

Related

• ^[1] — Asymmetric agency context
• ^[2] — Discovery call best practices
• ^[3] — Ben's transition to account management
• ^[4]