A Berlin Team’s Friday Demo Prep with AI Drafts

Set the Berlin team’s Friday demo target

It is Wednesday afternoon in Berlin. Your team has a product demo on Friday with a prospect whose working language is English, while their operations team is based in Japan. The goal is not to generate a polished presentation in one prompt. It is to prepare a demo that answers the prospect’s questions, uses language they can follow, and stays accurate when the conversation goes off-script.

Open a working document and write down three things: the audience’s roles, the decision they need to make, and the time available for the demo. For example: a product lead and two operations managers; decide whether a pilot fits their workflow; 25 minutes. Keep these details next to every AI draft so the model has useful context.

Gather the evidence before drafting

Collect the product facts you can stand behind: current capabilities, setup requirements, supported integrations, limitations, and one relevant customer outcome you are allowed to share. Add the prospect’s stated priorities from discovery notes. Mark uncertain claims as questions instead of letting a draft turn them into promises.

Make a short source sheet with headings such as “Verified,” “Not available,” and “Ask the client.” This gives you a practical way to check generated copy. Do not paste confidential customer data, credentials, or internal material into a prompt unless your organization has approved that use.

Draft the demo sequence in small pieces

Ask an AI model for a 25-minute outline based on your audience, decision, and verified facts. Request time boxes, a concrete screen or action for each section, and one question to ask the client. Then review the outline against the source sheet. Remove any unsupported capability, and shorten sections that do not help the prospect make their decision.

Draft the opening, transitions, and closing separately. A useful prompt is: “Write a 45-second opening for an English-language product demo with a Berlin product team and Japanese operations managers. Use plain international English, avoid idioms, and make no claims beyond these verified facts: [facts].” For a second pass, ask the model to flag ambiguous phrases and suggest simpler alternatives. You remain responsible for deciding which version sounds natural and accurate.

Prepare language for an international room

International English should be direct, not stripped of detail. Replace sports metaphors, regional slang, and jokes that depend on shared context. Expand unfamiliar abbreviations the first time they appear. Prefer “We will show how the approval step works” to “Let’s kick the tires on approvals.”

Ask the model to create a glossary of product terms with the wording you intend to use consistently. If you prepare translated notes for the client, treat them as a draft: check product names, numbers, dates, currencies, and technical terms with a fluent colleague or qualified reviewer. Keep the demo spoken in the language agreed with the client, and do not assume that translated slides alone make the session accessible.

Rehearse the uncertain parts

Run the outline aloud with a teammate acting as a skeptical attendee. Ask them to interrupt with realistic questions: What happens when an integration fails? Which steps require an administrator? What is not included in the pilot? Use AI to generate possible questions from the prospect’s priorities, then write your own answers from verified information. For anything you cannot confirm, prepare a clear follow-up rather than an improvised promise.

Check the timing with a real screen-share rehearsal. Leave space for questions, confirm that demo data is safe to show, and test the exact account and links you will use. Keep a backup screenshot or short recording for any environment-dependent step.

Check the final draft at the desk

Before Friday, read the script once for factual accuracy and once for spoken clarity. Confirm that names, time zones, dates, and next steps match the invitation. Share the agenda with the client in advance when appropriate, and ask whether the planned language and level of technical detail suit the attendees.

For teams that need flexible access to multiple AI models while preparing drafts, RelayVano offers a pay-as-you-go relay for models including Claude Sonnet 5.5 and Gemini 3.1 Pro. It works with Claude Code, Codex, Gemini CLI, and OpenAI-compatible SDKs using the API base https://api.relayvano.com. Sign up for RelayVano to try it in your drafting workflow. Keep the final judgment with your team: the best demo draft is the one you have checked, rehearsed, and can defend in the room.