Day 2: Deep competitor analysis

Analyze the following competitor of my startup:

Competitor: [NAME]
URL: [WEBSITE]
My product: [NAME AND DESCRIPTION]

Provide:
1. Main value proposition (how they position themselves)
2. Detailed pricing model (plans, amounts, what's included in each tier)
3. Negative reviews -- what customers complain about (use data from G2, Capterra, Trustpilot, App Store)
4. Gaps my product can exploit
5. Strengths I need to respect or overcome
6. How they acquire customers (channels, content strategy, SEO, outbound)
7. Size estimate (employees, revenue, funding rounds if applicable)

Be specific. I don't want generic analysis.

Day 3: Positioning

Based on the market mapping below, help me position [PRODUCT].

MARKET:
[PASTE YOUR MAPPING FROM DAY 1]

MAIN COMPETITORS:
[PASTE YOUR ANALYSIS FROM DAY 2]

MY DIFFERENTIATOR (from my perspective): [DESCRIBE]

Generate:
1. 5 positioning options (each with a different angle)
2. For each option: tagline, 1-paragraph value proposition, why it works
3. Recommendation of the strongest option with justification based on market data
4. Opening message for cold email or pitch (2-3 sentences)

Automated competitor analysis

To continuously monitor competitors without hiring an analyst:

Recommended stack for early-stage: Perplexity Pro + Similarweb free. Cost: R$110/month. Sufficient for weekly monitoring of top 5 competitors.

Part 2: Generating pitch decks with AI

The structure that works with Brazilian investors

Based on the criteria of funds like Kaszek, Monashees, Canary, ACE, and Redpoint eventures, the pitch deck for Brazil in 2026 follows this logic:

Prompt to generate content for each slide:

You are a pitch deck specialist for the Brazilian venture capital ecosystem.

My startup:
- Name: [NAME]
- Segment: [B2B / B2C / B2B2C]
- Sector: [SECTOR]
- Stage: [IDEA / PRE-SEED / SEED]
- Solution: [DESCRIPTION]
- Problem solved: [DESCRIPTION]
- Current metrics: [LIST WHAT YOU HAVE]
- Current revenue: [NUMBER OR "no revenue yet"]
- Team: [NAMES AND BACKGROUNDS]

Generate content for the [SLIDE NAME] slide.

Rules:
- Maximum 50 words of text on the slide (the rest is in your speech)
- 1 impactful number that anchors the slide
- 1 proof point or data that validates the claim
- Tone: direct, ambitious but honest
- Don't overuse jargon -- Brazilian investors prefer clarity

The traction slide: the most important one in your pitch

For Brazilian funds in 2026, traction is the number one filter. Here's how to present metrics with AI:

Analyze my metrics below and help me present my startup's traction in the most compelling way for a Brazilian seed investor.

METRICS:
[PASTE YOUR METRICS: MRR, users, MoM growth, churn, CAC, LTV, any data you have]

GENERATE:
1. The most impressive metric I have (with context on why it impresses)
2. The 3-act growth narrative (where I came from, where I am, where I'm going)
3. The number that anchors the slide (the most impactful one)
4. 2 market benchmarks that make my metrics look good in context
5. What NOT to show (metrics that weaken the pitch)
6. 1 sentence that summarizes traction in 10 words

Part 3: AI-Assisted MVP Creation

The Minimum MVP Framework for Brazilian Startups

The classic early-stage Brazilian founder mistake: building too much before validating. AI enables a much faster validation cycle.

Level 1: Paper MVP (R$ 0)

Use AI to create a visual representation of your product before writing a single line of code.

I need to create a paper MVP to validate [PRODUCT].

My target user: [PROFILE]
The problem it solves: [DESCRIPTION]
The main action the user takes in the product: [DESCRIPTION]

Generate:
1. User journey map (5-7 steps) from the moment they recognize the problem until they solve it with my product
2. List of the 3 most critical screens to validate the core hypothesis
3. For each screen: which UI elements are essential (form, button, list, chart) and which are distractions
4. Core hypothesis that the MVP needs to test (format: "We believe [user] will [action] because [motivation]. We'll know we're right when [metric]")
5. Success criteria: what needs to happen for the MVP to be considered validated

Level 2: Landing Page MVP (R$ 0-200)

Create the copy for a validation landing page for [PRODUCT].

Page objective: Collect emails from people interested in [PRODUCT] before building it.

My target user: [PROFILE]
Main problem: [DESCRIPTION]
Product promise: [WHAT YOU DO]

Generate:
1. Main headline (10-12 words maximum, benefit-focused, not product-focused)
2. Subheadline (1 sentence that expands on the promise)
3. 3 benefit bullets (format: "You [action verb] [specific result] without [frustration eliminated]")
4. Social proof placeholder (what to collect to add later)
5. CTA (button text + email capture form)
6. "How it works" section in 3 simple steps
7. Segmentation question for the form (1 question that gives you valuable data about the lead)

Tone: direct, no hype, honest about being early-stage. Brazilians prefer authenticity.

Level 3: AI-Assisted Functional MVP

For startups with some technical resources, use AI to accelerate development:

MVP stack for founders without a CTO: Bolt.new + Supabase + Vercel. Cost: R$ 110/month. You can have a working product in one week.

Part 4: Automated Customer Discovery

Customer discovery is the most important and most overlooked activity at the early stage. Most founders conduct 5-10 interviews when they should be doing 50-100.

AI doesn't replace talking to customers. But it helps you prepare better, analyze faster, and surface insights you wouldn't catch manually.

Preparing for Interviews

Prepare a customer discovery script for [PRODUCT].

Hypothesis I want to validate: [DESCRIPTION]
Interviewee profile: [ROLE, COMPANY, CONTEXT]
Available time: [30 OR 45 MINUTES]

Generate:
1. Opening script (30 seconds that explain the goal without influencing responses)
2. 5 main questions (open-ended, revealing behavior rather than opinions)
3. 3 follow-up questions for each possible response
4. Questions I should NEVER ask (those that lead the interviewee to confirm my hypothesis)
5. How to close the interview by asking for a referral (Mom Test technique)
6. What to observe in the interviewee's tone and language (signs of genuine interest vs politeness)

Principle: Never mention the product. Discover actual behavior, don't validate the solution.

Scaling Interview Analysis

Analyze the customer discovery interview transcriptions below and extract structured insights.

TRANSCRIPTIONS:
[PASTE INTERVIEW TRANSCRIPTIONS OR SUMMARIES]

ORIGINAL HYPOTHESIS: [DESCRIPTION]

Generate:
1. Most cited problems (ranked by frequency and intensity)
2. Exact language interviewees used to describe the problem (quote verbatim)
3. Current behavior to solve the problem (what they're using today)
4. What they said does NOT work in the current solution
5. Willingness to pay: what indications they give about how much they'd pay
6. Emerging segments (groups with different problems or contexts)
7. Validated hypotheses, invalidated hypotheses, and new hypotheses generated
8. Next 5 recommended interviews (different profiles to reduce confirmation bias)

Format: insight table with frequency (X of Y interviewees mentioned).

Automated Customer Discovery at Scale

For startups that need quick quantitative data before qualitative interviews:

Prompt to create a market survey:

Create a 10-question survey to validate [HYPOTHESIS] with [USER PROFILE].

Rules:
- Closed-ended questions with scale (1-5 or multiple choice) for easy analysis
- 2 strategic open-ended questions (the most important ones)
- First question: screening (eliminate respondents who aren't in the target audience)
- Last question: email (optional) for interview follow-up
- Avoid questions that induce positive responses
- Tone: short, direct, maximum 3 minutes to complete

For each question, explain what you expect to discover.

Distribution Tools:

Part 5: AI-Assisted OKRs and Strategy

Defining OKRs That Work for Early-Stage

The most common mistake: copying OKRs from a large company that make no sense for a startup.

I need to define OKRs for my startup for the next quarter.

COMPANY CONTEXT:
- Stage: [IDEA / PRE-PRODUCT-MARKET-FIT / POST-PMF]
- Current biggest risk: [PRIMARY RISK -- e.g.: "I still don't know if people will pay"]
- Most scarce resource: [TIME / MONEY / TEAM]
- What we learned last quarter: [DESCRIPTION]
- 12-month goal: [DESCRIPTION]

GENERATE:
1. 1 main Objective for the quarter (what we need to prove or achieve)
2. 3-4 Key Results for that objective (measurable, with number and date)
3. For each KR: why it matters, how we'll measure it, what could block it
4. Main initiatives (the 2-3 bets that move the KRs the most)
5. What NOT to do this quarter (conscious backlog)
6. Weekly check-in: 3 questions to review progress

Principle: For pre-PMF, the objective should be to learn, not to grow.

AI-Assisted Go-to-Market Strategy

Create a go-to-market strategy for [PRODUCT] in the Brazilian market.

PRODUCT: [DESCRIPTION]
PRIMARY SEGMENT: [PROFILE]
PRICE: [VALUE OR MODEL]
AVAILABLE RESOURCES: [TEAM + MONTHLY BUDGET]
6-MONTH OBJECTIVE: [SPECIFIC GOAL -- e.g.: "100 paying customers"]

GENERATE:
1. Primary acquisition channel (what will generate 80% of first customers)
2. Positioning message for that channel
3. Outbound playbook (if B2B): contact sequence, tools, templates
4. Inbound playbook (if B2C): content channel, frequency, topics
5. Partners or communities for initial activation (specific to Brazil)
6. Channel metrics: what to measure week 1, month 1, month 3
7. Channel pivot criteria: when to abandon the main channel and test another

Be specific to the Brazilian context (WhatsApp, Instagram, LinkedIn, SME market).

Part 6: Tool stack for early-stage startups (under R$5,000/month)

Tier 1: Essential stack (R$500-1,000/month)

Tier 2: Growth stack (R$1,000-2,500/month)

Tier 3: Pre-Series A full stack (R$2,500-5,000/month)

Part 7: Brazilian startup case studies using AI

Nubank: AI as a competitive advantage from day one

Nubank didn't use AI exactly as we've described, but the principle is the same one they adopted back in 2013: eliminate what banks do manually and replace it with intelligent automation.

Today, Nubank processes over 100 million credit analyses per month with proprietary models. Their approval rate is 40% higher than the banking average with 30% lower default rates. This didn't happen because Nubank is bigger—it happened because they took data seriously from day one.

What founders can learn:

• Treat data as a product from the MVP stage. Structure data collection before you need it.
• Credit AI, pricing, and risk assessment are defensible differentiators for any fintech.
• The data model is more valuable than the initial product.

iFood: Testing and iteration with AI in product

iFood uses AI in at least 5 critical dimensions: demand forecasting (for rider planning), dynamic delivery pricing, restaurant ranking, dish recommendations, and fraud detection.

The "combo recommendation" feature—when the app suggests adding fries to your burger—increased average order value by 12% without any change to the physical product.

What founders can learn:

• Behavior-based personalization is a low-cost revenue lever.
• Use AI for ranking and recommendations before any other application in marketplaces.

QuintoAndar: AI to eliminate friction in traditional markets

QuintoAndar used AI to solve Brazil's biggest rental problem: bureaucracy. Automated credit analysis (results in minutes vs. weeks), intelligent property pricing, and detection of underpriced properties for market capture.

What founders can learn:

• In regulated and bureaucratic markets (real estate, legal, financial, healthcare), AI to eliminate process friction is the most powerful entry point.

Startups from ACE Ventures and Brazilian Y Combinator alumni

Part 8: The pitfalls -- when AI becomes a distraction in early-stage

Pitfall 1: "I'll use AI to build first, then validate"

The problem: You spend 3 months building an AI-assisted product without talking to a single real customer. The product is technically impressive and nobody wants it.

The rule: Before using any AI tool to build, you need to have at least 20 conversations with potential customers. AI accelerates execution. It doesn't replace validation.

Pitfall 2: Spending on AI stack before product-market fit

The problem: Founder signs up for 15 AI tools (BRL 3,000/month) before having a single paying customer. High burn rate, low learning rate.

Pre-PMF rule: Use only the 3-4 essential tools. ChatGPT Plus + Cursor + Make + free HubSpot. Under BRL 500/month. The money left over goes toward customer interviews (coffee, transportation, founder time).

Pitfall 3: Using AI to create strategy instead of learning from the market

The problem: You use AI to create a perfect go-to-market plan on paper. The plan makes sense. The market disagrees.

The rule: AI generates hypotheses. The market validates them. Never execute an AI-generated strategy without validating the assumptions with real data.

Pitfall 4: Automating what you don't yet understand

The problem: You automate your sales process before you've made 50 manual sales. Your automated funnel repeats the error at scale.

Paul Graham's rule applied to AI: "Do things that don't scale" before automating. Understand the process manually before letting AI take over.

Pitfall 5: AI as a substitute for leadership

The problem: You use AI to make hard decisions (firing someone, changing product, pivoting) instead of developing judgment as a founder.

The rule: Use AI to collect and organize information. The final decision is yours. Founders who delegate strategic decisions to AI lose their capacity for judgment development -- the most valuable resource a founder has.

30-day plan to implement AI in your startup

Week 1: Foundation (days 1-7)

Days 1-2: Audit where your time goes

Before implementing any tool, do an honest mapping of where you spend your time as a founder.

List all activities you did in the last 7 days and classify them:
- [HIGH VALUE] Activities only I can do (customer conversations, product decisions, fundraising)
- [MEDIUM VALUE] Activities that require judgment but are partially delegable (analysis, documentation, research)
- [LOW VALUE] Repetitive and bureaucratic activities (formatting, scheduling, basic research)

Days 3-4: Set up your minimum stack

• Create ChatGPT Plus account (or Claude Pro)
• Create Make account (free plan)
• Set up HubSpot CRM (free)
• Set up Notion for documentation

Days 5-7: First prompts and automations

Use the prompts from this guide to:

• Market research (use the prompt from Part 1)
• Customer discovery script (use the prompt from Part 4)
• Define quarterly OKRs (use the prompt from Part 5)

Week 2: Accelerated customer discovery (days 8-14)

Goal: Conduct 10 customer interviews and analyze them with AI.

Day 8: Create a list of 30 potential interviewees (LinkedIn, Facebook groups, personal network)

Days 9-10: Send interview requests to all 30 (expected response rate: 20-30%)

Days 11-13: Conduct the interviews. Record with permission. Transcribe with Whisper or Otter.ai.

Day 14: Use the interview analysis prompt to extract insights. Identify the 3 most critical insights.

Week 3: Product and validation (days 15-21)

Days 15-16: With insights from interviews, use AI to refine your product hypothesis.

Days 17-18: Create an MVP landing page using the prompt from Part 3.

Days 19-21: Launch the landing page and collect your first 100 visits. Goal: 15-20% email conversion rate.

Week 4: Pitch and strategy (days 22-30)

Days 22-24: With validation data, generate pitch deck content using the prompts from Part 2.

Days 25-27: Define a 90-day go-to-market strategy using the prompt from Part 5.

Days 28-30: Review the plan, identify the biggest unvalidated risk, and plan the next 30 days.

Success metrics for the 30-day plan:

What comes next: AI for growth-stage startups

Once you hit Product-Market Fit, how you use AI shifts completely.

Pre-PMF: AI to learn faster. Post-PMF: AI to scale cheaper.

At the growth stage, the most impactful applications are:

• AI in sales: outbound automation, lead scoring, sales rep coaching
• AI in marketing: content personalization, ad optimization, SEO at scale
• AI in product: user behavior analysis, churn prediction, feature prioritization
• AI in support: chatbots, intelligent escalation, automated onboarding
• AI in operations: internal process automation, financial forecasting, supplier management

Each area has its own playbook. But it all builds on what you learned in the early stage: validate before building, measure what matters, and use AI to accelerate learning—not to avoid it.

Next steps

Ready to structure AI adoption in your startup with hands-on guidance and courses that take you from fundamentals to advanced application?

• Explore AI courses for business — modules tailored for founders, marketing, and product teams
• Browse strategy and product prompts — ready-to-use templates for the most common startup challenges
• Create your free account and start applying AI to your startup today

Founders who use AI don't work more hours. They learn more in fewer hours—and that's why they win.