Brain Cells: Accelerating Startups
Dylan Goldblatt
August 15, 2025
KSU Office of Research
Research at Startup Speed
From Manual Analysis to AI-Powered Insights
Why Startups Choose Brain Cells
Speed: Turn weeks of research into hours of automation
Privacy: Keep your competitive intelligence secure
Scale: Process thousands of data points in parallel
Flexibility: Adapt to any research need without coding
Protect Your Competitive Advantage
Your Research Stays Confidential
Startup-Safe Security
- Investor pitch data stays private
- Customer lists remain confidential
- Competitive analysis never leaves your machine
- Product roadmaps stay secret
- Pricing strategies remain internal
Start Researching Today
From Idea to Insights in 5 Minutes
The Fastest Path to Research
git clone https://github.com/ngoldbla/braincells.git
cd braincells
./start.sh
Power at Your Fingertips
What Can Brain Cells Actually Do?
Market Research & Competitive Analysis
"Analyze {{competitor_website}} and extract their value proposition"
"Generate TAM/SAM/SOM calculations from {{industry_report}}"
"Create investor pitch points based on {{market_gap}}"
"Identify underserved segments in {{customer_feedback}}"
Investor & Funding Intelligence
"Extract investment thesis from {{vc_portfolio_page}}"
"Identify check size and stage focus of {{investor_name}}"
"Find warm intro paths to {{target_investor}}"
"Parse {{investor_interview}} for funding criteria"
Customer Discovery & Problem Validation
"Categorize {{user_interview}} by pain point severity"
"Score {{survey_response}} for product-market fit signals"
"Extract jobs-to-be-done from {{customer_feedback}}"
"Identify early adopter characteristics in {{user_data}}"
Lead Generation & Enrichment
"Find decision maker titles at {{target_company}}"
"Enrich {{company_name}} with employee count and funding data"
"Identify tech stack from {{company_website}}"
"Generate personalized outreach for {{prospect_profile}}"
Built for Startups
Technology That Matches Your Pace
Startup-Optimized Stack
Hugging Face Integration
Access to thousands of models without individual APIs
One token unlocks everything
Community-validated models for every use case
Local-First Architecture
No cloud costs eating your runway
No usage limits blocking your research
No latency slowing your iterations
Docker Simplicity
Works on any laptop your team has
No IT department required
Deploy anywhere as you scale
Models for Every Research Need
Match your model to your research phase:
Customer Discovery
Phi - Process hundreds of interviews quickly
Llama 3.2 - Extract nuanced insights from conversations
Competitive Analysis
GPT-OSS - Deep reasoning about market positioning
Mistral - Pattern recognition across competitors
Market Research
Qwen - Process large industry reports
Llama 70B - Complex multi-factor analysis
Problem Discovery Research
"Extract specific problems from {{interview_transcript}}"
"Rate problem severity on scale of 1-10 based on {{description}}"
"Identify willingness to pay signals in {{customer_feedback}}"
Competitive Intelligence Gathering
"Analyze {{competitor}} pricing model and identify gaps"
"Extract features from {{product_page}} and categorize by importance"
"Compare our positioning vs {{competitor_messaging}}"
Market Expansion Research
"Identify new use cases in {{customer_success_story}}"
"Score {{industry_vertical}} for expansion potential"
"Generate ICP variations based on {{existing_customer_profile}}"
AI Startups: Training Data Generation
"Generate examples of {{use_case}} for model training"
"Create edge cases for {{product_feature}} testing"
"Synthesize customer queries in {{industry_domain}}"
"Generate labeled data for {{classification_task}}"
Getting Started
Your First Brain Cells Workflow
Step 1: Choose Your Starting Point
Generate from scratch: "Cities of the world with landmarks in Ghibli style"
Import your data: Upload XLS, TSV, CSV, or Parquet (up to 1,000 rows)
Import from Hub: Load any Hugging Face dataset directly
Step 2: Define Your Research Questions
Click "+" to add AI-powered research columns
"What problem does {{customer}} describe in {{interview}}?"
"Rate {{feedback}} on a product-market fit scale of 1-10"
"Extract the main value prop from {{competitor_website}}"
Step 3: Configure and Run
Select your AI model and provider for different performance
Toggle "Search the web" for up-to-date information
Click "Run" to generate results for all rows
Step 4: Iterate and Refine with Feedback
Edit cells: Click any cell to provide examples of preferred output
Like results: Use thumbs-up to mark good examples
Regenerate: Apply your feedback to improve all cells
Drag down: Generate more rows instantly from the last cell
Advanced Capabilities
Pushing the Boundaries
Multi-Stage Research Pipelines
Column 1: "Extract company name and description from {{website}}"
Column 2: "Identify industry and business model of {{company}}"
Column 3: "Score {{company}} as potential customer (1-10) based on ICP"
Column 4: "Generate personalized outreach for score >7"
Dynamic Research Strategies
"If {{company_size}} > 1000, research enterprise pain points
If {{funding_stage}} = 'Series B+', analyze scaling challenges
If {{industry}} = 'FinTech', check regulatory compliance needs"
Research Model Selection Strategy
Fast Iteration: Llama 3.2 for quick hypothesis testing
Deep Analysis: GPT-OSS for complex reasoning tasks
Bulk Processing: Phi for processing thousands of rows
Specialized Research: Domain models for industry-specific analysis
From MVP to Scale
Growing Your Research Capabilities
Research Evolution by Stage
Pre-Seed Research (1 Founder)
Local instance for customer discovery
Validate problem-solution fit
Identify beachhead market
Seed Research (Small Team)
Shared instance for competitive analysis
Market sizing and TAM validation
Pricing and positioning research
Growth Stage (Research Team)
Multiple instances for parallel research
Continuous competitive monitoring
Expansion opportunity analysis
Protecting Your Startup's Edge
- Pre-launch strategies stay confidential
- Customer lists never leave your control
- Pricing research remains internal
- Investor communications stay private
- Product roadmaps are never exposed
- Competitive analysis is fully secured
The Startup Advantage
Open Source for Maximum Flexibility
Why Choose Open Source
- No burn rate increase: Free to use, scale when ready
- Pivot friendly: Modify anything to match your needs
- Due diligence ready: Full code transparency for investors
- Integration flexible: Connect to your existing stack
- Community powered: Learn from other startups' workflows
Scale Your Research Operations
Template successful research: Export validated research workflows
Automate recurring analysis: Set up weekly competitive monitoring
Scale validated approaches: Apply proven research to new markets
Build research moat: Accumulate proprietary insights over time
Research Best Practices
Maximizing Your Intelligence ROI
Start Small, Validate, Then Scale
Test with 5-10 rows first
Research Validation Checklist
✓ Do the insights lead to specific actions?
✓ Can you make decisions based on this data?
✓ Would you pay for these insights?
✓ Do patterns emerge from the sample?
Research Velocity Optimization
Choose the right model for each research phase
Discovery Phase (Speed > Precision)
Use Phi or Llama 3.2 for rapid iteration
Test 10+ research questions quickly
Identify promising research directions
Validation Phase (Balance)
Use Mistral for deeper analysis
Process full sample sets
Refine research criteria
Production Phase (Precision > Speed)
Use GPT-OSS for final insights if you have more VRAM
Process complete datasets
Generate investor-ready analysis
The Research Roadmap
Features Designed for Startup Success
Near Term: Research Acceleration
- Competitor monitoring: Automated weekly intelligence reports
- Customer discovery templates: Pre-built interview analysis
- Investor research workflows: DD-ready data rooms
- Market sizing calculators: TAM/SAM/SOM with evidence
- Pitch deck generator: From research to slides
Medium Term: Intelligence Platform
- Research agents: Autonomous overnight analysis
- Trend detection: Alert on market shifts
- Integration hub: Connect to CRMs, analytics, data sources
- Research marketplace: Buy/sell validated research templates
- Collaboration tools: Share research with investors/advisors
Long Term: Autonomous Research Operations
The vision: Research that runs itself
- Wake up to completed competitive analysis
- Automatic customer insight extraction from all touchpoints
- Predictive alerts on market opportunities
- AI that learns your specific market dynamics
- Research moat that compounds with every query
The Future of Research for Startups
Every founder becomes a research expert
Every hypothesis gets validated with data
Every pivot is informed by intelligence
Every startup has enterprise-grade research
Brain Cells: Accelerating Startups
Dylan Goldblatt
August 15, 2025
KSU Office of Research