What is ActiveGenie?

Building with Generative AI can be chaotic. You get inconsistent outputs, unpredictable behavior, and a constant, time-consuming need to re-engineer prompts for different models. This makes it incredibly difficult to build features that are reliable enough for production.

ActiveGenie is your toolkit for taming that chaos. It's an enabler for creating reliable, production-ready GenAI features by offering powerful, model-agnostic tools that deliver consistent results from supported providers.

Instead of wrestling with the low-level complexities of LLMs, you can solve real-world problems with a simple, consistent API.

TIP

Convinced already? Jump straight to the Quickstart Guide to get started in minutes.

---

See It In Action

1. ActiveGenie::Comparator

Comparator conducts a structured "political debate" between two players to determine a clear winner based on your criteria, providing detailed reasoning for its choice.

player_a = "Implementation uses dependency injection for better testability"
player_b = "Code has high test coverage but tightly coupled components"
criteria = "Evaluate code quality and maintainability"

result = ActiveGenie::Comparator.call(player_a, player_b, criteria)
# => ComparatorResponse(
#      winner: "Implementation uses dependency injection for better testability",
#      reasoning: "Player A's implementation demonstrates better maintainability through dependency injection..."
#    )

Perfect for:

• Better copy: Choosing between marketing copy variants.
• Best product: Selecting the best product by abstract criteria.
• Balance Gamification: Compare player strategies to give more points to better ones.

2. ActiveGenie::Scorer

Scorer assembles an AI jury of specialists (e.g., a cardiologist, a senior developer, a marketing expert) to provide an objective quality assessment with detailed feedback from multiple perspectives.

content = "Patient shows significant improvement in cardiac function with ejection fraction increased from 45% to 62%"
criteria = "Evaluate medical accuracy, clarity, and clinical relevance"

result = ActiveGenie::Scorer.call(content, criteria)
# => {
#      "cardiologist_score" => 94,
#      "cardiologist_reasoning" => "Clinically significant improvement in ejection fraction...",
#      "medical_writer_score" => 87,
#      "final_score" => 90.7
#    }

Perfect for:

• Content quality: Evaluating articles, documentation, or marketing copy.
• Compliance: Scoring content against regulatory or brand standards.
• Assessment: Evaluating essays, reports, or technical answers.

3. ActiveGenie::Ranker

Ranker organizes a multi-stage tournament, combining initial scoring + ELO ratings + free for all debates to produce a reliably sorted list.

solutions = [
  "Uses modern design patterns with proper separation of concerns",
  "Implementation uses dependency injection for better testability",
  "Legacy code with tightly coupled components but working functionality"
]
criteria = "Evaluate code quality and software engineering best practices"

result = ActiveGenie::Ranker.call(solutions, criteria)
# => {
#      players: [
#        { content: "Uses modern design patterns...", score: 85, elo: 1245, rank: 1 },
#        { content: "Implementation uses dependency injection...", score: 82, elo: 1198, rank: 2 },
#        # ... more ranked results
#      ]
#    }

Perfect for:

• Candidate evaluation: Ranking job applicants based on qualifications.
• Product comparison: Ranking features, vendors, or solutions.
• Content curation: Prioritizing articles, posts, or media.

4. ActiveGenie::Extractor

Turn messy text into clean, structured data. Extractor uses your defined schema to parse unstructured text, even handling informal language and complex context with ease.

product_text = "Sony 65\" Class BRAVIA XR X95K 4K HDR Mini LED TV - $1999.99 (Save $500)"
schema = {
  brand: { type: 'string', description: 'Product brand' },
  display_size: { type: 'string', description: 'Screen size with units' },
  price: { type: 'number', minimum: 0, description: 'Current price' }
}

result = ActiveGenie::Extractor.call(product_text, schema)
# => { brand: \"Sony\", display_size: \"65\", price: 1999.99 }

Perfect for:

• E-commerce: Parsing product listings and reviews.
• Customer support: Extracting structured insights from support tickets.
• Content management: Automatically structuring documents and articles.

5. ActiveGenie::Lister

Lister simulates a "Family Feud" style survey to generate ordered lists based on popular options.

theme = "Factors consumers consider when buying a smartphone"
result = ActiveGenie::Lister.call(theme)
# => [ "Price", "Battery life", "Camera quality", "Storage capacity", "Brand reputation" ]

Perfect for:

• Market research: Understanding consumer priorities.
• Content strategy: Generating topic ideas based on popular demand.
• Tags and categories: Creating relevant tags for articles or products.

---

The Pillars of ActiveGenie: How We Guarantee Results

Our modules are built on a foundation of principles that ensure reliability, consistency, and flexibility.

Pillar 1: Consistency

For consistency we use a multi-layered approach to force LLMs to produce predictable, high-quality outputs.

• Reasoning Prompting: We use human reasoning techniques to control a model's thought process. Instead of just asking for an answer, we force the model to stage a political debate (Comparator) or form a jury of experts (Scorer). This structured reasoning provides far more consistent results.
• Overfitted Prompts: Each module uses highly specialized prompts that are "overfitted" for a single purpose. This allows for a higher degree of precision and control over the output, making the modules reliable for their specific tasks.

Pillar 2: Verifiable

With every release, we run a suite of 400+ test cases across multiple providers to validate performance. This prevents regressions, ensures quality, and helps you choose the most cost-effective model for your needs.

| Module | Best Precision | Recommended Model | |---|---|---|---| | Extractor | 94% | deepseek-chat | | Comparator | 96% | claude-sonnet-4-20250514 | | Scorer | 83% | deepseek-chat | | Lister | 68% | claude-3-5-haiku-20241022 | | Ranker | 67% | gemini-2.5-flash |

See our detailed benchmark results for full methodology and metrics.

Pillar 3: Model agnostic

The LLM landscape changes daily. ActiveGenie is designed to be model-agnostic, giving you the freedom to adapt without rewriting your code.

• No Vendor Lock-In: Switch between providers with a single line of configuration. Your application code remains unchanged, protected from breaking API changes and provider-specific quirks.
• Use the Best Model for the Job: Mix and match providers and models to optimize for cost, speed, or accuracy on a per-call basis.

# Set OpenAI as the default
ActiveGenie.configure do |config|
  config.providers.openai.api_key = "your-openai-key"
  config.providers.anthropic.api_key = "your-anthropic-key"
  config.providers.default = :openai
end

# Override to use Anthropic for a specific creative task
result = ActiveGenie::Lister.call(
  "Topics for a tech blog",
  config: { model: "claude-3-5-haiku-20241022" }
)

• No More Prompt Engineering Headaches: We handle the complexity of provider-specific formatting, token optimization, and best practices internally. You just call the module.

Supported Providers: OpenAI, Anthropic, Google, DeepSeek, and more coming soon.

---

Your Journey Starts Here

Ready to add reliable GenAI capabilities to your application?

1. Installation - Get ActiveGenie set up in minutes.
2. Quickstart Guide - Try our modules with hands-on examples.
3. Module Documentation - Deep dive into each module's advanced capabilities.

Or explore specific modules based on your needs:

• Comparator - For choosing between alternatives
• Extractor - For structuring unstructured data
• Scorer - For quality assessment and expert evaluation
• Lister - For market research and content planning
• Ranker - For multi-option evaluation and ranking