IntentRecognizer

Category: Specialized | Module: mycontext.templates.free.specialized

Recognizes the true intent behind a question or request by going beyond the surface-level interpretation. Performs multi-layer analysis: surface request → immediate goal → underlying goal → long-term goal → motivations → implicit assumptions → reformulated true intent. Returns the real question and the optimal response strategy.

When to Use

• Customer support — understand what users really need
• Building chatbots and AI assistants
• Pre-processing questions before routing or answering
• Product requirement clarification
• Sales and consulting qualification
• Improving RAG retrieval by understanding the actual need
• Designing user research questions

Quick Start

from mycontext.templates.free.specialized import IntentRecognizer

recognizer = IntentRecognizer()

ctx = recognizer.build_context(
    input="What's the best programming language to learn?",
    context="Asking for a friend who wants to switch careers from accounting",
    depth="comprehensive",
)
result = ctx.execute(provider="openai")
print(result.response)

Methods

build_context(input, context=None, depth="comprehensive")

Parameters:

Parameter	Type	Default	Description
input	str	""	The question or request to analyze
context	str | None	None	Who is asking and why
depth	str	"comprehensive"	Analysis depth

execute(provider, input, context=None, depth="comprehensive", **kwargs)

result = recognizer.execute(
    provider="openai",
    input="Can you help me write a resignation letter?",
    depth="comprehensive",
)

Multi-Layer Intent Analysis

The pattern reveals the full intent stack:

User asks: "What's the best programming language to learn?"

Surface: Information request about programming languages

Immediate goal: Get a language recommendation

Underlying goal: Pick the right language to start learning coding

Long-term goal: Change careers or add programming skills to land a better job

Motivation: Career advancement, higher income, more interesting work

Implicit assumptions: Assumes "best" is universal (it isn't), assumes they need 
to choose before starting, may not know that starting matters more than which

True intent: "Help me get started programming in a way that matches 
my career goals"

Optimal response: Ask about career goals first, then recommend Python/JS 
with a concrete 30-day starter path — not an abstract language comparison

8-Section Analysis

1. Surface Analysis — Explicit request, key terms, literal question type
2. Goal Inference — Immediate, underlying, and long-term goals; success criteria
3. Motivation Analysis — What's driving the request; pain points, constraints, urgency
4. Context Interpretation — Situation, background, stakeholders, environment
5. Implicit Assumptions — Unstated beliefs, potential biases, knowledge gaps, misconceptions
6. Need Classification — Information / decision / action / validation need
7. Reformulated Intent — True intent, real question, optimal response type
8. Recommendation — How to best respond, what to include, what to avoid

Need Classification

Need Type	Description	Optimal response
Information	They need facts or knowledge	Direct answer with explanation
Decision	They need to choose between options	Framework or comparison
Action	They need to do something	Step-by-step guide
Validation	They need confirmation or reassurance	Evaluation with honest assessment

Examples

Customer Support Intent

customer_message = "How do I cancel my subscription?"

result = recognizer.execute(
    provider="openai",
    input=customer_message,
    context="Paying customer, 8 months active, submitted ticket after failed upgrade",
    depth="comprehensive",
)
# Likely reveals: True intent is frustration with upgrade, not cancellation desire
# Optimal response: Acknowledge frustration, offer help with upgrade issue first

Sales Qualification

result = recognizer.execute(
    provider="anthropic",
    input="Do you have an enterprise plan?",
    context="Inbound inquiry from Fortune 500 company, VP Engineering",
    depth="comprehensive",
)
# Reveals: Not just pricing inquiry — evaluating whether vendor is enterprise-ready

Product Requirements

result = recognizer.execute(
    provider="gemini",
    input="Can we add a way to export data?",
    context="Request from power user who uses the tool daily for analytics",
    depth="comprehensive",
)
# Reveals: May want scheduled exports, or specific format for a specific tool

AI Assistant Pre-Processing

# Use IntentRecognizer as a pre-processing step for your AI assistant
def handle_user_query(user_input: str, context: str = ""):
    # Step 1: Understand the real intent
    intent = IntentRecognizer().execute(
        provider="openai",
        input=user_input,
        context=context,
        depth="comprehensive",
    )
    
    # Step 2: Use the analysis to route to the right handler
    # or to enrich the question before answering
    return route_to_handler(intent.response, user_input)

The Gap Between Asked and Needed

The stated question is rarely the full picture

Research in human-computer interaction consistently shows that users' stated questions are proxies for their actual goals. A question like "what's the weather tomorrow?" could mean:

• I'm deciding what to wear (action need)
• I'm deciding whether to cancel outdoor plans (decision need)
• I'm checking if my flight might be delayed (information need)

The intent determines the optimal response format and content.

API Reference

Method	Returns	Description
build_context(input, context, depth)	Context	Assembled context
execute(provider, input, context, depth, **kwargs)	ProviderResponse	Execute analysis
generic_prompt(input, context_section, depth)	str	Zero-cost prompt string