Understanding Context-Aware Alerts and AI Explanations

Context-aware filtering is Podscan's most powerful tool for reducing false positives. It uses AI to analyze the full conversation context around a keyword match, only triggering alerts when mentions are genuinely relevant.

How Context-Aware Filtering Works

1. A filter term is matched in a podcast transcript
2. The AI reads the surrounding context (not just the sentence)
3. Your context-aware question is answered by the AI
4. The alert only triggers if the answer is "yes"

Setting Up a Context-Aware Question

When creating or editing an alert:

1. Add your filter keywords as usual
2. In the Context-Aware Question field, enter a yes/no question
3. The AI will answer this question for each potential match

Examples of Effective Questions

Disambiguating common terms:

• Filter: "Apple" → Question: "Is this episode discussing Apple Inc., the technology company?"
• Filter: "Claude" → Question: "Is this episode discussing Claude, the AI assistant from Anthropic?"
• Filter: "Winners" → Question: "Does this discuss competition winners, not the clothing store?"

Ensuring substantive discussion:

• Filter: "climate change" → Question: "Does this episode discuss climate change for more than a minute?"
• Filter: "AI safety" → Question: "Does this episode discuss AI safety in relation to government regulation?"

Filtering context:

• Filter: "podcast marketing" → Question: "Is this a discussion about marketing strategies, not just a passing mention?"
• Filter: "Jeff" → Question: "Is this episode hosted by someone named Jeff?"

Viewing AI Explanations

When a mention passes context-aware verification, you can see exactly why.

1. In your mentions list, look for the Context Verified indicator (sparkle icon)
2. Click on the mention or select View full mention
3. The Episode Details page displays a special panel showing:

• The context-aware question that was asked
• The AI's explanation of why this mention was deemed relevant
• The specific transcript context that informed the decision

Benefits of AI Explanations

• Transparency: Understand the AI's reasoning
• Verification: Confirm the filter is working as intended
• Refinement: Improve your questions based on explanations
• Confidence: Trust that verified mentions are genuinely relevant

Writing Better Context-Aware Questions

Be specific: "Is this about our product X?" works better than "Is this relevant?"

Ask yes/no questions: The AI needs to give a binary answer.

Include distinguishing details: "Is this about Acme Corp the software company, not the cartoon company?"

Focus on what matters: If you only want detailed discussions, ask about depth: "Does this episode discuss X for more than 30 seconds?"

When to Use Context-Aware Filtering

Ideal for:

• Brands with common names (Apple, Amazon, Target)
• Terms with multiple meanings
• Names shared by different entities
• When you only want in-depth coverage, not passing mentions

May not be needed for:

• Unique brand names
• Very specific technical terms
• Niche topics with little ambiguity

Monitoring Filter Performance

If you're getting unexpected results:

1. Review AI explanations for false positives
2. Refine your question to be more specific
3. Consider adding exclusion terms in your main filter (using NOT operator)
4. Adjust restrictions like categories to narrow the search space

Updated on: 18/12/2025