Boost Targeting AI
Domain: Audience targeting, ad delivery optimization, embedding matching Integrated With:
BoostingModule
,TRNUsageOracle
,SemanticTrustOracle
The Boost Targeting AI is responsible for intelligently delivering boosted content to users most likely to resonate with it. It functions entirely within the TRN ecosystem—using embedded similarity, user content history, ad preferences, and performance pacing to maximize reach, relevance, and return on TRN spent.
It never manipulates visibility for personal gain and operates with full DAO-governed constraints.
🎯 Core Functions
Audience Discovery: Searches for user CIDs with
ads enabled
whose embeddings match the ad contentPrompt Interpretation: Reads the campaign goal as a natural language prompt from the booster and identifies key themes, categories, or emotional tone
Embedding Matching: Compares ad content embeddings (text, audio, video, image) to all known user engagement vectors
Similarity Filtering: Selects a pool of matching CIDs based on tight semantic relevance, adjusts wideness dynamically
Reach Feedback: Reports projected reach and similarity score range back to the booster before launch
Campaign Adjustment: If similarity pool is too narrow or wide, the AI expands or restricts the match band automatically
🧠 Workflow
User initiates a boost
TRNUsageOracle verifies TRN availability and deducts boost budget
User specifies targeting prompt, optional preferences, and spend speed
Boost Targeting AI:
Parses prompt
Fetches embeddings of ad
Matches against all users with
ads turned on
Filters based on region, freshness, content compatibility
Target list is queued
Boosted post is delivered in-feed with 10s delay and resonance bar
Burns, blesses, retrns, and skips update similarity feedback
Campaign depletes according to reach and pacing
⚙️ Parameters
Similarity Threshold
Dynamic, starts tight and expands if reach is too low
Pacing Mode
Fast / Balanced / Slow (affects spend rate)
Campaign Length
Min 1 day, Max 7 days
Spend Control
TRN depletes from booster vault via TRNUsageOracle
Minimum Reach Preview
Provided before launch
Timeout Bar
10s display, user cannot scroll or skip
Audio
On by default at device level
Resonance Feedback
Used for boosting economy and LottoModule input
📊 User Behavior Filters
Boosted posts shown only to:
Users with ads enabled
Users who haven’t seen this ad in the last hour
Users not flagged as AI bots (checked via
AI Bot Verifier
)
AI pacing adjusts based on:
Engagement velocity
Bless/Burn ratio
TRN remaining in campaign budget
🔄 Boost-AI Dialogue
The system allows boosters to interact with the AI conversationally:
Set goals like “target US creators who make motivational content”
Specify audience types: “show to people who recently retrned similar content”
Adjust reach: “go wider”, “focus tighter”
AI confirms projected reach and resonance bands
Once accepted, the budget is locked and the campaign proceeds
🔁 Retrns and Reposts
Retrns of boosted posts point to the original post, not the boosted copy
The original post continues earning TRN normally if retrned
Boosted versions expire independently and do not earn from retrns
🔒 DAO Oversight
The targeting AI cannot be customized to unfairly favor or exclude users
Constraints on semantic targeting are set by country NFTs or the DAO via:
CountryRulesetManager
SemanticTrustOracle
🧬 Related Modules
BoostingModule
Content Similarity Embeddings
SemanticTrustOracle
GeoOracle
LottoModule
Last updated