Tappy

Analyzes the semantic content and tone of a LinkedIn post (including text, engagement patterns, and implicit context signals) to generate contextually relevant comments that match the post's subject matter and professional tone. Uses language model inference to produce comments that reference specific details from the source post rather than generic responses, with post context passed as prompt context to the LLM backbone.

Provides a single-action workflow to generate and immediately insert a comment into LinkedIn's native comment box, with optional preview/edit capability before posting. Integrates with LinkedIn's DOM to detect the comment input field, populate it with generated text, and optionally auto-submit or require user confirmation. Reduces friction from generate-copy-paste-edit cycle to a single tap.

Detects the implicit tone, formality level, and engagement style of a LinkedIn post (e.g., casual vs corporate, thought leadership vs networking) and generates comments that match that tone rather than defaulting to a single generic voice. Analyzes post language patterns, emoji usage, hashtag style, and author profile signals to calibrate response tone, then conditions the LLM generation on detected tone parameters.

Implements a freemium model where free users receive a limited number of comment generations per month (e.g., 5-10), with paid tiers unlocking higher quotas or unlimited generation. Tracks usage per user account via backend state (likely tied to LinkedIn account or email), enforces quota limits client-side and server-side, and surfaces quota status in the UI with upgrade prompts when limits approach.

Allows users to rate generated comments (thumbs up/down or 1-5 star scale) and optionally regenerate if quality is poor. Feedback is collected and may be used to improve future generations (via fine-tuning or prompt optimization), though current implementation likely treats feedback as telemetry rather than real-time personalization. Regeneration triggers a new LLM inference with the same post context, potentially producing a different comment.

Extracts and parses LinkedIn post content (text, hashtags, mentions, links, engagement metrics) from the LinkedIn page DOM or via LinkedIn's API (if available) to provide structured input to the comment generation model. Handles various post formats (text-only, image captions, video descriptions) and normalizes extracted content for downstream processing. May use regex, DOM selectors, or LinkedIn's official API depending on integration approach.

Manages user identity and LinkedIn account linking via OAuth 2.0 or similar protocol, allowing users to authenticate with LinkedIn credentials and authorize Tappy to access post content and post comments on their behalf. Stores user session state and account linkage in backend database, with token refresh logic to maintain valid authentication across sessions.

Posts generated comments directly to LinkedIn on behalf of the user, either via LinkedIn's official API (if available) or via automated form submission (browser extension filling the comment box and clicking submit). Handles rate limiting, error handling (e.g., post deleted, user blocked), and optional confirmation before posting to prevent accidental spam.