Find the conversations worth joining.

An AI-powered radar for X posts where a useful reply can build visibility, credibility, and a better professional network.

Scheduling next scan…

No completed scans yet

X SearchCandidate PostsRelevance AnalysisEngagement AnalysisOpportunity RankingBest Conversations

Public post data sourced from X. This independent project is not affiliated with or endorsed by X.

Latest candidates

Sort by

Reading the latest signals…

01CollectX API
Searches the previous 12 hours for active conversations about AI agents, AI-assisted SDLC, software engineering, RAG, AWS/Bedrock, AI problems, and tools such as Claude, ChatGPT, Gemini, Copilot, Cursor, and Windsurf.
Why: The focused topic set matches Ali’s expertise, while minimum reply, like, or repost activity avoids spending the result set on unseen posts.
02ValidateRules
Removes Ali’s own posts, reposts, malformed results, and duplicates before scoring; low-value candidates remain visible.
Why: Every valid candidate is shown, while the score makes spam, promotions, and engagement bait easy to skip.
03DeduplicateSeen-ID cache
Excludes post IDs returned within the previous 24 hours, including the currently displayed snapshot.
Why: Each paid refresh prioritizes new conversations without making extra X requests solely to replace repeats; the short-lived ID cache expires automatically.
04UnderstandLocal analysis
Scores professional relevance, room to add value, and network fit with deterministic server-side analysis.
Why: X content stays inside the application by default. Gemini inference is available only after the disclosed processing is approved and paid-service data terms are confirmed.
05MeasureMetrics
Scores public views, interaction quality, activity velocity, and freshness from the metrics returned by X.
Why: Views and interactions are the dominant criteria because a relevant post still needs real distribution and active participation.
06RankHybrid score
Combines every signal into a 0–100 Opportunity Score and orders all candidates from strongest to weakest.
Why: Green means check (70+), orange means maybe (55–69), and red means skip (below 55). Views and interactions dominate, while a quality floor prevents irrelevant virality from being recommended.
07CacheRailway volume
Writes only the latest successful snapshot and usage counter to the persistent /data volume; page visits read that cache and never call X.
Why: A server environment variable controls the scan cadence without a code change. One Railway replica owns that scheduler and its concurrency lock, preventing duplicate paid scans.
08Owner scanSecure backend
Allows the owner to request an immediate scan and restart the configured countdown through a protected server workflow.
Why: Owner authorization and server-enforced usage controls keep the paid action private and spending bounded.

Opportunity Score weights

View reach: 32%
Existing interactions: 28%
Interaction velocity: 15%
Professional relevance: 12%
Ability to add value: 7%
Freshness: 4%
Audience value: 2%

How the radar works

Latest candidates