The brief is due in an hour. You have 14 tabs open.
SignalGuard fuses twenty-one live feeds — social chatter, weather, airspace, traffic, scanner feeds, NOTAMs, nearby events, federal advisories — into one severity-scored brief.
Built for the person who has to make the call.
No login required to try.
You're triaging by hand. You're trusting your memory. You're hoping the thing that matters bubbled to the top of whichever tab you were looking at when it landed.
"protest tonight 8pm trafalgar"
14:02 · 47 replies · scroll 3 deep to find this
!FDC 3/3941 NY VIP movement
issued 06:14 · TFR coming · ctrl-F for venue zip
"weird helicopter at MSG?"
3.2k upvotes · 217 comments · 4 hours old
"meet point 14:00 trafalgar — no phones"
channel you don't monitor · 8,400 subscribers
+ 10 more tabs · across federal alerts, weather, traffic, scanner feeds, and the four sources you forget to check.
Every event you're monitoring, scored against the same severity scale — risk-ranked, not date-ranked. Composite posture across Chatter, Environment, Movement, and Context. 24h alert feed on the right rail. Stale-scan flags before they become misses.
01 · SCAN
23 signals fan out in parallel. AI scores every post, fuses every layer, and writes the executive synthesis. 15–40 seconds, end to end. No account needed to try.
02 · MONITOR
Save the scan and SignalGuard re-runs it on cadence — 15 minutes during event-day mode. Severity escalation lights up your dashboard and routes alerts to Slack, Teams, email, or SMS.
03 · ACT
Layered situational map for the control room. Severity-scored PDF for the security lead. Audit log for after-action review. Every artifact ready to share before the doors open.
The same model layer the scan tool uses on every brief — surfaced here so you know what the AI is doing and where it stops.
Layer 01 · Per-post scoring
Each X post is read by Claude Haiku 4.5 and classified by severity (Low → Critical), category (impersonation, ticket fraud, threat language, scalper-coordination), and a one-line reason. The model's confidence percentage is shown alongside, so you can weight each call yourself instead of trusting the label blindly.
Layer 02 · Cross-signal synthesis
Once every feed returns — chatter, weather, advisories, traffic, airspace, crime baseline — a second AI pass reads them together and produces an executive narrative plus a short list of recommended actions for the briefing. The synthesis is a draft for the human briefer, not a verdict.
The AI surfaces signal. You make the call.
Confidence scores reflect the model's stated certainty and may be miscalibrated for sarcasm, idioms, or coded language. Always corroborate before acting.
Radius around the venue. Color is severity. Toggle the 4 signal classes independently. Click a pin, read the post — the source URL is always one click away.
"meet point 14:00 trafalgar — no phones on"
0.3mi NW · @protest_london_2026 · 18m ago
Critical signals get full real estate. Benign chatter packs into compact tiles you can dismiss at a glance.
Sample brief · Madison Square Garden
knicks vs heat · 2026-05-11 · 19:30 et · scan + 24h re-poll
Executive synthesis
Elevated. Counter-protest organizing on a Telegram channel not under direct monitoring; estimated 60–80 participants converging on the 7th Ave gate around doors. Severe-weather watch issued for the 17:00–19:00 window. FAA TFR active for VIP movement within 3nm. Crowd density and traffic within normal range for the venue and event size.
Recommended actions
Chatter · 8
↑ 2 elevatedEnvironment · 5
1 elevatedMovement · 5
1 elevatedContext · 5
0 elevatedGenerated by signalguard.app · Claude Haiku 4.5 · public-data feeds only · operator validation required before action
"Read the brief in five minutes. Walk the perimeter knowing what's actually out there."
Chatter · 8
X, Reddit, Bluesky, Mastodon, YouTube, GDELT global news, Telegram, TikTok — keyword + venue scoped.
Environment · 5
NWS weather alerts, Open-Meteo air quality, NOAA SPC severe-weather outlooks, NASA FIRMS wildfires, USGS earthquakes.
Movement · 5
OpenSky aircraft, FAA TFRs, FAA NOTAMs, TomTom traffic incidents, Broadcastify public scanner feeds.
Context · 5
DHS NTAS national advisory, FBI Crime Data Explorer baseline, OpenFEMA disaster declarations, Ticketmaster nearby events, Google Places POI density.
Scan results aren't kept after the session — re-run to refresh.
No tracking pixels, no third-party analytics SDKs in the brief view.
API keys (yours or ours) live in environment variables, never in the page.
PDFs are generated on-demand and streamed to your browser — never archived server-side.
Public data only · no PII collected · GDPR-aware
Most signals surface within minutes of publication. Social chatter (X, Reddit, Bluesky, Mastodon, YouTube, Telegram, TikTok) updates per scan. Weather, airspace, traffic are live. Federal disasters and crime baseline are dailies. The age of every signal is stamped on its card and on the PDF brief — you decide what's still relevant.
Every flagged post carries a confidence score, a written reason, and a link to the raw source. The AI surfaces signal — you make the call. We don't cry wolf for you. If something keeps misfiring, mute it per-signal in your notification prefs.
No. It gives them back the 3–4 hours a day they spend in 14 tabs. SignalGuard is a force multiplier for the human in the loop — the brief, the map, the alert. The judgment call stays with your team.
Yes — run a scan without an account. The single-event brief stays free. Paid tiers start at $19 per scan (PAYG) or $249/mo (Pro) if you want saved-event monitoring, dashboards, alerts, and the situational map.
T-minus · the brief is due
Run a free scan on your next event. 40 seconds. No card. No login. The PDF is yours either way.