Powered by open-weight Gemma · No API keys · Runs local

AlertCadence

The security brief your clients actually read.

It pulls CISA's list of actively-exploited vulnerabilities, keeps the ones that affect software small businesses run, and rewrites each in plain English — using a model that runs entirely on your own machine.

The problem

It's not a threat-comms problem. It's a translation problem.

Small businesses don't read CVEs. They skim. Security advisories are written for engineers — full of jargon, severity scores, and exploit detail that means nothing to the person who owns the bakery.

AlertCadence translates. It takes the same advisories a security team reads and turns each one into three plain sentences: what happened, why it matters to your business, and what's being done about it.

A raw advisory

  • "Path traversal via crafted request"
  • Severity 9.8, CWE-22
  • Client asks "do we need to worry?"

After AlertCadence

  • "An attacker could reach files they shouldn't"
  • "Your customer data could be exposed"
  • "We're patching and watching it now"

How it works

From raw feed to client-ready brief.

Four steps, all running on your own hardware. No cloud API, no account, nothing to leak.

01CISA KEV

Fetch

Pulls CISA's Known Exploited Vulnerabilities feed — every item is already confirmed exploited in the wild.

02Python

Filter

Keeps the last 30 days and the vendors small businesses actually run. Ransomware-linked items rank first.

03Gemma · local

Summarize

A local open model rewrites each advisory into What happened / Why it matters / What we're doing.

04HTML

Send

Renders one clean, white-labeled brief you review and send. (PDF, email & scheduling on the roadmap.)

Every item is already confirmed-exploited
Runs on CPU — no GPU required
No data ever leaves your machine

Why a local model

The open, local choice isn't a constraint — it's the point.

Privacy

Your data never leaves.

The model runs on your own machine with Ollama. Your client list, the advisories you process — none of it is sent to anyone's API.

  • ✓ Runs offline
  • ✓ No API keys
  • ✓ Nothing logged elsewhere

Cost

Nothing per brief.

Because the model is local and open-weight, generating a brief costs nothing — no per-token bill, no usage meter, no surprise invoice.

  • ✓ No cloud AI bill
  • ✓ $0.00 per brief
  • ✓ Open-source engine

Voice

Calm, not alarmist.

Built on the same plain-English philosophy as The Security Dictionary — it says "we're patching this," not "CRITICAL VULNERABILITY DETECTED."

  • ✓ Plain English
  • ✓ No panic
  • ✓ Built for non-technical readers

See a real brief.

This week's issue, generated from live CISA data by a local model.

Open this week's brief →