The most common question first-time users ask after deploying their first autonomous routine is:

“I configured my competitor tracker to run daily at 08:00 AM, but this morning I received no message on Telegram. Did the routine fail?”

The answer is almost always: No, the routine ran perfectly. Nothing changed on the target website, so Tishtar suppressed the notification.

In traditional software or crude cron-job scripts, a scraper blindly blasts an alert every day, even if the content is 100% identical to yesterday. Tishtar was engineered with an opposite philosophy: Silence is a feature. Your attention is finite, and notification fatigue kills productivity.

What is Delta Memory?

┌──────────────────┐
│ Target Webpage   │
│ (Raw HTML / Doc) │
└────────┬─────────┘
         │
         ▼
┌────────────────────────────────────────────────────────┐
│ Tishtar Pre-Processing & Cryptographic Hashing          │
│ • Strips tracking pixels, ads & dynamic timestamps     │
│ • Computes normalized content hash (SHA-256)           │
└────────┬───────────────────────────────────────────────┘
         │
         ├─── Hash matches previous run? ──► [EARLY EXIT: ZERO NOTIFICATION, ZERO SPAM]
         │
         └─── Hash differs? ──► Send to LLM Engine ──► Format Delta ──► [TELEGRAM ALERT]

Delta Memory is Tishtar’s stateful comparison engine. Unlike stateless LLM tools that treat every single execution as a blank slate, Tishtar remembers what it saw during the previous execution.

Normalized Content Extraction: When Tishtar visits a page, it first strips away rotating banner ads, cookie consent dialogs, and dynamic session tokens that create false alarms.
Cryptographic Checksum Verification: The system computes a normalized SHA-256 hash of the extracted payload.
Early-Exit Gate: If the checksum is identical to yesterday’s snapshot, the routine immediately exits. No redundant LLM analysis is invoked and no duplicate message is sent to your phone.
Semantic Delta Extraction: Only when the checksum differs does Tishtar wake up the language model to analyze what specifically changed and whether it meets your configured threshold.

3 Reasons Why Silence Protects You

1. Zero Channel Spam

If you monitor 10 competitors, 5 regulatory portals, and 20 supplier catalogs, a naive scraper would bombard your Slack or Telegram with 35 messages every single morning. Within a week, your team would mute the channel and miss the one alert that actually mattered.

With Delta Memory, you only receive a ping when something genuine shifts. When Tishtar speaks, you know it requires attention.

2. Streamlined Pipeline Efficiency

Repeatedly running heavy LLM inference on unchanged pages creates unnecessary latency and noise. By stopping the pipeline at the pre-processing layer when no changes are detected, Tishtar skips redundant analysis and keeps your destination channels strictly high-signal.

3. Proof of Execution (Execution Logs)

Wondering if your routine actually ran? You don’t need a spam message to verify it.

Open your Tishtar Dashboard -> Routines.
Look at the Runs History tab.
You will see the timestamp of every execution with status Skipped (No Delta Detected) or Dispatched (Delta Found).

How to Customize Notification Sensitivity

If you do want a heartbeat confirmation or need to adjust sensitivity:

Strict Thresholds: “Only alert me if the price changes by > $5.”
Informational Thresholds: “Alert me on any text change in the first paragraph.”
Summary Fallbacks: Inside Tishtar Studio, you can configure a weekly heartbeat summary (e.g. “Every Friday at 5:00 PM, send a recap confirming that all 10 sources were checked and remained stable”).