# Zooming In, Zooming Out
Case Study #2 from syNRGy
By Navya Rehani Gupta (NRG)
https://synrgy-nrg.com/zooming-in-zooming-out

---

A system that creates synergy between the big picture and the details, so leaders make better decisions, faster.

## Why I Built This

I've been talking about zooming in and zooming out for 5+ years, across two companies, at multiple conferences from San Francisco to Sydney. I strongly believe that the best leaders maintain the right horizon and the right depth in every situation.

I noticed a new opportunity that typically gets solved with a lot of manual analysis. In most companies, everyone has their own wealth of intelligence and their own understanding of the market. Sales had deep customer signal. Marketing tracked competitive narratives. Finance followed earnings reports and investment trends. Nobody had the same view, and nobody had the full picture.

We needed a system that consolidates all of these efforts, maintains both zoom levels simultaneously, and delivers a shared starting point every single day.

### Before vs Now

| Before | Now |
|--------|-----|
| 20+ min aligning on "what happened" | Meetings start at "what should we do" |
| Patterns lost between meetings | Cross-category patterns surface automatically |
| Recency bias drove the agenda | One shared synthesis, every morning at 5am |

## How It Works

Every morning at 5am, the system scans 100+ sources across 10+ categories in HR Tech, correlates signals across accumulated history, and delivers a briefing that connects the dots no single person could.

### The Daily Briefing includes:
- **TL;DR** with Big Picture, So What, and New Ideas
- **Top Market Moves** tagged as NEW, HOT, or tracked
- **Category sections:** AI & Workforce, Industry Landscape, and more
- **Self Score** across 5 dimensions: Signal, Pattern, Tags, Speed, Action

Delivered daily at 5am. Automated. Self-improving.

### The Flow
1000s of zoomed-out and zoomed-in signals > Noise > AI distills to signal > Better decisions, faster. That's the synergy.

"This is already the best newsletter in our industry." -- Lucas Martinez, CEO, Talent.com. On Day 1, after seeing this weekend project.

## Technical Architecture

### Generation
A shell script reads a structured source list, injects a rolling window of previous intel as context, and prompts an AI model with section structure, tagging rules, and self-scoring criteria.

### Validation
Output passes through multiple gates before delivery: minimum size thresholds, required section checks, and AI artifact detection. HTML extraction uses a primary parser with fallback.

### Write-Back Loop
After delivery, a script parses the briefing, extracts structured entries with tags, and appends to a running intel log. Today's output becomes tomorrow's input context, so each briefing builds on everything before it. The loop closes, and the system gets smarter every day.

### Pattern Engine
Fridays ingest the full week. The engine analyzes tag frequency, cross-category correlations, and thread continuity to surface patterns invisible when reading day by day. A market shift on Monday connects to a product launch on Thursday connects to a partnership the next month. The system sees the through-line.

### Self-Scoring
Every briefing scores itself across 5 dimensions: signal quality, pattern depth, tag accuracy, speed, and actionability. Low scores trigger prompt adjustments the next day. The feedback loop is continuous, not periodic.

### On-Demand Deep Dives
Custom commands spawn parallel research agents across multiple topics simultaneously, producing structured analysis that writes back to the intel log, feeding the next day's context window.

### File Structure

```
scripts/
  daily-briefing.sh        # orchestrator
  send-briefing-outlook.py # delivery
  update-briefing-history.py # write-back loop
  auto-deep-dive.sh        # parallel research
  catchup-missed-jobs.sh   # retry logic
data/
  competitors.json         # source of truth
  Market_Analysis.md       # intel log + history
config/
  launchd plists (5)       # scheduling
  .env                     # keys
```

### Stack
Bash, Python, Claude Code, launchd, Outlook API, Vercel

## What I Learned

Three things changed once the system was running:

1. **Decisions got faster.** The zoomed-out trends and zoomed-in signals were already correlated before the meeting started. Instead of spending 20 minutes aligning on what happened, we started with the synthesis and moved straight to decisions.

2. **"So what" became clear.** Before, five people would zoom in on different signals and come away with five different interpretations. Now everyone starts from the same synthesis. The debate moved from "what happened" to "what we should do about it."

3. **New ideas surfaced from gaps nobody was looking at.** The system zooms out across categories that no single person was tracking. Some of our best product ideas came from connections it made across seemingly unrelated signals.

### The Compound Effect

Each day's briefing builds on accumulated context. A zoomed-out market shift on Monday connected to a zoomed-in product launch on Thursday that connected to a partnership deal the next month. The system saw the through-line. No single person would have.

These patterns surface with no manual analysis needed. Noise removed. Recency bias removed. Signal, clear. Leaders walk into meetings already aligned on what matters, and spend time on what to do about it.

## Compare Notes

If you're building something similar or want to compare notes, I'd love to hear from you.
Connect on LinkedIn: https://www.linkedin.com/in/navyarehani/

---

(c) 2026 Navya Rehani Gupta