In fast-moving teams, timing is everything.
And sometimes, the difference between a great opportunity and a missed one lies buried deep inside — a new lead, a billing event, a sudden spike that no one noticed.
I wanted to change that.
What if there was an agent that could live inside the data, spot important moments the second they happened, and alert the team before anyone even knew they needed to look?
That’s how Pulse Agent was born.
“It’s like having a PM assistant in Slack — working for me 24×7.”
Every day, our systems recorded billing transactions, user signups, new leads...
But amidst the noise of thousands of entries, critical signals — new enterprise leads, major payments, sudden drops — would get buried.
Manually checking dashboards wasn't scalable.
Important moments were slipping through the cracks.
We needed an autonomous agent that could not just track changes — but understand what mattered.
So I decided to step in.
I used to be a developer.
And I still knew how to build.
Pulse Agent was designed with one principle: move as fast as the business does.
Here's how it came together:
Metabase powered structured access to live billing and lead data.
n8n orchestrated the flows — detecting new events, pulling fresh data, triggering smart notifications.
Google Sheets served as a lightweight state keeper — remembering the "last seen" lead or transaction.
Slack Integration made insights instant, not buried in reports.
All just to figure out if a lead was worth reaching out to.
TL;DR: Context was buried. Nobody had the full picture.
But we didn’t stop there.
When a new lead came in, Pulse Agent scraped additional details about the user from Apollo.io — company, designation, social profiles — and then passed all that into an LLM (GPT/Claude) to summarize the user profile into a crisp snapshot for Customer Success teams.
In seconds, the CS team didn’t just know someone signed up — they knew who it was, where they worked, and why it might matter.
Building Pulse Agent meant solving some tricky problems:
State Management: Tracking what had changed since last check, across datasets.
Data Freshness vs. System Load: Querying often enough without stressing systems.
Signal-to-Noise: Avoiding false positives while catching what mattered.
Lead Contextualization: Scraping and summarizing external data to add real-world context — without any manual research.
Making it work felt a lot like training a dog to bark at the right doorbell.
Pulse Agent fundamentally changed how we interacted with our data.
Real-time Slack alerts for new leads, billing transactions, anomalies
Auto-enriched leads with full context for CS teams — designation, company, LinkedIn links
Instant, LLM-summarized user profiles
Reduced manual digging by hours every week
Critical insights surfaced while they were still actionable, not after the moment passed
From an endless scroll of reports to a pulse of real-time action — that's the shift we unlocked.
Pulse Agent is just getting started.
The roadmap ahead:
Anomaly detection models to proactively spot trends and risks
Voice alerts for time-sensitive opportunities
Self-healing triggers that adjust based on business velocity
Multi-channel insights: from Slack to WhatsApp, Email, and beyond
Because in a world moving this fast, data should talk before someone asks.
Role | Tools |
|---|---|
Workflow orchestration |
|
Custom node logic, data cleaning |
|
Data |
|
Scraping APIs |
|
LLMs |
|
Real-time notifications |
|
Everything — solo.
Built the full system end-to-end
Designed enrichment + summary logic
Wrote custom scripts for data joining & tagging
Integrated LLMs and Slack output
Tested, tuned, and rolled out to CS team
Because the data was there.
The need was urgent.
And sometimes the fastest way forward… is to build it yourself.
