Monday, 7 AM.
I open HubSpot like I have every weekday for the past year.
43 open deals staring back at me.
I start at the top. Click into the first deal. Scroll to last activity. July 14th — two weeks ago. I try to remember: was that the SaaS guy? Did we have a call? I add a mental note, close the tab, move to the next one.
40 minutes later I close the laptop and start my day.
And the crazy part? I thought this was good pipeline management.
Here's what I didn't realize I was doing.
I wasn't reviewing my pipeline.
I was performing a ritual that made me feel like I was reviewing my pipeline.
The actual decision — "which deals need attention today?" — takes 30 seconds once you have the right information. The 40 minutes? That was me hunting for information manually. One deal at a time. In a platform that has all of it, but refuses to surface it for you.
HubSpot has everything. Last activity. Deal stage. Email opens. Every data point you'd need.
It just doesn't show you the answer. You're the analyst. Every morning. From scratch.
166 hours a year. Per rep. Just to answer one question: what should I call today?
Oh, and if you're on HubSpot Pro for full data access? $1,200/year per seat. Analyst-level pricing. Manual analyst work.
So I built a HubSpot pipeline automation script.
Runs every 30 minutes. Scores every open deal by value × days since last contact × stage. Sends a Slack message with the top 5 that need attention.
Not a dashboard. Not a new tab. A message.
5 deals. One sentence on each. Done before my coffee gets cold.
Pipeline review: 40 minutes → 4 minutes.
But here's what surprised me most:
The deals it flags aren't the ones I forgot.
They're the ones I planned to follow up on. The call went great. "I'll send that proposal Thursday." Thursday became Monday. Monday became three weeks of silence.
You didn't ghost your prospect. You just didn't have a system that caught the gap.
That's fixable.
How the HubSpot lead scoring works on the free tier
No Marketing Hub Pro required. The free CRM tier gives you API access to every deal and contact field you need.
The scoring engine weights five signals: job title seniority, company size, industry fit, engagement history, and time decay — so a lead who visited your pricing page yesterday outscores one who opened an email three weeks ago.
When a lead crosses the threshold, Slack fires instantly. When a deal goes 7 days without activity, it gets flagged in the morning briefing.
The config is a JSON file. Change the weights without touching code. Works on HubSpot Free, Starter, and Pro.
The real cost of manual pipeline review
166 hours — that's what a daily 40-minute manual pipeline review costs per year, per rep.
At a $50/hour fully-loaded cost, that's $8,300 in labour per rep just to answer the question: what should I follow up on today?
AI operator news
Intercom charges $0.99 per resolved ticket. Do the math.
Intercom just released Fin Apex 1.0. It beats GPT-4o and Claude on customer support benchmarks. It's genuinely impressive.
Then I ran the numbers.
500 tickets a month. $0.99 each. $495/month on top of your existing Intercom plan — which already starts at $74/month.
$570/month to answer customer questions automatically.
The indie operator version: build a support agent that handles your top 10 repeat questions. Node.js, your helpdesk API, zero per-resolution fees. One afternoon. Runs forever.
The lesson isn't "Intercom is bad." It's that enterprise AI pricing is designed for enterprises. If you're not one, don't pay like one.
My agent cost $12 in tokens per session by week 3
Not because the prompts were bad. Because every run dragged the full conversation history into context — and context compounds.
[29/03/2026 7:35 PM] Abbi AI: A technique called xMemory fixes this. Instead of truncating old context or keeping it all, it compresses. The model sees a summary of old context, not the raw transcript.
Result on a CRM agent processing 40 deals per run: token cost dropped from $12 to $4.80 per session. Same output quality. 60% cheaper.
Your company's AI roadmap is not the same as your AI agent
Executives see AI as leverage — same headcount, more output. ICs see it as a threat.
The operators actually shipping useful AI agents are the ones solving IC problems, not boardroom ones. Pipeline review that saves a rep 40 minutes. Dispute alerts that save a founder $400. Ticket deflection that means the support team isn't answering the same question for the 200th time.
Build for the person doing the work. Not the person approving the budget.
How to pull your HubSpot pipeline data with one API call
Pull your pipeline data in under 5 minutes — without building anything.
curl "https://api.hubapi.com/crm/v3/objects/deals?limit=10&properties=dealname,amount,closedate,hs_lastmodifieddate" -H "Authorization: Bearer YOUR_ACCESS_TOKEN"
Replace the token from HubSpot → Settings → Integrations → Private Apps. You'll see your raw deal data in JSON. First step before automating anything.
Automate your HubSpot pipeline review — $79, no subscription
I packaged the full HubSpot CRM automation into a deployable kit: API integration, lead scoring engine, stale deal detection, daily Slack briefing.
15 minute setup. $79 once. Works on HubSpot Free, Starter, and Pro.
More next week.
Pritesh
Building AI agents that actually do things.
Know someone who still does their pipeline review manually every morning? Forward this to them. It takes 30 seconds and might save them 166 hours.