RSS Amplifier

Rin's Newsletter · Jul 30, 2026

Cursor's sales team says you shouldn't build this. They are wrong.

0
Sign in to vote or save

Maria · Rin's Newsletter

Cursor took 45 minutes out of every SDR's day and tripled qualified meetings in the process. Most sales teams still treat those 45 minutes as part of the job.

For context, I’m Rinat — founder of getsally.io, a B2B outbound agency working with 25+ teams across the US and EU, which means most of my week is spent inside sales processes that someone designed two years ago and nobody has questioned since.

George Hou (Head of Growth, Cursor) sat down next to an SDR and watched them open seven tabs and spend 45 minutes assembling context on a single account, only to end up in front of a blank page with nothing written.

Before: the rep pieced together CRM records, call notes, Slack threads and product usage by hand, just to understand one account.

After: they open Slack and the day is already assembled.

  • Every meeting on the calendar arrives with context attached

  • Overnight account movement is flagged, from usage spikes to new signups

  • 25 outbound emails sit drafted, each tied to a real signal and paired with a landing page matched to that contact’s stack and role

  • Pre-call briefs go down to behavioral detail, including that a buyer dislikes being sold to aggressively

  • Stalled deals have already been followed up on, with updates logged back to the team

The tab-switching is gone, but the more important thing is that the blank page went with it. That was never a tooling problem, it was a context problem.

Nothing in the existing stack of tools got thrown out, which is why this took three people rather than a platform migration. Salesforce, Gong, Slack, etc. stayed where they were.

They built in stages, and the chat interface only appeared at the end, once AEs and support were already bringing questions to the system. They started with structure rather than with AI.

They piloted with 8 SDRs. Today the reps extend it themselves, adding 1000+ automations on top of those 500+ skills without engineering involved. Internally they call it their “Cursor moment,” the point where AI became useful because it finally knew their context.

  • Qualified meetings tripled, and the top two reps hit 4x

  • AE ramp time dropped by more than half

  • The original goal was making reps roughly 20% more productive

The advice George Hou (Head of Growth, Cursor) closes on is that most teams shouldn’t try this, because Cursor has its own product data, its own engineers, a build-over-buy culture and years of failed automation attempts behind it.

I think the opposite is true: every team needs to start building now, and the only real question is how deep you go.

Their advantage isn’t the engineering bench. The team was assembled AI-first from the start, so they built around context before any process had time to calcify. If your sales process is already working, that’s the argument for rebuilding it rather than against, because working usually means:

  • your reps compensate for the gaps manually

  • the knowledge sits in a handful of heads

  • the context is scattered across five systems

All three used to be unavoidable, and none of them are now that a machine can read that context directly.

Productivity is starting to look like a function of how many credits your team burns on frontier models, and customizing a process has gotten cheap enough that even planning can be handed over.

Don't copy the scale, copy the order. We run 20+ salespeople against 30 clients in unrelated industries, which means every step gets rewritten per project:

  • Put all account context in one place, so a rep stops opening seven tabs before a call

  • Turn segmentation into explicit logic. Those rules live in the head of whoever runs the project, and writing them down as a skill is what makes them reusable

  • Build narrow tooling for the signals that market runs on, not a general-purpose scraper

  • Add drafts with personalization rules attached, because a signal on its own doesn’t write the email

  • Introduce research agents once there’s structure for them to follow

  • Add convinient chats last, when people already rely on the system

Let the people using the system be the ones who build it. That’s how you get 500+ internal skills instead of a backlog of tickets.

Each new client starts the sequence again, though shallower, because the skills from the last one carry over. That compounding is the return, and it’s why nobody can sell this to you finished.

If your reps are still opening seven tabs before a call, the problem isn’t the reps.

P.S. If you’re running cold sales right now and struggling to cut through the noise, grab 30 min on my calendar. Let’s see if we can help with that.

If you want the mechanics in more detail, the original breakdown covers the architecture and the internal rollout:

No posts

Read the original on rinrin.substack.com

Comments

Nothing yet. Say the first thing.

    Sign in to join the conversation.