How I Ran My Entire GTM with Prompts for 8 Months

May 2026  ·  6 min read  ·  Sandbox

Eight months ago I was three weeks away from hiring a full-time GTM person. The job description was drafted. The number — $120K–$130K base — was in my head. I was this close.

I didn't hire. Not because the budget fell through, but because I decided to test something first: what if I replaced the actual work — not the strategy, not the judgment — just the execution? Could I run a real GTM motion with prompts instead of people?

Here's exactly what happened.

What I Was Trying to Replace

The GTM role I was about to hire for wasn't one job. It was four jobs compressed into one salary:

Function What it required weekly What I was doing instead
Outbound prospecting 8–10 hrs: list building, sequencing, sending Not doing it consistently
Follow-up management 4–6 hrs: tracking every open deal, timing re-engagements Remembering when I remembered
Content production 5–7 hrs: posts, blogs, email copy Writing when I had gaps (rarely)
Pipeline hygiene 3–4 hrs: CRM updates, stale deal flags, reporting Quarterly at best

Combined: 20–27 hours of execution work per week. I had 4–6. The math was never going to work without adding headcount — or changing what "running GTM" actually required from me.

The Experiment: One Prompt to Start the Week

The premise was simple. I'd write one brief at the start of each week — who I was targeting, what the message angle was, what follow-ups needed to go out. Then Sandbox would handle the execution: build the sequence, schedule the sends, produce the content brief, flag warm leads.

My Monday morning input looked like this:

Monday Brief — Week of Nov 4
Target: boutique consulting firm partners, 10–40 employees, Northeast US, have a service-delivery model that scales with headcount Angle: they're losing warm deals to follow-up lag, not losing on price Follow-up: anyone who opened 2+ emails in the last 30 days but hasn't replied — send a "still relevant?" check-in Content: short post on why consultancies plateau at $2M — referral ceiling + no repeatable outreach motion Tone: direct, no buzzwords, treat them like they've heard every AI pitch already

That was it. 15 minutes, once a week. The execution ran from that input.

Month 1: It Felt Wrong

The first month was uncomfortable. Outreach was going out every day and I wasn't writing each email. Content was publishing and I hadn't spent three hours on it. Sequences were running and I wasn't monitoring them hourly.

I kept waiting for it to break. It didn't.

By the end of month one: 87 cold contacts reached. 54% open rate on the first campaign. Zero replies — but I'd expected that. Email sequences take time to warm. What I hadn't expected: I had 4–5 hours back every single week I'd previously spent on GTM admin.

Months 2–4: The Engine Compounds

This is what nobody tells you about running a consistent GTM motion: it compounds in ways that sprints don't.

By month three, the campaigns had follow-up sequences running automatically on everyone who'd opened but not replied. By month four, I had 200+ contacts in active sequences, a content library of 40+ posts driving inbound discovery, and — for the first time — a pipeline that hadn't gone quiet during a heavy delivery month.

Open rate (month 4)
61%
Prospects in active sequences
340+
Content pieces produced
60+
Weekly founder hours on GTM
3–4 hrs

The content was doing something else I hadn't anticipated: warm inbound. Prospects I'd never contacted were finding the blog, reading it, and booking calls. Not a flood — but 2–3 per month from people who had already decided they liked how I thought about the problem before we ever spoke.

Months 5–8: What Still Required Me

I want to be honest about what prompts didn't replace.

Still Required My Judgment
Positioning decisions
When the messaging angle wasn't landing, I had to figure out why and reframe it. The execution layer could run any angle well — but I had to pick the right angle in the first place.
Still Required My Judgment
Closing conversations
The outreach opened doors. Every door still needed a human to walk through it. Demo calls, follow-up questions about our specific approach, final negotiation — none of that was delegatable.
Still Required My Judgment
ICP refinement
After two months I realized I'd been targeting a segment that opened emails but rarely bought. Adjusting the ICP — that insight came from me reviewing data and making a call. The system ran whatever I told it to run.

That's actually the point. I spent 3–4 hours a week on judgment work. The 20+ hours of execution work ran without me.

8-Month Summary: What Actually Happened

Before (hiring approach)
After (prompt-based execution)

What I'd Tell Myself at Month Zero

The thing that almost stopped me wasn't cost. It wasn't skepticism about whether AI could do this. It was the feeling that I should be more involved — that growth work required my direct time, not just my direction.

That belief was costing me. Not because I was bad at GTM, but because I was doing execution work that didn't require my specific judgment. I was the bottleneck in my own growth motion because I was treating execution like it was judgment.

The experiment that was supposed to last a month is still running. The GTM role I almost hired for? I never posted it.

8 months, one brief per week, 3–5 hours of founder time: 700+ prospects reached, 58–63% open rates, 175+ content pieces live, pipeline running continuously. The cost is $3K–$5K/month. The alternative was $160K+ per year and a 4-month ramp.

See what running GTM with prompts actually looks like — in 15 minutes.

Book a demo: cal.com/edgarinvillamar/15min

Or start directly: app.sandbox.co/signup

Questions? Email rob@sandboxgtm.com