Business as Code: What the Concept Actually Means for a 10-Person Company
“Business as code” sounds like something large companies do — a concept for engineering orgs with dedicated platform teams and DevOps budgets. If you run a 10-person operation, you probably filed it under “not relevant to me yet.”
It is relevant. And it’s simpler than the name suggests.
Here’s what it actually means at your scale: the repeatable parts of running your business should run without your attention. Not with less of your attention. Without it.
Where the Concept Comes From — and What It Doesn’t Mean
Infrastructure as code started in software engineering. Instead of manually configuring servers, you write scripts that define your infrastructure — repeatable, version-controlled, deployable on demand. The same server setup every time, without a human doing it by hand.
Applied to a business, it means the same thing: any process you run the same way more than once is a candidate for becoming “code.” Defined, schedulable, testable, running without your hand on it every cycle.
What it does not mean: complex automation, an engineering team, or a platform built for enterprises. At 10 people, the interface is a prompt. The config file is a Monday morning brief.
What Makes Something “Code” in This Context
Code has four properties that matter here:
- Repeatable — runs the same way every time, not differently depending on who’s doing it or how they’re feeling
- Schedulable — runs on a trigger or a clock, not on someone’s memory or availability
- Testable — you can see whether it ran, what it produced, and whether it worked
- Attention-independent — runs without someone actively starting it every cycle
Now look at your GTM function. Outreach cadence — is it repeatable and schedulable, or does it run when you have time? Follow-up timing — is it triggered by a condition, or by whether you remembered? Content — does it publish on a schedule, or only when inspiration and bandwidth overlap?
Most operators run their GTM the opposite of code. Every execution requires a decision, a block of time, and your attention. That’s why it stops when delivery picks up. The process is real, but it’s not code — it’s manual labor that looks like a system.
What Business as Code Looks Like in Practice
Here’s a concrete translation for a 10-person operator business. Each of these should be running on a schedule, not on your bandwidth:
Outbound Pipeline
25–30 emails per week to ICP contacts, running on a defined cadence regardless of your calendar. The brief tells the system what ICP you’re targeting and what the current message is. Execution runs without you deciding to start it each week.
Follow-Up Cadence
Day 3, 7, 14, and 30 after every conversation — triggered by the event, not a calendar reminder you’ll miss when you’re in a delivery sprint. The condition fires automatically. You only touch replies that need judgment.
Content on Cadence
3–4 content pieces per week, driven by what you put into the Monday brief. Topic direction, audience angle, channel — your judgment, once, on Monday. Execution and distribution happen without your involvement for the rest of the week.
Warm Re-Engagement
Every 45–60 days, dormant warm contacts get a re-engagement sequence — time-triggered, not attention-triggered. You don’t have to remember that someone went quiet six weeks ago. The system does.
The Monday Brief — Your Config File
In infrastructure as code, you write a config file that defines what you want the system to do. It’s declarative: you express intent, the system handles execution.
For a 10-person business, the Monday brief is the config file. It takes about 20 minutes. You express strategic intent — what ICP focus this week, what message angle, what you’re seeing in the pipeline, any positioning shifts — and the execution layer translates that into the week’s GTM output.
That brief produces the week’s outreach, follow-up cadence, content queue, and re-engagement triggers. You didn’t write 30 emails. You didn’t schedule content manually. You expressed what you wanted. The execution layer ran it.
What Stays Human: The Judgment Layer
Code doesn’t replace judgment. It runs the work that doesn’t require it.
The judgment layer — what you own, permanently:
- Positioning decisions — ICP refinement, messaging direction, channel mix. These require context and business instinct, not execution capacity.
- Reading replies — a difficult response, a nuanced “not now,” a warm signal buried in what looks like a rejection. Humans read these better.
- Closing — relationship, negotiation, contract. This is where operator presence matters.
- Strategic pivots — when something stops working and needs a rethink. The execution layer runs what you defined; you decide when the definition needs updating.
These are not code candidates. They require the kind of judgment that doesn’t reduce to a script — and they’re also the work that most operators want to be doing more of, not less.
What Changes When Execution Becomes Code
| GTM activity | Manual model (execution mixed with judgment) | Business as code (execution separate from judgment) |
|---|---|---|
| Outreach volume | Bursts when you have time; silence during delivery months | 25–30 emails/week, continuous — delivery sprint or not |
| Follow-up timing | When you remember — average 8–14 day lag after calls | Day 3, 7, 14, 30 after every conversation, on schedule |
| Content cadence | Posts when inspiration and bandwidth align — rarely both at once | 3–4 pieces/week, driven by Monday brief, no manual production |
| Warm re-engagement | When you go back into the list — which you rarely do | Every 45–60 days, automatically, on time-trigger |
| Pipeline during delivery | Goes dark — September empty after June crunch | Runs regardless of delivery load; pipeline stays warm |
| Founder time on GTM | 10–20 hrs/week (mostly execution and coordination) | 3–5 hrs/week (entirely judgment and closing) |
Why Most Operators Haven’t Done This Yet
The assumption that’s held most 10-person businesses back from this model:
- “This requires technical implementation.” It doesn’t. The interface is a prompt, not a codebase.
- “This requires a team or multiple hires to build.” It doesn’t. It requires one config session and a Monday brief going forward.
- “AI tools already do this — I use ChatGPT.” Tools make you faster at execution. An execution layer removes you from the execution loop entirely. That’s a different thing.
- “I’d need to learn a lot to get this running.” Operators who implement this typically go from kickoff to operational pipeline in their first week — not months.
The shift is conceptual, not technical. You go from “how do I find time for GTM this week?” to “what do I want GTM to accomplish this week?” The first question is a bandwidth question. The second is a judgment question. Only one of them is worth your time as a founder.
Pipeline becomes predictable when execution becomes predictable. That’s what “business as code” means at 10 people. Not a platform. Not a complex system. A config file you write on Monday, and execution that runs the rest of the week without you.
If this framing fits how you’ve been thinking about the problem — the 15-minute call is a diagnostic. We map which parts of your current GTM are execution-dependent, and show you exactly what your business looks like when those parts run as code instead.
Book here: cal.com/edgarinvillamar/15min
Or reach out directly: rob@sandboxgtm.com