The buyer is the org running ten or more services

The claim

The buyer is not the solo developer and not the FAANG monorepo team — it is the in-between organization running ten or more services across separate repositories, where every new AI-agent session today re-learns the project from zero.

The evidence

Stack Overflow’s 2024 Developer Survey reports 41% of professional respondents work in organizations of 100+ developers. GitHub’s 2023 Octoverse counted 28M+ active repositories owned by organizations (not individuals), and the median enterprise customer maintains dozens of repos in parallel (GitHub enterprise sales disclosures). Datadog’s 2024 State of DevOps notes the median Datadog customer monitors 7+ services in production, with the top quartile above 30. Every one of those services tends to live in its own repository in a polyrepo shop.

In one reference workspace, 31 repositories already share a single agent-context playbook — CLAUDE.md plus MILESTONES.md plus AGENTS.xml — driven by one paste-in bootstrap. That is the buyer profile in one snapshot.

The comparable

HashiCorp’s early growth came from the same shape of buyer: not the solo dev (who used a shell script) and not Google (which had Borg), but the in-between team running 10–100 services that needed Terraform to keep them coherent. HashiCorp reached $200M+ ARR by 2020 selling into that band (HashiCorp S-1, 2021). The agent-context category is the same band one cycle later.

What we ask for

Introductions to engineering leaders running 10–100 services in polyrepo shops — staff engineers, platform leads, head-of-engineering roles. The fastest signal in due diligence is asking them: “how does a new AI-agent session learn what your repo is for today?” If the answer is “we paste a long prompt every time,” that is the buyer.

pie
  title Engineering orgs by service count (Datadog 2024)
  "1-2 services" : 25
  "3-9 services" : 50
  "10-29 services" : 18
  "30+ services" : 7