Built the Digital Twin Universe; core member of the 3-person dark factory team strongdm
Dark Factories

Jay Taylor

Director of AI-Augmented Product Engineering at StrongDM. Built the Digital Twin Universe — faithful behavioral clones of GSuite, Salesforce, Okta, Jira, Slack, and other SaaS services in a matter of weeks. Core member of the 3-person team that runs StrongDM's Level 5 dark factory.

Jay Taylor joined StrongDM’s AI team on July 14, 2025 when Justin McCarthy formally constituted the dedicated factory engineering group. His core contribution: the Digital Twin Universe.

The Digital Twin Universe

McCarthy initially doubted Taylor could do it. His admission, quoted in Dan Shapiro’s follow-up post: “A year ago…I would have said: your enthusiasm is really welcome. Get back to work.”

Taylor built full behavioral clones of:

  • Google Docs, Google Drive, Google Sheets
  • Slack
  • Jira
  • Okta

Each clone is a self-contained Go binary built from public API documentation, then validated against the live service until no behavioral differences remain. The key prompting insight Taylor shared on Hacker News: use the top publicly available reference SDK client libraries as compatibility targets, with a goal of 100% API compatibility.

This creates external ground truth the agent can’t game — the reference SDK was written by humans, independently, and the agent has no access to it during development.

Why This Matters

Before the DTU, testing agent-written code against real SaaS services meant:

  • Rate limits that constrained test volume
  • API costs that accumulated with every test run
  • Inability to test dangerous failure modes safely
  • Flakiness from real network conditions

After the DTU: thousands of scenarios per hour, no rate limits, no costs, full control over failure modes.

The inversion — that cloning a SaaS service is now routine, not prohibitively expensive — is one of the fundamental economic shifts the dark factory enables.