Scaling AI coding at work is becoming a process problem
AI tool use is spreading inside a late-stage startup, but the day-to-day setup is messy rather than organized. The company is under pressure to keep growing 15% every quarter, and workers who have lived through re and layoffs treat AI as a way to protect their jobs. AI has also made the company’s physical product more important in the networking space, which pushed one internal manager into a broader AI rollout role.
The main problem is that many people built their own tool setups without IT oversight. The current fix is custom training, folders full of markdown notes, and reusable made with Claude. Sensitive access details, such as server connections, Cloud Run deployment, and , are being kept in a 1Password vault so people can work in a more secure setup.
The hardest part is teaching about 40 people who are mainly motivated by job security while also keeping normal work moving.
Key points
- AI coding adoption is already happening inside the company, but many setups are informal and inconsistent.
- Repeated work is being saved as markdown notes in .
- Claude is being used to turn work patterns into reusable .
- Server, deployment, and details are being managed through a 1Password vault.
- The bigger challenge is not only the tool, but training people under pressure and aligning their habits.