Petaflo is building Zipcode with a Google Cloud-ready AI infrastructure plan.

Company and public materials

What Zipcode does

• Runs close to a customer's repository so private source code does not need to be pasted into cloud-only assistants.
• Builds a traceable run history with prompts, model choices, source references, generated changes, test notes, and approval state.
• Exports code review packets, deployment checklists, implementation notes, and workflow reports for engineering teams.
• Uses cloud infrastructure for account management, artifact storage, observability, queue workers, model evaluation, and secure collaboration.

Planned Google Cloud usage

AI and model evaluation

Vertex AI and Gemini will be used to evaluate coding-agent prompts, compare model quality, generate structured code-review outputs, and build secure AI workflow prototypes.

Hosted product runtime

Cloud Run will host the API and workflow services. Cloud Tasks and Pub/Sub will coordinate long-running agent jobs, repository review queues, and report generation.

Storage, data, and observability

Cloud Storage will store encrypted artifacts and generated reports. Firestore or AlloyDB will hold run metadata and workflow state. Secret Manager, Cloud Logging, Cloud Monitoring, and Cloud Trace will support secure operations and reliability.

Near-term milestones

• Publish the public company website, Google Cloud plan, privacy notice, terms, and contact channel.
• Build a hosted MVP for repository intake, run status, source references, approval logs, and report export.
• Evaluate Gemini and Vertex AI prompts for code understanding, test planning, deployment analysis, and evidence-backed summaries.
• Prepare pilot workflows for engineering teams that need AI coding help without exposing private source to unmanaged services.