AI runtime
Provider testing across Azure OpenAI, Gemini / Vertex AI, and Amazon Bedrock.
Workflow data
Source storage, retrieval, queues, run history, and report exports.
Cost control
Usage monitoring, model routing, evaluation traces, and budget guardrails.
Product workloads
AI inference Document processing Vector retrieval Workflow queues Audit reports- Secure customer intake for documents, websites, code snippets, spreadsheets, and operational context.
- Multi-step AI runs that classify tasks, extract structured facts, identify risks, and generate customer-ready outputs.
- Persistent run history with source references, status tracking, approvals, and retryable execution.
- Cost controls, model routing, evaluation traces, and cloud observability for production reliability.
Planned cloud usage
Microsoft Azure
Azure credits will support Azure OpenAI / AI Foundry experiments, hosted APIs, storage, identity, and observability for early workflow automation.
Google Cloud
Google Cloud credits will support Gemini and Vertex AI evaluation, Cloud Run services, Firebase or managed storage for MVP workflows, and analytics as customer workflow data grows.
AWS
AWS Activate credits will support Bedrock model comparisons, containerized job runners, S3-style source storage, managed databases, and queue-based execution for production workloads.
Near-term milestones
- Maintain public website and company-domain email at admin@petaflo.com.
- Package the first workflow templates: operations audit, lead research, document review, and application support.
- Build a hosted MVP with customer intake, run history, source references, approvals, and exportable reports.
- Run provider-by-provider AI quality and cost comparisons across Azure, Google Cloud, and AWS.