Week 7 of 8 · Build AI
Documents, channels, process and cloud
Documents and spreadsheets, chat channels beyond the web, the process model under the agent, and the first deployments to a neo-cloud and a major cloud from Terraform.
What the week builds toward
Invoices, contracts, letters and spreadsheets handled; WorkBot on Slack, WhatsApp, SMS and iMessage; explicit execution models with queues and events; deployed on a neo-cloud and on a major cloud, from Terraform through a plan, review and apply pipeline.
The 5 days
- Day 31
Documents and spreadsheets: PDF, DOCX, XLSX and large tables
Native versus pre-parsed document input, Office formats, and SQL over tables too large for a prompt.
You ship: A document intake service for PDF, DOCX and XLSX, with accuracy and cost measured per path.
- Day 32
Multi-channel chat: web, SMS, WhatsApp, iMessage and Slack
One agent and one session store across web, Twilio SMS and WhatsApp, iMessage and Slack.
You ship: The agent on 4 channels with shared sessions, delivery callbacks and retries.
- Day 33
Process lifecycle: queues, pub/sub, SSE, WebSockets and serverless
Request, stream, job or workflow for every operation, with queues, events and the limits of serverless.
You ship: A processing layer with a fair queue, dead letters, versioned events and streams across 2 instances, load tested.
- Day 34
Cloud services: AWS, GCP, Azure and the neo-clouds
The services an AI product needs on each cloud, the neo-cloud alternatives, and how to choose for a customer.
You ship: The monorepo deployed on a neo-cloud and on a major cloud, with cost estimates and a recommendation.
- Day 35
Infrastructure as code: Terraform and deployment targets
The deployment as Terraform, reproduced from a clean account, on Fargate, Kubernetes and an Azure VM, through a pipeline.
You ship: WorkBot's infrastructure as Terraform on 3 targets, through a plan, review and apply pipeline, with a demonstrated rollback.