9AI

The product

9AI is Contained’s standalone multi-LLM platform for bringing artificial intelligence into existing business operations without forcing teams to replace their processes or IT stack. It combines operational data, unstructured documents, and company-specific rules so people can investigate information in natural language and automate work that would otherwise depend on repetitive reading, rekeying, and validation.

The platform can ingest inputs such as packing lists, invoices, phytosanitary documents, manifests, PDFs, scans, and spreadsheets. It then transforms and validates that information for use in reporting, forecasting, customs documentation, and other operational workflows.

Business inputs

• PDFs and scans
• Spreadsheets
• Operational data
• Company rules

9AIInterpret · reason · actMulti-LLM orchestration with retrieval and agent workflows

Business outputs

• Validated data
• Customs documents
• Forecasts and reports
• Natural-language answers

My contribution

As a full-stack developer on the team, I worked across the product surface and the AI-enabled backend: building React and TypeScript interfaces, Python and Flask services running on AWS EC2, PostgreSQL and Redis data flows, and the integration paths that turn model output into usable application behaviour.

• Developed user-facing workflows for configuring, running, and reviewing AI-assisted business processes.
• Connected backend services with multiple model providers and retrieval-based data flows.
• Helped turn document ingestion, validation, business rules, and generated outputs into cohesive end-to-end product experiences.
• Contributed to the special integration that lets 9AI work with BlueRing’s connected, auditable supply-chain data.

The engineering approach

9AI uses a multi-model architecture rather than tying the product to a single provider or task. OpenAI and Anthropic models sit behind orchestration built with LangChain and LangGraph, while retrieval-augmented generation grounds interactions in business data and documents. Python and Flask services run on AWS EC2, with PostgreSQL and Redis supporting persistent and fast-access application data; React and TypeScript provide the interfaces where users guide and inspect the work, with AWS Amplify supporting frontend delivery.

That architecture supports different kinds of intelligence within one product: conversational access to business data, document extraction and transformation, agent-style processing, anomaly detection, forecasting, and automated reporting. The product can apply bespoke business rules so generated outputs reflect how a company actually works instead of producing generic model responses.

The BlueRing integration

9AI and BlueRing are independent products with a deliberate integration boundary. BlueRing provides the connected operational record—manifests, quality data, stock, distribution, farm records, finance, telemetry, and audit history. 9AI provides the intelligence and automation layer that can reason across that record, trace goods in seconds, generate reports, and prepare government workflows.

Because the integration preserves those responsibilities, either platform can deliver value independently. Together, they let teams move from collecting trustworthy data to questioning and acting on it without disrupting their existing systems. See my work on BlueRing.

Operational value

Contained positions 9AI for growers, importers, receivers, freight forwarders, and internal business teams. Its use cases include forecasting and waste reduction, automated ESG reporting, claims analysis, fraud detection, and the preparation of customs workflows for systems such as CDS/HMRC and IPAFFS/DEFRA.

The result is a platform aimed at shortening the distance between receiving information and making a decision: fewer keystrokes, less report hunting, clearer auditability, and automation that remains connected to the source data behind it.