Agentic AI on an AI Development Platform and Azure OpenAI

Taller's engagement with the global technology company is a client-internal-platforms partnership: work delivered against the global technology company's own corporate-systems estate rather than against the global technology company's customers. Pods are sourced primarily from Argentina and Brazil and have produced engineers the global technology company has formally recognized in its own quarterly awards program.

The global technology company's own AI investment requires building on the AI substrate the firm also sells — their AI development platform, Azure OpenAI, the broader Microsoft AI surface — at a depth that exceeds what most external partners can deliver. The work requires substantive depth in multi-agent system design, RAG pipelines, and the framework integrations that the multi-agent architecture requires, with the operational discipline an internal client deployment expects.

Taller's agentic AI pod develops multi-agent system designs, enterprise AI copilots, custom knowledge-base integrations, RAG pipelines built against vector databases (Pinecone, Weaviate), and the framework integrations (LangChain, AutoGen, CrewAI) that the multi-agent architecture requires. The implementation substrate is ASP.NET Core, Entity Framework, SignalR, and Minimal APIs on the .NET side; FastAPI, Django, and Flask on the Python side; Docker for the deployment surface; Azure Service Bus or RabbitMQ for the messaging spine between agents. Building agentic AI against the same platform the client ships to the rest of the world is a particular kind of engineering responsibility — Taller's agents have to operate inside the global technology company's own governance posture, with the same security baselines, compliance constraints, and observability discipline the firm expects of its commercial customers.

Taller's pods are actively developing against the legal department help portal, AI processing solutions for D365, email categorization, and the broader AI workload platform. Engineers operate early on Foundry features released within the last one to two months; work that requires intensive study of the documentation because little external information yet exists.

Working against the global technology company's own AI platform at this depth is the strongest possible commercial reference for the underlying engineering capability. The client has chosen to invest the AI-platform learning curve into the pods that have demonstrated they can absorb it. This is the property that distinguishes this engagement from any other technology-platform partnership in the market.