Author Archives: Allen Rodriguez

Cloud SQL for PostgreSQL can store and search embeddings with the pgvector extension, letting application teams add semantic retrieval without introducing a separate vector database. For Generative AI Leader, the design question is whether semantic retrieval belongs beside transactional data or should be separated into a purpose-built retrieval platform. Google Cloud also integrates Cloud SQL […]

Citations make enterprise search answers easier to verify but do not repair weak retrieval. Anthropic’s Claude API can return citations that point to source passages from documents, and search-result content blocks can carry source attribution for retrieval-augmented generation. Those features help users inspect why an answer is supported, but the quality of the experience still […]

Claude batch processing is an execution-mode choice for work that can trade immediacy for lower cost and higher-volume asynchronous processing. Anthropic’s Message Batches API currently charges batch usage at 50 percent of standard API pricing. Most batches are documented as completing much sooner than the maximum window, but applications should design for asynchronous completion rather […]

Claude audit logging is a multi-source evidence problem rather than one feature with one data source. Anthropic now exposes several layers of operational and governance data, and they answer different questions. The Compliance API provides per-event activity records for security, legal, and compliance workflows. Analytics APIs provide aggregated usage and adoption metrics. OpenTelemetry can stream […]

Streaming changes a Claude integration from “wait for one response” into a sequence of events that the application must interpret correctly. Anthropic’s Messages API uses server-sent events (SSE) when streaming is enabled. That lets text, tool-use information, and other deltas arrive incrementally, which can reduce perceived latency and keep long-running HTTP connections active. It also […]

Claude API rate limits are best treated as capacity signals rather than arbitrary request failures. Anthropic separates monthly spend controls from rate limits, and the API can constrain both request frequency and token throughput. A system that watches only requests per minute can therefore look healthy while still exhausting its input- or output-token budget during […]

Idempotency becomes important the moment a Claude-powered workflow can be retried. The production issue is whether another attempt merely repeats model work or can repeat an external side effect. A network timeout, worker restart, queue redelivery, or user double-submit can cause the same logical job to reach the application more than once. If the workflow […]

Claude Code CI workflows are a trust-boundary design problem as much as an automation feature. Claude Code can participate as an interactive collaborator triggered from issues and pull requests or as an automated worker launched by a GitHub Actions event. Anthropic’s official GitHub Action packages Claude Code for this environment, but the surrounding workflow still […]

“custom commands” is still a useful way to think about repeatable Claude Code workflows, but the current product model is broader: Anthropic has merged custom commands into skills. The useful takeaway is how this behavior changes the workload boundary. Existing Markdown files under .claude/commands/ continue to work, while new reusable workflows are generally better created […]

Microsoft’s agent platform is no longer one product with one runtime. The same platform boundaries are central to Microsoft AI-103 preparation, where agent design is best understood as architecture rather than as isolated product features. Teams can choose how much orchestration Microsoft manages and how much they own. Microsoft Foundry Agent Service can run prompt […]

Microsoft AI-103: A2A Protocols for Enterprise Agents Agent-to-agent communication becomes difficult when every team invents its own invocation contract. One agent exposes a REST endpoint, another expects a function call, a third returns a proprietary task object, and none of them describe capabilities in the same way. The Agent2Agent protocol is meant to reduce that […]

Microsoft AI-103: Agent Conversation State Conversation state is what lets an agent behave as though the second turn belongs to the first. It sounds simple until the application has to decide which messages belong together, how long they should persist, what happens when two requests arrive at once, and whether the conversation is allowed to […]

Microsoft AI-103: Agent Retry Policies Retries are one of the easiest ways to make an agent system look more reliable while quietly making it less safe. A transient 429 from a model deployment may deserve another attempt. A malformed tool request does not. A timeout after a payment API call is especially dangerous because the […]

Microsoft AI-103: AI Gateway Semantic Caching Semantic caching changes the economics of an AI gateway by allowing it to reuse an earlier model response for a new prompt that is similar in meaning, not merely identical in text. Azure API Management supports this pattern with LLM semantic cache lookup and store policies backed by an […]

Microsoft AI-103: AI Gateway Token Quotas Large language model capacity is usually measured in tokens, not only requests. Two API calls can consume radically different amounts of model capacity because one contains a short question and the other includes a long document, several tool definitions, and a large response. Azure API Management addresses this mismatch […]