https://folarin.dev AI engineering tutorials, deep dives, and project walkthroughs, agentic systems, LangChain, LangGraph, RAG, LLM integration, vector databases, and full-stack AI. en Sun, 02 Aug 2026 23:00:00 GMT https://folarin.dev/blog/a-skill-is-a-dependency-that-runs-code https://folarin.dev/blog/a-skill-is-a-dependency-that-runs-code Sun, 02 Aug 2026 23:00:00 GMT An Agent Skill is not a prompt template. It carries injectable instructions and executable scripts, so treat it like any package you install. Folarin Akinloye Agentic AIAgent SkillsProductionAgentic AIArchitecture https://folarin.dev/blog/skills-vs-mcp-vs-tools-vs-subagents https://folarin.dev/blog/skills-vs-mcp-vs-tools-vs-subagents Sun, 02 Aug 2026 22:30:00 GMT Four different answers to "extend the agent," each solving a different problem. They stack, they do not compete. Folarin Akinloye Agentic AIAgent SkillsMCPTool CallingAgentic AI https://folarin.dev/blog/what-are-agent-skills-progressive-disclosure https://folarin.dev/blog/what-are-agent-skills-progressive-disclosure Sun, 02 Aug 2026 22:00:00 GMT A skill is a folder of know-how an agent loads only when a task needs it. The mechanism underneath is the whole point. Folarin Akinloye Agentic AIAgent SkillsAgentic AILLMArchitecture https://folarin.dev/blog/designing-ux-for-a-confidently-wrong-intern https://folarin.dev/blog/designing-ux-for-a-confidently-wrong-intern Sun, 02 Aug 2026 21:00:00 GMT Traditional UX assumes the system is correct and fast. An AI feature is neither guaranteed. Design for that. Folarin Akinloye AI System DesignAI Product EngineeringProductionArchitectureLLM https://folarin.dev/blog/the-reference-architecture-for-a-production-ai-app https://folarin.dev/blog/the-reference-architecture-for-a-production-ai-app Sun, 02 Aug 2026 19:00:00 GMT The model call is one box out of ten. The other nine are the engineering. Folarin Akinloye AI System DesignAI Product EngineeringArchitectureProductionLLM https://folarin.dev/blog/who-owns-quality-benevolent-dictator https://folarin.dev/blog/who-owns-quality-benevolent-dictator Sun, 02 Aug 2026 17:00:00 GMT On an AI product, quality is a judgment call that needs one owner, and an incident whose postmortem output is a new eval example Folarin Akinloye AI System DesignAI Product EngineeringProductionEvaluationLLM https://folarin.dev/blog/detect-contain-monitor-ai-risks https://folarin.dev/blog/detect-contain-monitor-ai-risks Sun, 02 Aug 2026 15:00:00 GMT You do not eliminate AI-product risk. You bound the blast radius and watch the rate. Folarin Akinloye AI System DesignAI Product EngineeringProductionLLMArchitecture https://folarin.dev/blog/unit-economics-of-an-ai-product https://folarin.dev/blog/unit-economics-of-an-ai-product Sun, 02 Aug 2026 13:00:00 GMT Every request costs real money. Why the unit is cost per task, not per call, and how to keep it profitable. Folarin Akinloye AI System DesignAI Product EngineeringProductionLLMArchitecture https://folarin.dev/blog/llmops-and-the-data-flywheel https://folarin.dev/blog/llmops-and-the-data-flywheel Sun, 02 Aug 2026 11:00:00 GMT Launch is the start of the work. How you operate an AI product, and how production usage compounds into a moat. Folarin Akinloye AI System DesignAI Product EngineeringProductionEvaluationLLM https://folarin.dev/blog/ship-on-containment-not-perfection https://folarin.dev/blog/ship-on-containment-not-perfection Sun, 02 Aug 2026 09:00:00 GMT How to expose a non-deterministic feature to real users safely, without waiting for a flawlessness that never comes Folarin Akinloye AI System DesignAI Product EngineeringProductionArchitectureLLM https://folarin.dev/blog/treat-the-model-as-a-flaky-dependency https://folarin.dev/blog/treat-the-model-as-a-flaky-dependency Sat, 25 Jul 2026 17:00:00 GMT The vendor changes pricing, retires models, and swaps tokenizers under your live product. Wrap it accordingly. Folarin Akinloye AI System DesignAI Product EngineeringProductionLLMArchitecture https://folarin.dev/blog/does-your-problem-need-an-llm https://folarin.dev/blog/does-your-problem-need-an-llm Sat, 25 Jul 2026 15:00:00 GMT The fit gate to run before you build, and what to do when the answer is no Folarin Akinloye AI System DesignAI Product EngineeringLLMArchitectureProduction https://folarin.dev/blog/evals-are-the-product-discipline https://folarin.dev/blog/evals-are-the-product-discipline Sat, 25 Jul 2026 13:00:00 GMT Why eval-driven development done like TDD fails, and what error-analysis-first evaluation looks like instead Folarin Akinloye AI System DesignAI Product EngineeringEvaluationLLMProduction https://folarin.dev/blog/ai-is-not-a-moat https://folarin.dev/blog/ai-is-not-a-moat Sat, 25 Jul 2026 11:00:00 GMT The model is electricity, not the factory. What actually makes an AI product defensible. Folarin Akinloye AI System DesignAI Product EngineeringProductionArchitectureLLM https://folarin.dev/blog/the-afternoon-and-the-quarter https://folarin.dev/blog/the-afternoon-and-the-quarter Sat, 25 Jul 2026 09:00:00 GMT Why an AI demo takes an afternoon and an AI product takes a quarter, and what lives in the gap Folarin Akinloye AI System DesignAI Product EngineeringProductionArchitectureEvaluationLLM https://folarin.dev/blog/evaluating-agents-locally-with-arize-phoenix https://folarin.dev/blog/evaluating-agents-locally-with-arize-phoenix Sun, 12 Jul 2026 23:13:00 GMT Trace and judge your agent on your own machine, no hosted service required Folarin Akinloye Agentic AIEvaluationObservabilityAgentic AIPythonLLM https://folarin.dev/blog/what-is-agentops https://folarin.dev/blog/what-is-agentops Sun, 12 Jul 2026 23:12:00 GMT How MLOps and LLMOps extend to autonomous, multi-step agents Folarin Akinloye Agentic AIAgentic AIProductionObservabilityEvaluationMulti-Agent https://folarin.dev/blog/opentelemetry-for-agent-observability https://folarin.dev/blog/opentelemetry-for-agent-observability Sun, 12 Jul 2026 23:11:00 GMT The vendor-neutral standard your agent traces should be built on Folarin Akinloye AI InfrastructureObservabilityArchitectureProductionLLM https://folarin.dev/blog/tracing-in-agent-observability https://folarin.dev/blog/tracing-in-agent-observability Sun, 12 Jul 2026 23:10:00 GMT Traces, spans, and why a readable trace is the fastest way to debug a misbehaving agent Folarin Akinloye AI InfrastructureObservabilityAgentic AIProductionArchitectureLLM https://folarin.dev/blog/what-the-research-says-about-prompting-reasoning https://folarin.dev/blog/what-the-research-says-about-prompting-reasoning Fri, 03 Jul 2026 09:50:00 GMT The taxonomy of reasoning elicitation, and the uncomfortable retrieval-vs-reasoning debate every agent builder should sit with Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluationAgentic AI https://folarin.dev/blog/a-field-guide-to-model-specific-prompting https://folarin.dev/blog/a-field-guide-to-model-specific-prompting Fri, 03 Jul 2026 09:40:00 GMT How prompting actually differs across Claude, GPT, Gemini, Llama, and Mistral, and what transfers Folarin Akinloye LLM IntegrationPrompt EngineeringLLMProductionArchitecture https://folarin.dev/blog/graduate-job-classification-a-prompt-engineering-case-study https://folarin.dev/blog/graduate-job-classification-a-prompt-engineering-case-study Fri, 03 Jul 2026 09:30:00 GMT What a production classification system teaches about which prompt tweaks actually move the needle Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluationProduction https://folarin.dev/blog/prompt-functions-prompts-as-reusable-functions https://folarin.dev/blog/prompt-functions-prompts-as-reusable-functions Fri, 03 Jul 2026 09:20:00 GMT A 2023 chat trick that quietly became how we build LLM systems Folarin Akinloye LLM IntegrationPrompt EngineeringLLMTool CallingArchitecture https://folarin.dev/blog/generating-code-with-prompts https://folarin.dev/blog/generating-code-with-prompts Fri, 03 Jul 2026 09:10:00 GMT Comments to code, SQL from schemas, and the verification habit that separates working code from plausible code Folarin Akinloye LLM IntegrationPrompt EngineeringLLMPythonProduction https://folarin.dev/blog/dataset-diversity-tackling-repetitive-generations https://folarin.dev/blog/dataset-diversity-tackling-repetitive-generations Fri, 03 Jul 2026 09:00:00 GMT Why your generated dataset all sounds the same, and the seeding trick that fixes it Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluationPython https://folarin.dev/blog/building-agents-that-actually-work https://folarin.dev/blog/building-agents-that-actually-work Thu, 02 Jul 2026 11:00:00 GMT The practical habits that separate a reliable agent from a flaky demo: simpler workflows, better information flow, and a debugging ladder Folarin Akinloye Agentic AIAgentic AIArchitecturePythonTool CallingLLM https://folarin.dev/blog/generating-a-synthetic-dataset-for-rag https://folarin.dev/blog/generating-a-synthetic-dataset-for-rag Thu, 02 Jul 2026 09:25:00 GMT Use a big model to train a small retriever, for about $55 instead of a month of labeling Folarin Akinloye RAGRAGRetrievalEmbeddingsPrompt EngineeringPython https://folarin.dev/blog/generating-synthetic-data-with-prompts https://folarin.dev/blog/generating-synthetic-data-with-prompts Thu, 02 Jul 2026 09:20:00 GMT Turn a model into a data factory for tests, evals, and cold-start training, without spending a month labeling Folarin Akinloye LLM IntegrationPrompt EngineeringLLMPythonEvaluation https://folarin.dev/blog/bias-in-prompting https://folarin.dev/blog/bias-in-prompting Thu, 02 Jul 2026 09:15:00 GMT The distribution and order of your few-shot examples can quietly bias the model. Here is how to test for it. Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluation https://folarin.dev/blog/factuality-prompting-to-reduce-hallucination https://folarin.dev/blog/factuality-prompting-to-reduce-hallucination Thu, 02 Jul 2026 09:10:00 GMT Three prompt-level moves that cut made-up answers, and where prompting stops being enough Folarin Akinloye LLM IntegrationPrompt EngineeringLLMRAGEvaluation https://folarin.dev/blog/adversarial-prompting-injection-leaking-jailbreaking https://folarin.dev/blog/adversarial-prompting-injection-leaking-jailbreaking Thu, 02 Jul 2026 09:05:00 GMT The three attacks that break LLM apps, and the defences that actually hold up Folarin Akinloye LLM IntegrationPrompt EngineeringLLMProductionArchitecture https://folarin.dev/blog/graph-prompting-explained https://folarin.dev/blog/graph-prompting-explained Thu, 02 Jul 2026 09:00:00 GMT What it means to prompt a model with nodes and edges, and when it actually helps Folarin Akinloye LLM IntegrationPrompt EngineeringLLMArchitectureRAG https://folarin.dev/blog/code-agent-vs-tool-calling-agent https://folarin.dev/blog/code-agent-vs-tool-calling-agent Thu, 02 Jul 2026 09:00:00 GMT The difference between an agent that writes Python and one that emits JSON, with strengths, limits, use cases, and when to pick each Folarin Akinloye Agentic AIAgentic AITool CallingPythonArchitectureLLM https://folarin.dev/blog/prompting-reasoning-models https://folarin.dev/blog/prompting-reasoning-models Wed, 01 Jul 2026 09:05:00 GMT Why the step-by-step prompts you learned in this series can hurt an o3 or Claude with thinking on Folarin Akinloye LLM IntegrationPrompt EngineeringLLMAgentic AIProduction https://folarin.dev/blog/multimodal-chain-of-thought-explained https://folarin.dev/blog/multimodal-chain-of-thought-explained Wed, 01 Jul 2026 09:04:00 GMT A two-stage framework that generates a rationale from text and image first, then infers the answer from that rationale Folarin Akinloye LLM IntegrationPrompt EngineeringLLMArchitecture https://folarin.dev/blog/program-aided-language-models-pal https://folarin.dev/blog/program-aided-language-models-pal Wed, 01 Jul 2026 09:03:00 GMT Program-Aided Language Models offload the actual computation to an interpreter, so arithmetic bugs stop wrecking your reasoning Folarin Akinloye LLM IntegrationPrompt EngineeringLLMPythonTool Calling https://folarin.dev/blog/directional-stimulus-prompting-explained https://folarin.dev/blog/directional-stimulus-prompting-explained Wed, 01 Jul 2026 09:02:00 GMT A small tuneable policy model generates hints that steer a frozen black-box LLM, no access to its weights required Folarin Akinloye LLM IntegrationPrompt EngineeringLLMArchitectureProduction https://folarin.dev/blog/active-prompt-explained https://folarin.dev/blog/active-prompt-explained Wed, 01 Jul 2026 09:01:00 GMT Use the model's own uncertainty to pick which questions are worth a human-written reasoning chain Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluationPython https://folarin.dev/blog/automatic-reasoning-and-tool-use-art https://folarin.dev/blog/automatic-reasoning-and-tool-use-art Wed, 01 Jul 2026 09:00:00 GMT Automatic Reasoning and Tool-use, and why it reads like an early sketch of the agents we build today Folarin Akinloye LLM IntegrationPrompt EngineeringTool CallingLLMAgentic AI https://folarin.dev/blog/generated-knowledge-prompting https://folarin.dev/blog/generated-knowledge-prompting Mon, 29 Jun 2026 09:10:00 GMT Have the model write down what it knows first, then answer using that as context Folarin Akinloye LLM IntegrationPrompt EngineeringLLMRAG https://folarin.dev/blog/self-consistency-prompting https://folarin.dev/blog/self-consistency-prompting Mon, 29 Jun 2026 09:08:00 GMT Run the same reasoning prompt several times, then take the majority answer Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluation https://folarin.dev/blog/chain-of-thought-prompting https://folarin.dev/blog/chain-of-thought-prompting Mon, 29 Jun 2026 09:06:00 GMT Standard and zero-shot CoT, when step-by-step reasoning actually helps, and what it costs Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluation https://folarin.dev/blog/zero-shot-vs-few-shot-prompting https://folarin.dev/blog/zero-shot-vs-few-shot-prompting Mon, 29 Jun 2026 09:04:00 GMT When examples help, how many to use, how to pick them, and where few-shot quietly breaks down Folarin Akinloye LLM IntegrationPrompt EngineeringLLMEvaluation https://folarin.dev/blog/the-anatomy-of-a-good-prompt https://folarin.dev/blog/the-anatomy-of-a-good-prompt Mon, 29 Jun 2026 09:02:00 GMT Instruction, context, input, and output indicator, plus the design habits that make them work Folarin Akinloye LLM IntegrationPrompt EngineeringLLMProduction https://folarin.dev/blog/llm-settings-that-change-your-output https://folarin.dev/blog/llm-settings-that-change-your-output Mon, 29 Jun 2026 09:00:00 GMT Temperature, top_p, max tokens, stop sequences, and the penalties, and when to touch each Folarin Akinloye LLM IntegrationPrompt EngineeringLLMProduction https://folarin.dev/blog/building-a-react-agent-loop-from-scratch https://folarin.dev/blog/building-a-react-agent-loop-from-scratch Sun, 28 Jun 2026 09:00:00 GMT Reason, act, observe, repeat. The whole engine is about 20 lines, and one counting rule explains all of it. Folarin Akinloye Agentic AIAgentic AITool CallingPythonArchitectureLLM https://folarin.dev/blog/git-worktrees-for-parallel-ai-agents https://folarin.dev/blog/git-worktrees-for-parallel-ai-agents Thu, 25 Jun 2026 09:05:00 GMT Worktrees have existed since 2015. They got popular in 2026 because we stopped working one branch at a time. Folarin Akinloye AI InfrastructureGitArchitectureAgentic AIMulti-Agent https://folarin.dev/blog/claude-tag-agent-in-your-slack https://folarin.dev/blog/claude-tag-agent-in-your-slack Thu, 25 Jun 2026 09:00:00 GMT Anthropic put an agent in the channel where the work already happens. Here is why that matters more than the model behind it. Folarin Akinloye Agentic AIAgentic AIMulti-AgentClaudeSlackProduction https://folarin.dev/blog/monitoring-model-deprecation-in-production https://folarin.dev/blog/monitoring-model-deprecation-in-production Wed, 24 Jun 2026 10:00:00 GMT The model you pinned will be switched off one day, probably with less notice than you would like. Here is how to find out from CI instead of from your users. Folarin Akinloye AI InfrastructureProductionLLMArchitecturePython https://folarin.dev/blog/observability-for-llm-apps https://folarin.dev/blog/observability-for-llm-apps Wed, 24 Jun 2026 09:25:00 GMT Your 500s and latency graphs will look fine while the product quietly gives wrong answers. LLM observability is about catching the failures that do not throw. Folarin Akinloye AI InfrastructureProductionLLMEvaluationAgentic AIArchitecture https://folarin.dev/blog/graph-vs-state-machine-vs-loop-for-agents https://folarin.dev/blog/graph-vs-state-machine-vs-loop-for-agents Wed, 24 Jun 2026 09:20:00 GMT Most agents start as a while loop and should stay that way. Here is how to know when you have actually outgrown it. Folarin Akinloye Agentic AIAgentic AILangGraphArchitectureMulti-AgentProduction https://folarin.dev/blog/caching-agent-tool-calls https://folarin.dev/blog/caching-agent-tool-calls Wed, 24 Jun 2026 09:15:00 GMT Prompt caching saves you tokens. Caching tool calls saves you the slow, flaky, expensive thing the tool actually does. Folarin Akinloye AI InfrastructureTool CallingAgentic AIProductionLLMArchitecture https://folarin.dev/blog/building-a-multi-tenant-rag-system https://folarin.dev/blog/building-a-multi-tenant-rag-system Wed, 24 Jun 2026 09:10:00 GMT The day one customer retrieves another customer's documents, it is over. Here is how to make sure that never happens. Folarin Akinloye RAGRAGRetrievalVector DatabasesArchitectureProduction https://folarin.dev/blog/streaming-tool-calls-in-langgraph https://folarin.dev/blog/streaming-tool-calls-in-langgraph Wed, 24 Jun 2026 09:05:00 GMT Token streaming is the easy part. Streaming what a tool is doing, mid-call, is what makes an agent feel alive. Folarin Akinloye Agentic AILangGraphStreamingTool CallingAgentic AIPython https://folarin.dev/blog/five-projects-to-master-ai-engineering https://folarin.dev/blog/five-projects-to-master-ai-engineering Tue, 23 Jun 2026 09:25:00 GMT Skip the toy chatbots. Build the five systems that teach the hard parts: retrieval, agents, evals, cost, and reliability. Folarin Akinloye AI InfrastructureRAGAgentic AIEvaluationProductionArchitecture https://folarin.dev/blog/human-in-the-loop-in-deepagents https://folarin.dev/blog/human-in-the-loop-in-deepagents Tue, 23 Jun 2026 09:20:00 GMT Why some tool calls need a person, and how to gate them with interruptOn, decisions, and a checkpointer Folarin Akinloye Agentic AIDeep AgentsAgentic AIProductionTypeScriptLangChain https://folarin.dev/blog/skills-in-deepagents https://folarin.dev/blog/skills-in-deepagents Tue, 23 Jun 2026 09:15:00 GMT Progressive disclosure, the three load levels, and how to write a SKILL.md the agent actually activates Folarin Akinloye Agentic AIDeep AgentsAgentic AILangChainTypeScriptTool Calling https://folarin.dev/blog/memory-in-deepagents https://folarin.dev/blog/memory-in-deepagents Tue, 23 Jun 2026 09:10:00 GMT Filesystem-backed memory, user versus agent scope, and background consolidation, with the TypeScript to wire each one Folarin Akinloye Agentic AIDeep AgentsAgentic AILangChainTypeScriptProduction https://folarin.dev/blog/subagents-in-deepagents https://folarin.dev/blog/subagents-in-deepagents Tue, 23 Jun 2026 09:05:00 GMT How the task tool works, when to delegate, and the config that decides whether subagents help or hurt Folarin Akinloye Agentic AIDeep AgentsMulti-AgentAgentic AITypeScriptLangChain https://folarin.dev/blog/context-engineering-in-deepagents https://folarin.dev/blog/context-engineering-in-deepagents Tue, 23 Jun 2026 09:00:00 GMT What goes into a deep agent's context, what the framework manages for you, and the knobs you actually control Folarin Akinloye Agentic AIDeep AgentsAgentic AILangChainTypeScriptArchitecture https://folarin.dev/blog/prompt-caching-for-llm-apps https://folarin.dev/blog/prompt-caching-for-llm-apps Mon, 22 Jun 2026 09:05:00 GMT How providers cache your prompt prefix, the real discounts, and the prompt structure that decides whether you save anything Folarin Akinloye LLM IntegrationLLMProductionArchitectureRAG https://folarin.dev/blog/langgraph-state-checkpointing-and-persistence https://folarin.dev/blog/langgraph-state-checkpointing-and-persistence Mon, 22 Jun 2026 09:04:00 GMT How a graph remembers: state channels, checkpointers, threads, and the time travel you get for free Folarin Akinloye Agentic AILangGraphLangChainAgentic AIArchitectureProduction https://folarin.dev/blog/cutting-llm-cost-and-latency https://folarin.dev/blog/cutting-llm-cost-and-latency Mon, 22 Jun 2026 09:03:00 GMT Measure first, then reach for caching, routing, smaller models, and the structural fixes that actually move the numbers Folarin Akinloye LLM IntegrationLLMProductionArchitectureEvaluation https://folarin.dev/blog/guardrails-and-safety-for-agents-in-production https://folarin.dev/blog/guardrails-and-safety-for-agents-in-production Mon, 22 Jun 2026 09:02:00 GMT Defense in depth for agents: input rails, output checks, tool limits, and the injection problem that will not fully go away Folarin Akinloye Agentic AIAgentic AIProductionTool CallingLLMArchitecture https://folarin.dev/blog/streaming-llm-responses-end-to-end https://folarin.dev/blog/streaming-llm-responses-end-to-end Mon, 22 Jun 2026 09:01:00 GMT Why SSE won, how the token stream flows from the provider to the browser, and where it breaks Folarin Akinloye LLM IntegrationLLMStreamingFastAPIProductionFull-Stack https://folarin.dev/blog/learning-production-ai-by-dissecting-open-source https://folarin.dev/blog/learning-production-ai-by-dissecting-open-source Mon, 22 Jun 2026 09:00:00 GMT There are no good production-grade courses, so I turn the codebases real teams run into my curriculum Folarin Akinloye AI InfrastructureProductionArchitectureAgentic AIFull-Stack https://folarin.dev/blog/vercel-eve-deep-dive-research-agent https://folarin.dev/blog/vercel-eve-deep-dive-research-agent Sun, 21 Jun 2026 09:00:00 GMT What Eve is, every capability that matters, a full TypeScript research agent, and how it compares to LangChain's Deep Agents Folarin Akinloye Agentic AIEveVercelTypeScriptAgentic AIDeep Agents https://folarin.dev/blog/agent-memory-short-term-vs-long-term https://folarin.dev/blog/agent-memory-short-term-vs-long-term Sat, 20 Jun 2026 09:25:00 GMT Threads are short-term memory. Stores are long-term memory. Most agent memory bugs come from confusing the two. Folarin Akinloye Agentic AIAgentic AILangGraphLLMArchitectureProduction https://folarin.dev/blog/structured-outputs-and-function-calling https://folarin.dev/blog/structured-outputs-and-function-calling Sat, 20 Jun 2026 09:20:00 GMT How to make an LLM return data that always fits your schema, and how function calling is the same idea wearing a different hat Folarin Akinloye LLM IntegrationLLMTool CallingStructured OutputsPythonProduction https://folarin.dev/blog/choosing-a-vector-database-in-2026 https://folarin.dev/blog/choosing-a-vector-database-in-2026 Sat, 20 Jun 2026 09:15:00 GMT Most teams should start with pgvector and only move when they have a real reason. Here is how to know when that day comes. Folarin Akinloye AI InfrastructureVector DatabasesRAGRetrievalProductionArchitecture https://folarin.dev/blog/embeddings-explained-for-engineers https://folarin.dev/blog/embeddings-explained-for-engineers Sat, 20 Jun 2026 09:10:00 GMT What embeddings really are, why cosine similarity works, and how to pick a model without trusting a leaderboard blindly Folarin Akinloye RAGEmbeddingsRetrievalRAGVector DatabasesLLM https://folarin.dev/blog/reranking-in-rag https://folarin.dev/blog/reranking-in-rag Sat, 20 Jun 2026 09:05:00 GMT A reranker can rescue a mediocre retriever or waste 200ms on an already-correct answer. Knowing which is the whole skill. Folarin Akinloye RAGRAGRetrievalEmbeddingsProductionEvaluation https://folarin.dev/blog/chunking-strategies-for-rag https://folarin.dev/blog/chunking-strategies-for-rag Sat, 20 Jun 2026 09:00:00 GMT Why recursive splitting is the right default, and the few cases where you should reach for something fancier Folarin Akinloye RAGRAGRetrievalEmbeddingsVector DatabasesProduction https://folarin.dev/blog/deep-agents-langchain-customer-support-agent https://folarin.dev/blog/deep-agents-langchain-customer-support-agent Fri, 19 Jun 2026 14:00:00 GMT Why Deep Agents exists, what it gives you out of the box, and a full support agent you could actually ship Folarin Akinloye Agentic AIDeep AgentsLangChainAgentic AIMulti-AgentPython https://folarin.dev/blog/evaluating-agents-with-langsmith https://folarin.dev/blog/evaluating-agents-with-langsmith Fri, 19 Jun 2026 13:00:00 GMT Why "it worked when I tried it" is not evaluation, and how to measure agents properly with tracing, datasets, evaluators, and experiments Folarin Akinloye Agentic AIEvaluationLangSmithLangChainAgentic AIPython https://folarin.dev/blog/building-mcp-servers-with-fastmcp https://folarin.dev/blog/building-mcp-servers-with-fastmcp Fri, 19 Jun 2026 12:00:00 GMT The anatomy of a server, the rules that matter, stdio vs remote, the pitfalls I hit, and how to test and ship it Folarin Akinloye Agentic AIMCPFastMCPPythonTool CallingArchitecture https://folarin.dev/blog/giving-agents-tools-and-mcp https://folarin.dev/blog/giving-agents-tools-and-mcp Fri, 19 Jun 2026 11:00:00 GMT How agents actually do things, how to design tools they use well, and where MCP fits Folarin Akinloye Agentic AIAgentic AILangChainMCPTool CallingPython https://folarin.dev/blog/agentic-rag-vs-traditional-rag https://folarin.dev/blog/agentic-rag-vs-traditional-rag Fri, 19 Jun 2026 10:00:00 GMT When retrieval stops being a fixed step and becomes something the agent decides Folarin Akinloye RAGRAGRetrievalAgentic AILangGraphArchitecture https://folarin.dev/blog/what-are-ai-agents-and-multi-agent-systems https://folarin.dev/blog/what-are-ai-agents-and-multi-agent-systems Fri, 19 Jun 2026 09:00:00 GMT A plain-English tour of agents, why people wire several together, and how to build one in Python Folarin Akinloye Agentic AIAgentic AIMulti-AgentLangGraphLangChainArchitecture