Everyday Data Science
Latest
Agentic workflows now power a third of surveyed enterprise automationAfrica's AI startup ecosystem posts record funding yearNew benchmark results reshape the coding-agent leaderboardNigeria launches national AI strategy with major investment planRwanda's sovereign AI cloud enters public betaThe future of AI agents: from tools to teammates

Building Agentic AI Systems

A practical path from why agents matter to building your first multi-agent pipeline: the landscape, the key design decisions, and the hands-on build.

The lessons

  1. 1

    Part 1 · 7 min

    The Agent Revolution Is Here, and Most Organizations Are Not Ready

    Gartner says 40%+ of agentic AI projects will be cancelled by 2027, and only ~130 of thousands of "agentic" vendors are real. Here's what the research actually shows, and what separates the teams shipping from the ones stuck in pilot purgatory.

  2. 2

    Part 2 · 7 min

    State of Coding Agents: Who Actually Wins on Real-World Tasks?

    Agents score 90%+ on SWE-bench. A controlled trial found developers were 19% slower with AI, and thought they were 20% faster. Why both are true.

  3. 3

    Part 3 · 8 min

    RAG vs. Fine-Tuning: A 2026 Decision Framework for Practitioners

    Stop arguing. Here's a decision tree grounded in cost, latency, and drift.

  4. 4

    Part 4 · 6 min

    Building Multi-Agent Pipelines with LangGraph: A Practical Guide

    LangChain now recommends the tool-based supervisor over create_supervisor. Here's the pattern to build, the three context decisions that decide if it works, and published benchmarks on which architecture actually costs less.

  5. 5

    Part 5 · 7 min

    LangGraph vs AutoGen vs CrewAI: The 2026 Comparison Is Already Out of Date

    The popular three-way comparison is stale: AutoGen is now maintenance-only, folded into the Microsoft Agent Framework (1.0, April 2026). The real live choice is LangGraph vs CrewAI, and the honest first question is whether you need a framework at all.

  6. 6

    Part 6 · 6 min

    Agent Accuracy Went Up. Agent Reliability Barely Moved.

    Twelve metrics, four dimensions, 14 agentic models. The finding: reliability gains lag well behind capability gains. Why a single success rate hides the flaws that actually break deployments.

  7. 7

    Part 7 · 10 min

    Switching Agent Harnesses Changed 13% of Tasks. Rerunning the Same One Changed 13% Too

    Five models, three production harnesses, 447 real GitHub issues. Claude Code and mini-SWE-agent land within five points of each other, a harness swap flips as many tasks as a rerun, and cost per task differs up to 3x. The wrapper decides the bill, not the score.

  8. 8

    Part 8 · 14 min

    Agent Memory Is Three Systems, Not One. Five Papers Prove the Split

    Five papers split agent memory into three jobs: factual recall, behavioural rules, and the cost of remembering. Dense retrieval wins long horizons, compaction beats raw transcripts until wording matters, and corrections need rules, not search.

  9. 9

    Part 9 · 7 min

    Passed Once Is Not Reliable: ThinkingBox Grades Agents on the Database, Not the Demo

    ThinkingBox checks the final database state after an agent finishes, not its summary. Across 121,680 trials, 79,853 failed, and 67% of failures ended with no tool error. One good demo is not reliability.

  10. 10

    Part 10 · 11 min

    A True Tool Result Can Still Be Wrong: ReplayBench Found a 64.9% Error Rate

    TRAP replayed real, unmodified tool results into the wrong task. Agents acted on them anyway: mean final-state error 0.649, and 0.989 across the full benchmark.