Ignite26/27
Back to all phases
Phase 03

November–December 2026

Build

Build is the technical core. You implement RAG against real data and establish an eval pipeline that catches regressions before they reach the client. At the same time you learn to classify the solution under the EU AI Act and to justify it in a business case — technical quality that cannot be defended commercially rarely gets bought.

Learning objectives

  • Implement RAG against enterprise data with measurable quality
  • Establish an eval pipeline that runs in CI
  • Classify a solution under the EU AI Act and document the assessment
  • Build a business case for an AI initiative

Core topics

  • 01RAG engineering: chunking, retrieval, re-ranking, hybrid
  • 02Eval as a discipline: golden sets, rubrics and LLM-as-judge
  • 03Eval in CI — regressions caught before delivery
  • 04EU AI Act: risk classification, DPIA and documentation
  • 05Business case modelling for AI delivery
  • 06Robust AI services: retries, circuit breakers, idempotency (Builder)

Technologies and frameworks

  • LangGraph · Pydantic AI
  • Langfuse · OpenTelemetry
  • Anthropic SDK · OpenAI SDK · Bedrock
  • Vector DBs: pgvector, Pinecone, Weaviate
  • MCP Inspector · custom MCP servers

Sessions in this phase

  • Masterclass: 'how to set up an eval pipeline'
  • Pre-Christmas demo day: show what you've built
  • Dojo on first production challenges
  • Frontier Friday on new agent frameworks

Artefacts Ignitere deliver

  • 01Running agent with tools and HITL approval
  • 02RAG system with eval report
  • 03Eval framework in Git with golden set and automated testing

Courses in this phase

Shared coursesAll Ignitere

  • Business case for an AI initiative

    Build an ROI model that survives the meeting with the client's CFO.

  • RAG engineering and hybrid search

    How to find the right context and deliver grounded answers.

  • EU AI Act risk classification

    Place a system in the right category and document why.

  • Evaluation in practice

    Golden sets, LLM-as-judge and an eval rig you can trust.

Track courses · Ignite Builder

  • Robust AI services in production

    Streaming, retries, circuit breakers — the patterns that keep agents alive in production.

  • Quality gates in the delivery pipeline

    Regression-test prompts and agents — break the build when quality drops.

Track courses · Ignite Tuner

  • Evaluation craft: rubrics and LLM-as-judge

    When the machine does the judging — calibrate it against people you trust.

  • Context engineering: what the model actually sees

    Manage the context window like an engineer, not like a writer

Track courses · Ignite Shaper

  • The EU AI Act in practice: DPIA and documentation

    Produce the documentation pack before the regulator knocks.

  • Process redesign with AI

    Map the existing process, find the AI leverage points, redesign the value stream.

Cross-cutting threads

  • EU AI Act practice
  • Consultant craft
  • Evaluation
Phase 02Discover
Phase 04Operate