Skip to content
Nerdy beta

Effective AI

Add to Space
 Flow AI Learning

Your cohort leaves with something real they built. The engagement starts with the problem you pick.

Contact us

Our Effective AI is an immersive workshop, not just another tooling course. Your teams pick a real problem before we start, then spend the engagement building something that solves it: a development harness, a set of reusable skills, a Claude Project, an agentic workflow. Along the way they learn prompt, context, and loop engineering, how to decompose their own work into AI-augmentable tasks, and how to turn tacit knowledge into explicit instructions a model can follow.

Sessions split roughly half instruction, half building. Builder time sits between them, and a senior Nerd/Noir engineering collaborator works hands-on alongside the cohort the whole way. The structure holds constant; the audience, material, intensity, and target problem adapt to your goals and budget.

Who It’s For

We run Effective AI for four kinds of cohorts. Each accommodates up to 20 people.

  • Business and domain experts moving past one-off chats to durable systems built around their real work: reusable projects, custom instructions, and agents that compound in value.
  • Product teams rethinking how they discover, design, and deliver, making AI a first-class part of the product development lifecycle rather than a bolt-on.
  • Engineering cohorts adopting a shared harness and reusable skills, so AI-assisted work is consistent across the team instead of varying by individual.
  • Centers of Excellence and pilot teams evaluating platforms, models, and tooling by building working proofs of concept, leaving with evidence rather than vendor slides.

How It Works

  1. Before kickoff. We work with you to establish the problem direction, so the cohort starts aligned.
  2. Framing. The cohort clarifies the problem, the outcomes worth having, and how it will know it got them.
  3. Working sessions. Focused instruction paired with hands-on building, applied to the cohort’s actual problem.
  4. Build time. Between sessions, participants keep building. This is where the habits form.
  5. Ship and continue. The final session demos progress and sets a plan for growing the practice after we leave.

Format

  • Standard engagement: four 2-hour sessions, delivered two per week for two weeks or one per week for four weeks, with builder time in between
  • Extended engagements and full-time, dojo-style challenges for more ambitious goals

What You’ll Walk Away With

  • A backlog of valuable use cases with working progress on at least one, built to grow and compound. Depending on the cohort, that might be a development harness, reusable skills, a Claude Project, or an agentic workflow.
  • The skills and judgment to keep building: prompt, context, and loop engineering, workflow decomposition, and the practice of turning tacit knowledge into explicit, reusable instructions
  • Working evals for what you built, so it stays honest when the model changes or the instructions drift
  • A read on where humans belong in your AI loops, and a playbook of FinOps practices to better manage costs
  • Access to the knowledge graph our collaborators use at nerdnoir.ai
  • A set of working principles and an action plan for the next use cases to pursue
  • A digital whiteboard with all workshop curriculum, examples, exercises, and templates

Example Sessions

The following is an example of a standard engagement. We customize the program around your cohort, problem, and goals.

Session 1: Framing the Problem

We sharpen our problem focus, determine desired outcomes, and identify the measures that will indicate success.

  • Take stock of the AI infrastructure the cohort already has in place
  • Find the highest-value insertion points for AI in the work the cohort does today
  • Build: stand up a working environment and take a first pass at the problem

Session 2: Engineering Fundamentals

The core technical skills, applied to the cohort’s problem rather than a canned example. The session’s real work is turning tacit knowledge into explicit in the form of reusable instructions.

Session 3: Compounding the System

Durability. Everything the cohort has built so far has to survive and evolve outside the room it was built in.

  • Projects, skills, harnesses, and agents — the forms a setup takes when it has to last
  • Evals that keep the system testable when a new model lands or the instructions change
  • Loop engineering: tool granularity, stopping conditions, and where human judgment stays in the loop so workslop is never the result
  • Measuring value in the terms your finance team uses: time saved, token cost, output quality
  • Build: extend, harden, and prepare to demo

Session 4: Ship and Continue

The cohort demos what it built and why it matters. This is the session that decides whether the practice outlives the engagement.

  • Each team writes its working principles for effective AI in your organization
  • Constructive critique from the Nerd/Noir collaborator on what the cohort has built
  • A clear roadmap of next steps for continuing, refining, and applying learnings
  • Each team prioritizes the use case backlog and plans its next use case

What We Ask of Sponsors

This workshop works when leaders treat it as serious work rather than training. Participants are building something with a number attached: time saved, token cost reduced, an innovation opportunity opened. We ask sponsoring leadership to:

  • Grant participants builder time between sessions, and defend it against competing priorities
  • Clear access to the tools, licenses, and systems the cohort needs
  • Help select a problem worth solving, before kickoff
  • Show up for the demo session
  • Experiential Learning — the dojo model this program runs on, and why doing beats instruction
  • Loop Engineering — the third leg of the curriculum, alongside prompt and context engineering
  • Prompt Frameworks — six named prompt structures with a decision matrix, applied in Session 2
  • Compounding Engineering — building AI setups that get better with use instead of resetting every session
  • Evaluations — how the cohort makes its system falsifiable in Session 3
  • AI Cost-Savings Scorecard — the FinOps side as a self-assessment you can run before we ever talk
  • Value Stream Mapping — for surfacing handoffs and bottlenecks in a recurring workflow
  • AI Without ROI — the problem this program answers
  • mcp-server — connect your AI tools directly to this knowledge collection
  • Jimmy Parker, “Behavior Change at Scale” (Hard Boiled Software, 2026) — the episode behind the training-application number above