FDE ForgeBeta
Start free

0:00 / 3:00

Guess first

They asked for everything. Ship one thing.

Northwind wants an AI assistant for risk, compliance, support and HR. You have six weeks. Which one do you build first, and how will you know it worked?

Wish list4 departmentsCostliest task8-hour reportMVVUnder 15 minutes
Lesson notesMVV, not MVP: the short version+

Week 1 · FDE Mindset & Toolkit · about 3 minutes

What you will be able to do

Scope the smallest integration that proves money saved, with a number agreed before coding.

The idea

An MVP asks: what is the smallest set of features? A Minimum Viable Value asks: what is the smallest integration that proves money saved? (MVV is our handbook's name for it; customers and interviewers usually say "success metric" or "business case".)

Find it in three steps. Pick the manual task that costs the client the most. Take only the data that task needs, often three tables, not the whole warehouse. Agree the success number before you write code.

A signed-off number turns scope creep into one simple question: does this request move the number?

# Minimum Viable Value: Northwind Bank
Task:     Daily credit-risk report, built by hand
Cost:     8 analyst-hours a day
Data:     positions, limits, counterparties (3 tables)
Success:  report ready in under 15 minutes, every day
Owner:    Head of Credit Risk
Not now:  compliance, support and HR assistants

If you can't put a number on it, it isn't the MVV yet.

Key takeaways

  • Start from the most expensive manual task.
  • Ingest only the data that task needs.
  • Agree the success number before writing code.

Part of the Forward Deployed Engineer roadmap. New to the role? Start with what a Forward Deployed Engineer does.