Applied AI / Forward Deployed Engineer · Forward-Deployed Craft & Behavioral · Lesson 2 of 4
Communicating with customers & non-engineers
Half the job is talking to people who aren't engineers: customer ops leads, executives, domain experts. This is a scored skill in FDE loops.
- Translate: explain technical tradeoffs in business terms ('this cuts response time in half but raises cost ~20%'), not jargon.
- Manage expectations: be honest about what AI can/can't do. Set realistic accuracy expectations and explain that LLMs are probabilistic, not perfect. Overpromising kills trust.
- Discovery: ask about the workflow and the pain, not the feature they think they want. Find the real problem worth solving.
- Demo well: show working software early; a live demo builds more confidence than any slide.
- Handle pushback: when a stakeholder is frustrated or skeptical, listen, acknowledge, show data/a fix, and follow up. Stay calm and credible.
You are a translatorPicture yourself standing between a brilliant, literal, sometimes-wrong model and a busy human who just wants their problem solved. Your value is translating in both directions — and being trusted by both.
◆ Lock it in
- Translate tradeoffs into business terms; manage expectations honestly (LLMs are probabilistic).
- Do discovery on the workflow/pain, not the requested feature; demo working software early.
- Handle skepticism with listen → acknowledge → show → follow up.
Feynman drill — say it out loudAn exec asks 'why can't the AI just be 100% accurate?' Answer in plain, trust-building language.
Step 1 rate your confidence · Step 2 pick your answer
A non-technical customer expects the AI to be right 100% of the time. Best response?
Step 1 — how sure are you?