Prompt Craft & AI Fluency for Professional Services Firms
Six modules. Every team member leaves knowing how to write prompts that produce reliable output - and with access to a shared library they helped build during the session.
01
How to Think About AI - Not Just Use It
- What LLMs actually do (and don't do) - setting accurate expectations
- The difference between ad hoc use and systematic use
- Why prompt quality is a leverage point, not a nice-to-have
- Common failure modes: vague prompts, wrong tools, no shared system
02
Writing Prompts That Actually Work
- The anatomy of a structured prompt: role, context, task, format, constraints
- Writing for reliable output vs. one-off output
- How to include firm-specific context without overloading the prompt
- Iteration as a skill: how to diagnose and fix a bad response
- Live examples from professional services workflows: research, drafting, synthesis
03
Sourced, Verifiable Output
- Why hallucination happens and how to reduce it
- Techniques for prompting with citations and evidence requirements
- When to trust the output and when to verify
- Structuring prompts for legal, financial, and advisory accuracy standards
04
Custom GPTs and Claude Projects
- What Projects and custom GPTs are and when to use each
- Setting up a Project with firm context baked in: instructions, files, tone
- Building task-specific assistants for high-volume workflows: client summaries, research briefs, proposal drafts
- Hands-on: each participant sets up at least one working Project by end of session
05
Building Your Firm's Prompt Library
- Why a shared library is the difference between individual productivity and firm-level compounding
- What goes in a prompt library: templates, tested prompts, use-case tags
- How to structure it so people actually use it, not just save it
- Governance: who updates it, how prompts get vetted, version control basics
- Hands-on: contribute at least one prompt to the firm library
06
Putting It Into Practice
- Mapping your highest-volume workflows to what they look like with AI in the loop
- Building a personal habit: where AI fits in your weekly work
- How to tell if it's actually working: tracking time saved and output quality
- What's next: bridge to 202 - Skills, Cowork & Data Connections
Delivery Format
- Half-day workshop - 6 modules in a single focused session
- 6 × 60-minute sessions - spread across two to three weeks
- Built for the full firm team - not just power users
- Designed for firms with 10 to 150 people
What You Walk Away With
- Firm Claude Project - set up and loaded with your context before the session ends
- Shared prompt library - seeded during the session, with a clear system for adding to it
- Personal workflow map - each participant identifies where AI fits in their weekly work
Want to run this with your team?
Book 30 minutes and we'll figure out what a 101 engagement looks like for a firm your size.