JB Training Solutions is a corporate training company. Their trainers travel to client organizations to deliver workshops like Coaching Fundamentals and Leading to Greatness, each supported by a facilitator's PowerPoint deck.
At the end of every workshop, participants take home a post-training worksheet: a one-page, branded document with reflection and action questions tied to that specific session. It's designed to be revisited over the following weeks, helping people apply what they learned. Every one of those worksheets has to feel specific to the workshop it comes from, not a generic set of prompts recycled across programs. With 44 distinct programs in the catalog, that meant building a new worksheet from scratch for nearly every session.
When a Growing Content Library Becomes a Bottleneck
Building that worksheet meant pulling from the deck's structure, applying the internal template, and manually balancing actionable and reflective questions. One employee, one deck at a time, over 3 hours per worksheet.
The opportunity cost added up. Dozens of programs sat ready to teach, yet turning each one into a participant-ready worksheet meant repeating that same manual process every time. They had already built a valuable library of training content. The bottleneck was turning it into something participants could actually use.
How One Working Session Was All It Took to Start
The team came to SoftSnow with a direct goal: get more value out of what they'd already built.
SoftSnow's team started by asking for real examples: a few trainer decks paired with the worksheets JB's own staff had built from them. Those examples showed exactly how a skilled employee made decisions at each step of the process. From there, the team could identify where AI could reliably take on that work.
"Because they had examples of a few different decks and the corresponding checklists they'd made, we were able to work our way backwards and figure out the process." — Trevor Miller, Contractor Engineer at SoftSnow
How Claude Learned to Replicate a Manual Process, and Improve It
Using Claude Opus, SoftSnow's team reverse-engineered the client's manual process, mapping what a skilled employee did when building a worksheet from scratch. Claude then refined its own output until it could reliably match what had been produced by hand
The resulting skill lives in a shared Claude plugin marketplace. Point it at any trainer deck, and it returns a branded, ready-to-use worksheet. Every update SoftSnow makes rolls out automatically, so no one has to reinstall or reconfigure anything.
Three principles shaped the build:
- No hallucination. Missing or unclear content gets flagged for human review instead of being invented.
- Branded output. Every worksheet matches client’s template, with the logo anchored to prevent layout drift.
- One-page constraint. If content risks overflowing, the skill asks the user what to cut before rendering.
From an Hour of Manual Work to Minutes, at Scale
"What used to take hours we can do in just a couple of minutes." — Courtney Templin, President, JB Training Solutions
The post-training worksheet skill gave JB Training Solutions a direct path from trainer deck to participant-ready document, with every program in their library now a deployable asset in minutes instead of hours.
That relationship doesn't end at launch. AI models and capabilities continue to change quickly, and SoftSnow actively maintains and updates the skill so it keeps delivering reliable results as that landscape shifts.
For B2B service companies with deep content libraries, that kind of leverage adds up quickly. It's a small example of a pattern we see across every AI transformation we lead: change starts with one workflow, for one person, before it reaches the rest of the organization.
Frequently Asked Questions
- Why does JB need a new worksheet for every workshop instead of one template?
Each program covers different frameworks and skills, so the reflection questions have to match that specific content. A generic worksheet can't do that; the skill reads each new deck and builds questions from what was actually taught.
- What makes a custom Claude skill different from a generic AI tool?
A custom skill is built around one team's process, template, and guardrails, in this case the worksheet format, question style, and brand standards specific to the client, rather than a general-purpose prompt.
- Does the Claude skill need to be maintained after launch?
Yes. AI models and capabilities change quickly, so SoftSnow actively maintains and updates each skill after launch to make sure it keeps performing reliably.
Ready to turn your own content library into a scalable asset? Let's talk about what that could look like for your team.
