What Makes an Associate "AI-Fluent"? Inside the Taskforce Academy

"AI-fluent" gets thrown around a lot right now, usually without much behind it. I run talent acquisition and management at Taskforce, which means I'm the person responsible for making sure that when we say an associate is AI-fluent, it means something specific and verifiable — not a line item on a résumé nobody checked.

It starts before day one

AI training isn't something we bolt on after an associate is placed with a client. It's part of the Taskforce Academy, which every associate completes before they're ever matched to an engagement. That sequencing matters: we're not testing AI skills on your time, at your expense, while you find out what someone can and can't actually do.

The three things we actually train for

1. Tool-specific strengths, not generic prompting

Knowing how to write a reasonable prompt is table stakes. What we train for is knowing which platform to reach for — Claude, ChatGPT, or Gemini — for a given task, based on real differences in how each handles long documents, structured research, or drafting under specific constraints. An associate who defaults to one tool for everything is an associate who hasn't been trained properly.

2. Verification, not blind trust

AI output is only useful if someone knows when to double-check it. We train associates to treat AI-generated research, summaries, and drafts as a strong first pass that gets verified before it reaches you — not a finished product. This is the single biggest gap we see in self-taught "AI skills," and it's the one most likely to actually cost you something if it's missing.

3. Folding AI-assisted work into professional output

There's a specific, recognizable texture to work that's obviously AI-generated and lightly edited. Part of training is closing that gap — an associate should be able to use AI tools heavily in their process while producing a final deliverable that reads like it came from a skilled professional, because it did, with AI as part of their toolkit rather than a replacement for their judgment.

How we know it's working

Training that isn't tested is just a slide deck. Associates are assessed on realistic scenarios — drafting under time pressure, synthesizing research from messy sources, handling a request that requires picking the right tool rather than the familiar one — before they're considered ready for placement. It's the same principle behind the rest of our screening process: we'd rather catch a gap internally than have a client catch it first.

Why this is where we're investing

The floor for "capable administrative support" has moved. Two years from now, AI fluency across multiple platforms won't be a differentiator — it'll be assumed, the way basic English fluency and spreadsheet literacy are assumed today. We'd rather be early to that standard than catching up to it, and that's the honest reason the Academy looks the way it does.