How many people are using AI at work?

Hiya 👋

In this issue:

• What most U.S. adults are (and aren’t) using AI for

• A legal tech app that is a good example of being “AI-first”

• Forrester breaks down where to keep humans in the loop

Let’s get into it 👇

A big poll of American’s across backgrounds shows a clearer sense of where adoption stands, and how far we are from deep, effective use in everyday life and work.

Here’s what stood out to me:

• 60% of U.S. adults say they use AI to search for information

• Only 37% use AI for work-related tasks

• 40% say they’ve used it to generate ideas

• The most active age group? Adults under 30, at 74% usage overall!

We’re in the awareness era, not the proficiency era. People are using AI like they used Excel in 2002: everyone says they use it deeply, but few know about Pivot tables and macros.

🧠 Why it matters:

We're witnessing a new kind of digital divide, not access, but fluency.

Even among younger adults (under 30), where 62% say they’ve used AI to brainstorm ideas, only a minority are using it consistently or for more complex outputs.

Most people now have access to AI. What they don’t have is:

• Clear guidance on when to use it

• Confidence in its reliability

• Real integration into their workday

But with the launch of ChatGPT Study Mode, the productization of AI is showing how quickly a new tech paradigm is being adapted into people’s daily lives.

👉 Go deeper: Read the full AP-NORC article here

We all see “legacy” SaaS talking about being “AI-first” but they’re reorienting their products around AI, not the other way around.

Companies are trying to think about this too. In a world where only humans are doing things, how do we adapt processes & data when they don’t need to do it all?

I think Harvey is a good software paradigm to look at in what “AI-first means”.

First off, some context: we’ve been working with a few legal teams recently, and the one app that’s getting both General Counsel and paralegals genuinely excited?

👉 Harvey

🤔 What makes it different?

• They built custom AI models, not just customized prompts - Harvey partnered directly with OpenAI to train models on legal case law from scratch. While most "legal AI" tools are basically ChatGPT with fancy prompting, Harvey actually rebuilt the underlying intelligence for legal work.

• It thinks like a law firm, not like a chatbot - Instead of one big model trying to do everything, Harvey chains together specialized models for different tasks (one for clause extraction, another for research, etc.) - just like how a partner would distribute work to specialists.

• It’s built for workflows where humans don’t have to do everything - Harvey gets trained on your firm's actual work product, templates, and past cases. It's not just learning generic legal knowledge - it's learning how your firm operates.

• Built for security in an AI world - Enterprise-grade security from day one with zero data training policies, but also factoring in their business being built around AI not letting the wrong people see things they shouldn’t just because they asked a chat bot.

✨ Why this matters

Harvey didn't take existing legal software and sprinkle AI on top - they asked "If we could rebuild legal work from scratch with AI at the center, what would that look like?" That's the difference between being AI-enhanced and AI-first.

For your business, the real opportunity isn't "how do we add AI to what we already do?" It's:

• What processes could disappear entirely if AI handled them?

• What expertise could you scale across your whole organization?

• What would your team focus on if the routine stuff just... happened?

🔗 Go deeper: Gabe Pereyra’s take on why Harvey is the real deal →

A new Forrester report lays it out clearly:

The most successful AI strategies aren’t tech-led. They’re people-led.

That means making AI easier for employees to trust, control, and critique, not just automating around them.

🧐 What the research found:

• Companies with high-performing AI programs embed human oversight

• AI success is tied to clear communication and training, not just deployment

• Trust grows when employees feel like they’re co-piloting AI, not being replaced by it

“This isn’t a technology deployment. It’s an experience transformation.”
- Forrester, Ground Your Workforce AI Strategy In Human Experience

🛠 What we’re seeing too

In practice, it looks like this:

• A task that took 10 steps now takes 3

• But: a human still checks the AI’s output

• That check isn’t inefficiency, it’s QA and buy-in

Teams that keep people in the loop are more likely to:

• Spot errors early

• Maintain trust in outputs

• Help employees feel essential, not left behind

✅ Key takeaway

The best AI automations don’t eliminate people, they reframe their role.

Call it AI-assisted QA. Call it human-in-the-loop. Just don’t call it “fully autonomous.”

“AI succeeds when the humans using it feel like they still matter.”

👉 Read the full Forrester report: Ground Your Workforce AI Strategy In Human Experience

We’re growing at Switchboard for two roles that quietly run the show behind the scenes, and we figured you might know someone.

🔧 Technical Project Manager

The person who can turn a messy roadmap into clean execution.
See the full role

🧩 Business Systems Strategist

The person who sees a 50-step process and says “this could be 5.”
See the full role

We’re remote, collaborative, fast-moving, and building tech & systems that ship.

Know a fit? Nudge them. We'd love to hear from them.

Know someone who’d enjoy this newsletter?
Forward it along, or send us a note if there’s a topic you’d like us to unpack next.