Navigating the AI Era in EdTech: Strategic Insights from Times Square

Standing in the middle of Times Square, with screens, sound, motion, it’s impossible not to think about information overload. That’s where I think AI in education is today. There’s constant buzz, real opportunity, and a need for clarity.

EdTech leaders are asking better questions. Not just what AI can do, but where to use it, where to draw a line, and what success looks like quarter to quarter.

In a recent episode of EdTech on the Street, we dove into some of the most urgent conversations. Education platforms, publishers, and workforce providers need a clear way to use AI to move faster while keeping the human context that education depends on. At Magic, we’ve tested, broken, rebuilt, and documented much of this. Here’s how we operate today, and what we believe others can take, apply, and make their own.

1. Who Edtech Builds for vs. Who We Market To

My stance is simple: don’t target students in marketing. In most of EdTech, students aren’t the buyers; your audience is teachers, parents, schools, or publishers. Design for learners, but market to the actual decision‑makers. That’s both a practical and ethical line. Before strategy, before architecture, before prompts, start with usage rules that keep your teams and your learners safe. This has nothing to do with chasing the next model. It’s just about knowing what you will and won’t allow.

 What’s Actionable Now

• Keep learner experience first in product decisions.
• Keep student data out of growth experiments.
• Route marketing toward institutional buyers and guardians.

Example of a Tactic Removed Because It’s Aimed at Students

• Removed: “Back-to-school” Instagram ads aimed at class hashtags used by high schoolers.
• Reason: Indirect micro-targeting of minors.
• Replaced with: Teacher resources landing page promoted to verified educator communities and district newsletters.
• Owner: Social Marketing Lead, reviewed by Privacy Counsel.

 2. Build a Team That Knows What Not to Do

Yes, there are laws (our SaaS platform is compliant with U.S. regulations), but the real guidance is a moral compass. Here are my two rules:

1. Respect IP & paywalls. Don’t use material “behind paywalls” as AI inputs.

2. Don’t paste proprietary data into open LLMs. In the “heat of the moment,” guard against sharing anything sensitive.

How to Train AI-Literate Teams

• Employees go through a 60-minute “safe prompting” module.
• Tool-specific training is provided only for approved apps.
• SMEs are trained to review AI-generated scaffolds and not just approve them.
• Prompt logging is redacted at source and monitored monthly.

3. AI and Marketing: Speed up the Making, Keep Humans on the Message

We use AI heavily in marketing, but not to write our content. Don’t use AI to author thought leadership. Content with context is something only humans can do. We’ve automated video editing, design resizing, transcript summarization, research compilation, and variant generation. What we have not automated is voice, tone, positioning, and context. Those stay human. It’s a clear line we draw at Magic.

How We Manage Production

• A co-pilot policy defines what AI is allowed to assist with.
• Human review is a mandatory step for any public asset.
• Each output includes a note on whether and how AI was used.

For every task, we track baseline cycle time and AI-assisted time. If it saves time and passes review, it stays. If it fails to review more than once, it’s out.

4. Jobs and Skills: The Refrigeration Moment

AI will eliminate some roles and create others, just like the internet did. In my conversations with edtech leaders, I like to use the “refrigeration” analogy: the tech exists; now we have to invent the products. In the near term, we expect more reviews by experts, not fewer. If AI can draft 80% of code or content, a human expert still has to verify it’s contextual and correct.

What’s Actionable Now

• Encourage everyone to “use AI” to remove administrative load, but keep human review as a hard rule.
• Hire/enable reviewers and editors who can validate AI‑assisted work.
• Build a team culture around curiosity and critical thinking.

5. Cost Savings Now, Revenue Later

Two years into the wave, my prediction is that AI will shift from buzz to more realism. Cost efficiencies are already showing up (faster GTM, less mundane work). Top‑line lift will follow after better product thinking and adoption, a few quarters down the line. We’re still in the middle of that journey.

What’s Actionable Now

• Use AI to speed go‑to‑market and reduce operational toil immediately.
• Keep the revenue thesis grounded in real usage and outcomes. Resist vanity launches.

Three Decisions Edtech Leaders Can Make This Quarter

AI is here to take the mundane off our plates and free us to do more meaningful work. If we keep learners at the center, market to the right buyers, and run with a moral compass, we’ll ship faster without losing the human touch that education requires.

My advice to any professional in EdTech today: stay curious. The people who thrive won’t be those who resist change, but those who understand how to use AI as an amplifier for their own ideas and expertise.

So whether you’re an executive, an educator, or a developer, remember: this isn’t about mastering the machine. It’s about using it responsibly to build better human outcomes.

Filmed live in Times Square, NYC. Part of the “EdTech on the Street: Real Talk in the City That Never Sleeps” video series by Magic EdTech.