Annie Tsai

There’s a shift happening in how work gets done. This week, we are going to step away from the very real job-loss narrative. The story arc I’d like to explore with you all today is this question of who is getting amplified in this new era of AI-everything.

For the last two and a half years at Interact, I’ve lived inside this shift. My job has increasingly become less about orchestrating and doing work and more about designing and building systems that do the work with and for our teams. The old question was “How much can one person handle?” – a classic capacity modeling challenge. The new question today and for the foreseeable future is now this:

“How much output can one person deliver when paired with reasoning systems that scale infinitely?”

This is the leverage mindset I’ve been driving home in the “Practical AI” and “AI Agent” workshops I’ve been running over the past few years. Heading into 2026, I don’t think anyone can ignore this anymore as it is very bluntly the difference between being left behind and having your contributions be wildly understood and celebrated.

AI Agents As Co-Workers, Not Tools

Most people still think of AI as automation. Push button, get output. The difference with agents is that simple automation follows instructions where reasoning agents deliver outcomes.

A real agent has a goal, determines best steps, decides when to retry, and knows when something looks wrong. It operates much more like a junior team member than like a macro.

I see this inside Interact every day. Our moderation pipelines. Our attribution clean-up logic. Our enrichment engines. Our workflow decisioning models. None of these are single actions. I designed these systems to make decisions, monitor themselves, escalate only when necessary, and deliver consistent outcomes.

This work isn’t happening in a vacuum. BCG reports that agentic systems are now reducing low-value work by 25–40 percent and accelerating core business processes by 30–50 percent, with early adopters seeing large improvements in service quality, risk reduction, and customer conversion.

This is why the conversation has to move beyond “AI as tool.” We’re onboarding co-workers and teams that infinitely scale.

The Macro Backdrop: The Money Has Already Moved

If you want to know where the economy is headed, follow the capital flows.

Gartner projects that in 2026 global IT spending is set to surpass $6.08 trillion, with nearly 20 percent growth in data-center systems, almost entirely driven by AI infrastructure build-out.

They’re building the operating substrate for a world in which agentic systems are standard requirements to exponentially scale enterprise productivity and outcomes.

Applying a Leverage Mindset

This is the part that operators misunderstand most often. The leverage mindset is not “apply AI everywhere.” It’s:

1. Start with outcomes, not tasks
2. Map the reasoning chain including appropriate human-in-the-loop escalations and gate checks
3. Expose the data and permissions
4. Assign the agent
5. Iterate like you would with a new hire

This process is how we’ve grown internal systems: process by process, with intention. What started out mostly in Zapier has grown up to be full agentic systems built in python and living in Google Cloud. It’s a journey that I think many are in the very early innings of but nearly every company will dip into over the next 18 months.

BigCo Consulting agrees. McKinsey reports that AI in distribution/logistics reduces inventory 20–30%, cuts logistics costs 5–20%, and reduces procurement spend 5–15% — entirely through well-structured outcome-owned workflows.

PwC’s 2025 manufacturing and operations analysis shows nearly 70% of leaders expect at least a three-point profit increase from AI — but only 8% capture full value, because system-level redesign is the real unlock.

Customer Experience: The Sharpest Expression of Leverage

As my career is rooted in customer experience, I think personalization is one of the easiest, most measurable and most impactful leverage point where organizations can start.

McKinsey finds that companies who excel at personalization see 10–15% revenue lift (with some reaching 25%) and generate 40% more revenue from personalization than their peers.

BCG shows that personalized recommendations increase conversion and cross-sell rates by 30–40%.

Companies have for years done the difficult work of bringing disparate systems and datasets together to ingest relevant information and build workable personalization engines. And truthfully, for most companies it is still more dream than reality.

Today however, agentic systems either built in-house by your resident tinkerer like what I did or paying a subscription to access some form of an agentic system that does this for you is entirely possible and incredibly effective – take your pick but the leverage is ready and waiting.

Close: Audit Your Leverage Point

Here’s your practical starting point:

Identify one outcome in your business that — if scaled with AI agent — would materially change your workflow, speed, and cost profile.

Not a task. An outcome. That is where your next AI co-worker should be hired.

If you need frameworks, here are the two strongest available today:

• CIGen’s weighted scoring model for prioritizing AI use cases — scoring business value, feasibility, data readiness, risk, and time-to-value – is really interesting.

• OpenAI’s Impact/Effort matrix for identifying quick wins and coordinating deeper systems transformation.

Operations is becoming the discipline of designing leverage. The sooner you begin, the greater your advantage.