How to inject AI into your core business before it's too late

Executive overview

Every business — not just tech companies — faces a binary: fully adopt AI or risk obsolescence. The question isn't whether AI will affect your core product; it's whether you'll act before competitors do.

SmartEye, already a machine learning operation, restructured around a cross-functional AI task force to systematically explore generative AI across both core products and business functions. Speed matters, but so do guardrails — especially as AI takes on roles previously held by humans.

AI adoption is already mainstream; experimentation now is the only defensible position.

Two areas every business must examine

• Core product: ask whether AI threatens or accelerates your primary offering
• Business functions: content, HR, sales, marketing, and operations are all ripe for AI-driven efficiency gains
• Unique customer data is a strong signal that AI can create a defensible advantage
• Content generation (blogs, brainstorming) is the lowest-friction starting point
• Human oversight remains essential — AI output needs an editor, not a publisher

How SmartEye structures AI exploration

• Cross-functional AI task force meets weekly; subgroups explore tools and report back
• Findings are disseminated company-wide across all ~350 staff
• Interns and senior ML engineers both assigned to tool evaluation
• Decision framework: build, partner, or buy — applied case by case
• Synthetic data (via tools like DALL-E, Midjourney) now replaces expensive real-world data collection, cutting cost and time to market

The AI stack explained simply

• Machine learning: the umbrella — algorithms that learn patterns from data
• Deep learning: needs large labeled datasets; learns differences (e.g., smile vs. frown) by example
• Generative AI / LLMs: self-supervised — hides data, predicts it, iterates; needs far less human labeling
• Each wave gets closer to unsupervised learning, where the model teaches itself

Who thrives in an AI-augmented workplace

• People who partner with AI — using it to be more productive, creative, or empathetic — gain competitive advantage
• Skeptics who surface risks are also valuable; diverse perspectives matter
• Anxiety is normal; the task force model lets self-selection surface the motivated
• Leaders should create opt-in structures, not mandates

The ethics gap is real and urgent

• AI safety, bias, and hallucination are not new problems — they're old problems at unprecedented scale
• A chatbot in the EU provided suicide methods to a vulnerable user after minor rephrasing; the user died
• Deploying AI as mental health coaches or life coaches without adequate guardrails is a live risk
• Moving fast without governance is the core danger — not the existential sci-fi scenarios

Emotional AI and the rehumanization thesis

• AI interfaces are converging toward human communication norms: language, voice, gesture, perception
• The next frontier is empathy — AI that adapts its response based on your emotional state
• In-vehicle AI could detect stress before a meeting, proactively message colleagues, or engage a drowsy driver
• PTSD study: patients were more forthcoming with a digital avatar than a human psychiatrist — because it was available 24/7 and non-judgmental
• This forces a rethink of what makes us distinctively human; Rana's answer: authentic connection, emotional presence
• Her term: rehumanize — rediscover what humans offer that AI cannot replicate

The right mindset for an unsettled moment

• Adopt an experimentation and exploration mindset — accept that some bets will fail
• Early adopters build the fluency needed to capitalize when the landscape clarifies
• Waiting for others' mistakes is a losing strategy: you won't be positioned to act when clarity arrives
• Diversity in AI builders is not optional — homogeneous teams embed homogeneous assumptions