General-purpose AI can solve most problems.
But in industries like aerospace, semiconductors, manufacturing, and energy, "most" isn't enough.
In this episode of AI in Action, we sit down with Arun Subramaniyan, Founder & CEO of Articul8, the enterprise AI company spun out of Intel, to discuss why domain-specific AI could define the next phase of artificial intelligence.
We explore why general LLMs hit a ceiling, why domain models don't have to be small, how enterprises chase "five nines" reliability instead of impressive demos, and why Arun believes the real AI winners are solving difficult industry problems while many others are simply selling a dream.
We also dive into Articul8's work on India's heritage languages, where Sanskrit and Tamil are treated as technical domains to unlock knowledge hidden in centuries of literature.
In this episode:
• Why general AI gets you only 70 to 90% of the way
• Why domain-specific AI is fundamentally different from general LLMs
• The misconception that domain models are "small"
• Why enterprises need near-perfect accuracy
• The future of AI in aerospace, manufacturing, semiconductors, and energy
• Why Arun thinks many AI startups won't survive
• Articul8's vision for India's heritage AI models
• Lessons from building an enterprise AI company after Intel
