In this episode of UNPAK3D, we sit down with Waseem Alshikh, co-founder and CTO of Writer, to unpack the reality of deploying Generative AI in the enterprise. While the consumer market chases the latest AGI hype, Waseem explains why the enterprise roadmap requires a different approach: predictability, domain specificity, and a shift from simple content generation to executing complex actions.
We dive deep into Writer's "full-stack" strategy—owning the LLM family (Palmera), the graph-based context layer, and the application layer—and why this architecture is critical for security and accuracy in regulated industries like healthcare and finance. Waseem also shares his contrarian take on "small" models, the necessity of change management to unlock real ROI, and why he defines AGI not as General Intelligence, but as "Artificial Governance Intelligence".
00:15 — Introductions & Writer’s core mission
03:25 — The enterprise AI gap: predictability vs. hype
07:40 — Domain-specific models and why they outperform general models
12:55 — Synthetic data, grounding, and how Writer trains models
18:10 — Building the enterprise context layer (knowledge graphs)
22:45 — On-prem, small models & deployment realities
28:15 — How enterprises are shifting their AI buying behavior
33:30 — Competing with OpenAI & managing roadmap pressure
39:35 — Multimodality and why it’s emerging inside the enterprise
45:25 — Unit economics, ROI & end-to-end task automation
51:15 — Competing for talent + Waseem’s view on AGI
55:55 — Decompression, customer stories & closing thoughts
