WRITER

New pod drop on UNPAK3D Podcast 🎧 The enterprise AI race isn’t about the model anymore. It’s about who owns the context. Me and Vignesh Ravikumar had my friend Waseem Alshikh (Co-Founder/CTO, WRITER) on UNPAK3D Podcast to go deep on defensibility in enterprise AI. Writer’s winning by doing the opposite of the hype: stability over shiny. In consumer, we want the newest toy. In enterprise, that “toy” breaks production. One major healthcare client is still on 2.5-year-old Palmyra 3 because it’s stable, cheap, and predictable. Enterprises don’t want a model that changes personality overnight. Here’s what actually matters: The context moat Models commoditize. Synthetic data is solvable. The real moat is your enterprise context—knowledge graph, policies, governance, org structure. Writer’s bet: own that context layer and make it irreplaceable. The ROI reset AI augmentation = expensive compute vs. cheap labor. Bad math. AI automation = one advisor supervising agents across 100+ accounts, not 2. That’s where 150%+ productivity and real unit economics show up. Humans become supervisors, not doers. The verticalization play Not horizontal. Deep domain models (Palmyra Fin, Palmyra Med) trained on industry DNA. Their 70–80B parameter models beat bigger general models on grounding because domain knowledge kills hallucinations. What enterprises actually want: ❌ Latest and greatest ✅ Predictable performance ✅ 1–2 year roadmap clarity ✅ On-prem deployment ✅ Zero model drift Talent angle that stuck with me: in a market with insane AI comp, Waseem competes on vision and legacy. His decompression is reading customer wins—like disability claim reviews going from weeks to under 24 hours. My TAKE >>> The stack is inverting: Model as moat → Context as moat Augmentation → Automation Horizontal → Vertical Hype cycles → Time-to-value The winners will be full-stack platforms that own workflow, models, knowledge graphs, deployment—and prove ROI via process transformation, not minor productivity boosts. Writer’s doing that by being boring, predictable, and enterprise-first in a market obsessed with flashy. Question for founders/operators: when you evaluate AI vendors, are you still chasing the newest model—or prioritizing stability, control, and hard ROI? Full episode dropping soon on UNPAK3D Podcast. Geodesic Capital #ai #contextengineering #fullstack #ROI #AIfounders #founders #value #enterpriseAI #scaling #resilientAI Spotify: https://lnkd.in/gYWP-JS3 Apple: https://lnkd.in/gDD2YsyR Substack: https://lnkd.in/gwBwqwgm

What if your LLMs could plan, reason, and specialize without sacrificing efficiency? Next Tuesday, December 2nd, I'll be at NeurIPS ’25 to discuss WRITER’s research on developing single deployed models that can shift into domain-specific modes. I'll share a sneak peek of our new “Enterprise Brain” research and explore the future of the next generation of LLM’s. Looking forward to connecting with you in San Diego!

After billions in spending, only 1 in 3 companies is seeing real returns. The experimentation phase is over – it's time to prove value. The good news? Some enterprises are crushing it: 333% ROI | $12M value | 6-month payback Our latest newsletter dives into the Forrester TEI study and reveals: ✨ What separates winning AI initiatives from the 40% that get scrapped ✨ How to build AI agents without developers ✨ American Eagle Outfitters Inc.'s escape from novelty to results ✨ Governance frameworks that accelerate deployment Read the full breakdown. 👇

WRITER is heading to #AWSreInvent 2025! 🌟 Join us at Booth 1770 to see how enterprises are rethinking AI governance in the agentic era. Catch: ✔️ Live platform demos showing how to empower business teams to build while giving IT visibility and control ✔️ Real customer stories of democratized AI with centralized oversight ✔️ Swag and tactical conversations about scaling AI confidently Plus, don't miss CEO May Habib on the keynote stage on Dec 2nd announcing what's next for WRITER + Amazon Web Services (AWS). See you in Vegas! https://lnkd.in/dZxfurg7

“If we don’t self-disrupt, someone else is gonna do it for us.” That's the mindset that drove Chris Meads, CCO of VOIS, and Alexandra Pott, Consumer Digital Director at Vodafone UK, to lead one of the boldest AI transformations in enterprise. The result? → 50% less copywriting time → 2x content performance → A full day per week saved for every person on the team In a recent session, these two AI transformation leaders sat down with Diego Lomanto, CMO at WRITER to reveal how they moved past the hype to drive real business value — and why waiting for the "perfect" AI plan is the biggest risk of all 👉 https://lnkd.in/dvszyb8y

Here's the thing about giving an LLM access to 200+ tools: it starts forgetting which one to use. That's context rot in action — when too much information in a prompt actually degrades reasoning. Functions blur together. Even advanced models struggle to pick the right tool. Our Developer Advocate, Ashley Weaver published a deep dive on RAG-MCP, a retrieval framework that solves this by selecting, compressing, and isolating only the most relevant tools before reasoning. The approach more than triples tool-selection accuracy while cutting prompt tokens by 50%+. Full breakdown + working prototype 👇

Consumer-grade AI doesn't work for the Fortune 500. Why? Because enterprise AI isn't a plug-and-play solution. It requires a nuanced understanding of existing workflows, processes, and organizational culture. Our Chief Strategy Officer Kevin Chung sat down with Marcelo De Santis to discuss how WRITER takes a collaborative approach to scaling AI in the world's top enterprises. Catch the full episode of Executive House’s The Tech Series podcast to hear how the top leaders are scaling agentic AI for real efficiency and speed (linked in the comments 🔗).

"We used to call it transformation when we took 12 steps and made them nine. That's optimizing the world as it is. We can now create a new world." 👆 Our CEO May Habib on the opportunities the new Writer Agent makes possible for every enterprise employee — and why we built AI that doesn't just assist, but executes. Michael Nuñez breaks it all down in this piece by VentureBeat: https://lnkd.in/gdYVv3yD

Most “enterprise AI” tools are still built like consumer apps with an enterprise price tag. They optimize for personal productivity. Single-shot generations. Isolated tasks. Individual convenience. But enterprises don't run on convenience. They run on repeatable processes, governed workflows, and systems that scale across thousands of people. So over the last few months, we rebuilt the agent from first principles. Today, we're launching WRITER Agent – a new design pattern for how AI should operate in the enterprise. It democratizes AI agent creation so teams closest to the work can build and deploy the workflows they run every day, freeing up engineers to focus on the infrastructure that makes it all secure, scalable, and governed. Here's what's different:  Playbooks = Your recurring workflows, turned into natural-language templates. The BDR running 50 sequences a week? They can now codify their best process and scale it across the team. The data analyst? They can turn their reporting flow into a reusable pattern. This is how tribal knowledge becomes systematized.  Routines = Those playbooks, running automatically. Daily, weekly, monthly – whatever you need.  Connectors = Real context from your calendar, CRM, data warehouse. We built our own MCP gateway for this – governed, auditable, enterprise-ready. WRITER Agent actually carries work forward across systems.  Personality profiles = Every output sounds like you, not a chatbot. When you're producing thousands of deliverables a week at scale, brand compliance isn't optional.  Enterprise controls = Govern web, data, and system access. The pattern is clear: AI won't just assist knowledge work. It will operate it. And for that to happen at enterprise scale, we need the people closest to the work building their agentic workflows. WRITER Agent is available today. The future of enterprise AI isn't about smarter prompts – it's about better tools in the right hands. Link to the full story in comments. Would love to hear how you're scaling AI in your org.