In the Zone | Our Investment in NaceAI

Our experience suggests that enterprise workflows demand structure, reliability, and traceability. These workflows consist of interconnected micro-decisions that require domain-specific expertise, adherence to regulatory frameworks, and alignment with internal policies. This is why we’ve maintained a measured perspective on the use of large, generalized models for end-to-end automation.

Over the past year, we’ve seen many Fortune 500 companies struggle to move AI initiatives beyond proof-of-concept stages. The primary obstacles we’ve identified include lack of clear ROI, complexity of integration, and lack of explainability/auditability. These challenges stem from a fundamental mismatch: most current AI models weren’t designed for complex business process automation. 

The Opportunity We See

We believe the future of enterprise AI lies not in adapting business processes to general-purpose models, but in creating many specialized, purpose-built models that can be combined to execute complex business workflows. 

In our search for companies addressing this need, we looked for teams building infrastructure that could enable small, highly-specialized AI models—models that are: 

• Designed specifically for the business processes they serve 
• More cost-effective 
• Faster to deploy 
• Inherently safer through their limited scope 

We envision forward-thinking companies managing large portfolios of fine-tuned, business-specific models/agents that can work together to deliver meaningful business outcomes and clear ROI. 

Why NaceAI?

NaceAI is building the infrastructure that will power this vision of the future. Its breakthrough is the MetaModel—an architecture that generates custom small models in minutes, each tightly tuned to a narrow task and business context. Non-technical users can create specialized, highly performant agents in minutes by simply describing key tasks and connecting relevant data sources. 

This isn’t theoretical. NAVI, its first product, is already live with leading enterprises who use MetaModel to generate lightweight agents that support audit, compliance, and operational intelligence functions. Mountain America Credit Union, for example, is using NAVI to automatically review credit loan applications, cross-check them against policy and regulatory requirements, and deliver explainable decisions. 

What Distinguishes NaceAI

Three factors particularly impressed us:  

• Technical Architecture: NaceAI leverages an ensemble of over thousands of specialized models and a proprietary meta-learning engine. This approach allows them to generate actual model weights—not just RAGs—that reflect the specifics of business tasks and knowledge bases.  
• Lifecycle Management: Generated models continuously evolve through continuous learning with MetaModels, feedback loops, and business telemetry. NaceAI provides ongoing management throughout the model lifecycle.  
• Team Expertise: We first met Dos Baha at NeurIPS years ago and were immediately struck by his clarity of vision for an agent-enabled future. Long before “agents” became a buzzword, Dos understood where the puck was going. He’s since pulled together an exceptional founding team: Sudha Valluru, a three-time founder and former enterprise operator from Google, Cisco and Apple; Zhanibek Datbayev, the former AI infra lead at Meta; and Amine Mohamed Aboussalah, a professor with deep research chops in meta-learning and generative models. Together, they bring both scientific depth and enterprise pragmatism. 

Our Investment

We’re pleased to participate in NaceAI’s seed round. Their approach resonates with our thesis that the most valuable enterprise AI will not be a single, all-encompassing model but rather networks of highly specialized agents working together. NaceAI makes it feasible for every enterprise, not just the tech elite, to create small, intelligent models that custom fit their business use cases like audit and compliance.

NaceAI is developing what we see as a crucial new layer in the AI stack—one that has the potential to democratize AI adoption across enterprises of all sizes and technical capabilities as companies look to build out many custom agents. 

We look forward to supporting the NaceAI team as they continue to develop this vision of purpose-built, enterprise-grade AI agents. 

In the Founder’s Words, from Dos Baha, NaceAI’s Co-Founder & CEO: