Jul 20, 2026, 1:59 PM ETFireside Chat, Interview

CJ Desai, MongoDB & Laura Modiano, OpenAI in Conversation | RAISE Summit 2026

MongoDB, founded in 2007, operates as a cloud-native, "run anywhere" database serving 67,500 customers ranging from AI-native startups (e.g., Eleven Labs) to large enterprises (e.g., JPMorgan Chase).
The company generates $2.5 billion in revenue and positions itself as the "unsung hero" of AI, arguing that the data layer must scale in tandem with AI models to prevent application bottlenecks.
MongoDB distinguishes itself as the only hybrid multi-cloud database capable of running on-premises, in public clouds, or across multiple environments simultaneously to address data sovereignty concerns (e.g., French regulations).

Founder Pitfall: CJ identifies a common failure mode where founders prioritize application logic and model selection but neglect to select a scalable data layer early, leading to infrastructure choking as applications grow.
Architectural Flexibility: MongoDB advocates for schemaless (NoSQL) architectures to accommodate rapid model evolution and agent demands, contrasting with rigid infrastructure that slows innovation.
Real-Time Transactional Priority: The company asserts that real-time transactional databases (OLTP) constitute the "high ground" for agentic applications, rather than purely analytical databases.
Vector Integration: MongoDB's key differentiator is the unification of auto-embeddings, vector database capabilities, and operational data within a single layer, avoiding the complexity of managing separate vector stores.
Innovation Velocity: AI model iteration cycles have compressed from 15 months to six weeks (referencing the transition from version 5.5 to upcoming 5.6), necessitating a data layer that can keep pace with continuous changes.

OpenAI's Codex desktop application, launched in February, has reached over 5 million weekly active users in a few months.
Adoption rates for non-technical users are three times faster than for technical users, expanding value delivery to sales, marketing, and finance teams.
Codex is enabling "vibe coding," allowing builders to create complex applications (e.g., logistics or CRM) via natural language prompts, reducing the need for rigid infrastructure.

A specific customer framework for becoming "AI-forward" enterprises rests on four pillars:
- Innovation Speed: Can AI tools (like Codex) shift the product roadmap left to innovate faster?
- Revenue Growth: Can AI be leveraged to sell more effectively?
- Service Quality: Can AI enable proactive diagnostics and better customer success?
- Operational Efficiency: Can AI reduce internal costs and improve overall efficiency?
Enterprises are moving beyond the "early innings" of AI prototyping toward the "2026 phase" of deploying production-grade agentic architectures at scale.

CJ employs a "70-30 rule" in customer interactions, dedicating 70% of time to listening to solve-for-challenges (e.g., talent gaps requiring forward-deployed engineers) and 30% to talking.
The company prioritizes "conviction over optimization," advising founders to aim for 10x improvements based on customer possibilities rather than incremental 10% gains based on existing data.
MongoDB provides forward-deployed engineers to assist customers in bridging capability gaps, evidenced by immediate responses to capacity issues (e.g., a prediction market CEO in Paris).

The transition from "prompting to production" is identified as the critical strategic phase for 2026 (accelerated from a previous 2025 estimate) to solve the challenge of building production systems at scale.
The industry trend suggests a move away from "prototype treadmills" where tools remain in development, toward fully realized agentic architectures supported by robust data layers.