This interview was excellent. I enjoyed the shared vision from @shishir and Rahul regarding the latest business combination and the technical implications of a world where Grammarly, Coda, and Superhuman collaborate to drive higher forms of productivity.
I took the interview and asked my AI to predict (or hallucinate) about the future. Feel free to posit other questions. Here’s what it said.
1. Probability of Acquiring Additional Surface Technologies
Conclusion: High
Grammarly is highly likely to acquire more ‘surface technologies’—core applications like chat or project management platforms. This strategy is central to their stated goal of building a comprehensive “AI native productivity suite.” Analytical Basis: • Explicit Strategic Goal: CEO Shishir Metra’s objective is to “build the AI native productivity suite” by “tak[ing] ownership of the most important surfaces that people work in” (24:26). The acquisitions of Coda (for documents) and Superhuman (for email) are the first steps in this broader strategy. • Identified Gaps: Metra explicitly identifies chat as a key communication surface yet to be addressed (24:39), making it a logical and probable next acquisition target. • Aggressive M&A Mandate: When questioned on a potential limit to acquisitions, Metra’s stance was unequivocal, signaling a well-funded and open-ended strategy:
“I don’t think I have a limit… we’re ready to pull in whatever parts of the suite we need to to complete that experience.” (24:07 - 24:44)
2. Probability of Acquiring an Agentic Browser
Conclusion: Low
The acquisition of a browser is improbable as it contradicts Grammarly’s core platform strategy, which is designed to be ubiquitous and browser-agnostic. Analytical Basis: • The “AI Superhighway” Model: Grammarly’s platform is conceptualized as an “AI superhighway” (31:54) that enables agents to run on any existing infrastructure (browsers, OS, desktop apps). Their strategy is to provide the “cars” (AI agents), not to own the “road” (the browser). • Focus on Ubiquity, Not Exclusivity: The platform’s value derives from its ability to operate across over 500,000 applications (31:29). Acquiring a single browser would limit this universal reach and undermine their core value proposition. • Sufficient Existing Integration: Grammarly’s technology can already “read what’s on your screen, be able to annotate it… and then be able to take action on your behalf” (31:48). This provides the necessary deep integration without requiring browser ownership.
3. Probability of Acquiring Dia, Fellou, or Den
Conclusion: Indeterminate
The source material (interview transcript) contains no mention of the companies Dia, Fellou, or Den. Therefore, a meaningful analysis of their acquisition probability by Grammarly cannot be performed based on the provided evidence.
4. Probability of Acquiring or Hiring ‘Pack-Makers’
Conclusion: Very High
Grammarly will almost certainly acquire or hire “pack-makers” (agent creators) and their teams. This is a direct operational necessity for successfully launching and populating their planned agent marketplace. Analytical Basis: • Marketplace Strategy: Grammarly has confirmed it is building an “agent marketplace” (33:24) on the foundation of Coda’s existing “PAX marketplace.” • Leveraging Existing Assets: Coda’s marketplace already contains approximately 9,000 “packs” or agents (33:37). • Direct Integration Plan: The strategy involves migrating these existing assets. Metra states that “many of those agents already exist and we’re taking those and bringing them over to the Grammarly platform” (33:43). This migration necessitates direct engagement with top creators through partnerships, acqui-hires, or outright acquisitions to ensure quality and a successful transition.
5. Agent Evaluation Challenges & Coda’s Role
Key Insight: Grammarly Lacks an Agent Evaluation Strategy
While Grammarly has a clear vision for an agent-driven ecosystem, it has not publicly articulated a strategy for the critical function of agent evaluation and quality control. Grammarly’s Agent Vision: • Autonomous & Context-Aware: Agents that can reason, solve problems, and maintain personalized user context. • Interoperable: Agents that communicate seamlessly with each other and with external systems (e.g., CRM, ATS). • Ubiquitous: Agents that operate across a user’s entire digital footprint, discoverable via a central marketplace. Unaddressed Challenge: Agent Evaluation The transcript provides no information on the frameworks, metrics, or processes Grammarly will use to evaluate third-party agents for safety, accuracy, performance, and quality. This is a significant operational and reputational risk that remains unaddressed in the provided source. Coda’s Defined Role: Coda’s function is foundational but limited to providing infrastructure, not evaluation. A Workspace for Agents: Coda acts as a primary “home for interacting with those agents” (22:53) within a document-centric context. A Marketplace Platform: Coda’s “PAX marketplace” provides the technical foundation for the broader Grammarly agent marketplace (33:32).
6. Superhuman Integration & Innovative Capabilities
Conclusion: Superhuman establishes email as a core pillar of Grammarly’s “AI native productivity suite.”
This integration moves beyond simple writing assistance to transform the inbox into an agent-driven command center. New Capabilities Enabled by the Integration: The partnership creates a paradigm where email becomes an active “workbench for working with your agents” (28:41), enabling a new class of workflows. Capability | Description
Automated Inbox Triage | An agent organizes and prioritizes the user’s inbox Contextual Response Drafting | Agents draft email responses that mimic the user’s authentic personal voice and tone (26:46). Intelligent System Sync | Agents connect to external systems of record (e.g., ATS, CRM) to pull in or update data directly from the email interface (26:57). Natural Language as API | Users can command agents to perform complex multi-step actions using plain English, replacing manual processes (27:02). Proactive Agent Actions | Agents can independently execute tasks like scheduling meetings and surfacing relevant insights without direct user command (26:51). Agent-to-Agent Communication | The ultimate vision is a system where a user’s agent can negotiate and resolve tasks directly with another person’s agent, streamlining communication entirely (40:53).
