When a Feature Becomes a Threat
Notion’s AI-powered database features have moved well past novelty. With its latest rollout of AI property types – fields that can automatically summarize, classify, and generate content directly inside database cells – Notion is no longer just a note-taking app that grew up. It is now a direct competitor to the structured, formula-driven workspace that Coda has spent years building its identity around.
Coda built its reputation on being the tool for power users who wanted something between a spreadsheet and a custom app. Its “Packs” ecosystem, two-way data connections, and deeply nestled formula logic gave it credibility with operations teams, product managers, and startups that needed real workflow automation without hiring engineers. That positioning held up fine when Notion was mainly a hierarchy of pages. It holds up less convincingly now.
Notion now ships AI-native database properties as a core product feature, not a bolt-on.
What Notion’s AI Databases Actually Do
The technical gap between the two products has narrowed faster than most observers expected. Notion’s AI properties let users create database columns that automatically pull context from linked pages, generate summaries on demand, extract key dates or action items, and classify entries by content type – all without leaving the database view. For teams that were using Coda’s formula columns to accomplish similar outcomes, the question now is whether Coda’s added complexity is still worth the learning curve.
Coda’s strength has always been depth. Its formula language can reference across tables, trigger actions, build interactive buttons, and connect to external services through Packs. None of that disappears because Notion added AI columns. But the user who originally chose Coda because it could do more than Notion now has to ask whether Notion has caught up enough for their specific use case – and for a growing number of those users, the answer is yes. Notion’s AI features cover the most common operations: summarization, tagging, status inference, and content generation. Those are not edge cases. They are the bulk of what most database power users actually do day to day.
Notion has also made pricing part of the pressure. AI features are bundled into its Plus and Business plans rather than priced as a separate add-on, which means teams already paying for Notion get AI database functionality at no additional line-item cost. Coda’s AI features, by contrast, live behind a separate AI add-on tier. For a team evaluating both platforms, the total cost math shifts noticeably when one product bundles AI and the other charges extra for it.
Coda’s Loyalty Problem
The users most likely to churn are not Coda’s newest customers. They are the mid-size teams that adopted Coda two or three years ago because it offered genuine structural advantages over whatever they were using at the time. These teams have built out workflows, trained their colleagues, and accumulated institutional knowledge inside Coda’s ecosystem. Migration friction is real, and Coda benefits from it. But friction only holds for so long when the alternative keeps closing the capability gap.
Coda has not been standing still. Its AI features include a document-level AI assistant, formula suggestions, and a growing library of automations. The company has also leaned into its “doc that thinks” framing – positioning Coda AI as something that understands the structure of your document rather than just generating text on top of it. That framing has genuine merit. Coda’s AI can reference table relationships and trigger actions in ways Notion’s AI properties currently cannot. But that advantage is technical, and technical advantages require users who care enough to notice them.
The risk for Coda is not that it becomes a bad product. The risk is that it becomes a great product for a shrinking audience. If Notion continues expanding its AI database capabilities – and every signal suggests it will – Coda’s differentiation moves further and further toward the edge cases: complex multi-table logic, Packs integrations, button-triggered automations. Those features matter deeply to a specific kind of user. They just do not matter to most users, which is exactly the problem. This is a dynamic worth watching in the broader enterprise knowledge tool consolidation playing out across the productivity software category.
Coda’s strongest defense right now may be its formula language – the thing that made it feel like a real development platform rather than a document tool. But Notion is reportedly building toward more programmable logic inside its database layer, and once that arrives, even that last moat gets harder to defend. The question is not whether Coda survives. It will. The question is whether it ends up as a specialist tool used by a few thousand highly technical teams while Notion takes everything above and below that ceiling.
