Here’s what’s on our plate today:

• 🧪 Grammarly, Coda, and Superhuman merged. Here's what the survival playbook looks like.

• 🗞️ Musk's Tesla confession, hypersonic fighters, and robot training gigs.

• 🗓️ A weekend to-do list for testing the tools at the center of this week's story.

• 🗳️ Can the cross-ecosystem bet actually hold off Microsoft and Google?

Let’s dive in. No floaties needed…

The Laboratory

TL;DR

• ChatGPT broke Grammarly’s moat: Revenue growth dropped from 43% to roughly 12% as free LLMs replicated core writing assistance, and Microsoft Copilot and Google Gemini started bundling the same features natively into enterprise productivity suites.

• Superhuman hit the same wall from a different angle: At $35M ARR and $30/month per seat, a premium email interface was a tough sell once Gmail and Outlook shipped their own AI features as defaults.

• The rebuild was structural, not cosmetic: Grammarly acquired Coda in late 2024 for its workspace and knowledge graph, then Superhuman in mid-2025 for email workflow intelligence. The combined entity rebranded as Superhuman Platform Inc. and launched Superhuman Go, an agentic assistant that works across 1M+ apps via a browser extension.

• The cross-ecosystem bet: Microsoft Copilot lives inside Microsoft, Google Gemini lives inside Google, and Superhuman Go works across both and everything in between. The wager is that enterprises do not live in a single vendor’s world.

• Moving fast has costs: A class-action lawsuit in March 2026 forced the removal of an “Expert Review” feature that used the likenesses of authors and journalists, including Stephen King and Kara Swisher, without consent. Rebuilding a brand while outrunning your own ethical guardrails is a pattern that compounds reputational risk fast.

How Superhuman became the blueprint for surviving the AI shift

For end users, technological progress rarely arrives without a learning curve. Each new wave of tools demands a degree of upskilling, requiring people to understand not just how the technology works, but how to use it effectively enough to keep pace with the productivity gains it promises. That process can be disorienting, especially when both the technology and its use cases continue to evolve in real time.

This adjustment is not limited to individuals. Enterprises, too, are forced to rethink how they operate, adopting new systems, retraining teams, and reorganizing workflows to capture those same gains. But for companies whose products are built on technology itself, the transition is more complex. They are not only integrating new tools into their operations but also reworking their business models to ensure they remain relevant as the underlying technology shifts beneath them.

One such company that has successfully transitioned from pre-AI-era software to the adoption of agentic AI is Superhuman Platform Inc. This parent company brings together Grammarly, Coda, and the original Superhuman email client under a single brand.

Each of those three products was built before the generative AI wave reshaped the software industry, and each faced its own version of the same existential question: what happens when the technology on which your product is built becomes a commodity?

From spell-check to survival

Grammarly was founded in Kyiv in 2009 by Max Lytvyn, Alex Shevchenko, and Dmytro Lider as a subscription grammar-checker aimed at students, built on rule-based natural language processing that flagged spelling errors, grammatical mistakes, and awkward phrasing. The founders bootstrapped the company for eight years, nearly running out of money multiple times, before raising a $110M round led by General Catalyst in 2017. By that point, Grammarly had crossed 7M daily users and was already profitable, according to CNBC. A freemium browser extension drove viral adoption through the late 2010s, a Grammarly Business tier extended the model to enterprise teams in 2019, and by November 2021, a $200M raise from Baillie Gifford and BlackRock valued the company at $13B with 30M daily active users.

Then, in late 2022,  ChatGPT’s successful launch started threatening Grammarly’s core value proposition, since a free tool could now perform many of the writing-assistance tasks Grammarly charged for, according to Plagiarism Today. Revenue growth, which had been running at roughly 43% year-over-year, slowed to approximately 12% by 2023 to 2024, according to Sacra. Grammarly responded in March 2023 with GrammarlyGO, a generative AI layer built on OpenAI’s GPT-3, but adding generative capabilities to a grammar-checker did not resolve the deeper problem: Microsoft was embedding Copilot across its Office suite, Google was integrating Gemini into Workspace, and both offered writing assistance as a native feature within the productivity environments where enterprise workers already spent their days. The question was no longer whether Grammarly could add AI to its product, but whether a standalone writing assistant could survive as a category at all.

Around the same time, Superhuman, the email client, faced a similar constraint.

A different product, the same problem

Founded in 2014 by Rahul Vohra, Superhuman was built around a single defining attribute: speed. The product promised sub-100 millisecond response times, paired with aggressive keyboard shortcuts and a design philosophy shaped by Vohra’s background as a professional game designer. He spent years resisting a public launch, focusing instead on refining the product and developing a now widely cited framework for measuring product-market fit, built around a single survey question and later detailed in a First Round Review essay that became a reference point for startup teams.

The company cultivated deliberate exclusivity, with 20 full-time onboarding specialists conducting individual calls with every new user, and the email signature’Sent via Superhuman’ functioning as a status signal that drove roughly 30% of website traffic. By 2024, Superhuman had over 50k paying customers at $30 per month, including teams at Netflix, Brex, and Spotify, but total ARR (annual recurring revenue) reached only an estimated $35M by mid-2025. As Gmail and Outlook added their own AI capabilities as default features, Superhuman’s differentiation as a premium interface layer was narrowing.

Rebuilding the business, not just the product

To counter the constraints, Grammarly chose expansion over optimization. In December 2024, it acquired Coda, a collaborative workspace with $41.1M in 2024 sales, in an all-stock deal where Grammarly shareholders held approximately 90% of the combined entity. The critical element was leadership: Coda CEO Shishir Mehrotra, a former chief product officer at YouTube, became Grammarly’s new CEO, bringing a vision he had titled “the AI-native productivity suite” in his own planning memo. Coda contributed Coda Brain, an AI layer running across 800+ integrations, and Coda Docs, providing Grammarly with a workspace surface and knowledge graph infrastructure it had never had.

In May 2025, the company secured $1B in non-dilutive financing from General Catalyst, using a revenue-based structure that preserved its $13B valuation. Two months later, Grammarly acquired Superhuman, bringing Vohra and his team into the company as part of a broader push into AI-powered productivity, with email already its largest use case, accounting for roughly 50M messages edited each week. By 29 October 2025, the combined entity had rebranded as Superhuman Platform Inc. and introduced Superhuman Go, an agentic AI assistant, according to BusinessWire.

The agentic bet

Superhuman Go works across more than 1M apps and websites via a browser extension, orchestrating AI agents that pull context, suggest actions, and take steps without the user having to switch tools. At launch, it included over 100 agents, from connector agents for Google Workspace, Salesforce, and Jira to partner agents from Radical Candor, Fireflies, and Quizlet, with an Agents SDK in closed beta, according to Reworked. The technical foundation draws on all three acquisitions: Coda Brain provides the knowledge layer, Grammarly’s browser distribution provides reach, and Superhuman Mail contributes reactive interface design and email workflow intelligence.

The competitive positioning is deliberate: Microsoft Copilot works inside Microsoft apps, Google Gemini works inside Google apps, and Superhuman Go works across both ecosystems and everything in between, betting that enterprises do not live in a single vendor’s world.

For the new existing entities, the transition has not been frictionless. In August 2025, the company launched Expert Review, a paid feature that generates AI writing feedback attributed to real authors and journalists, including Stephen King and Kara Swisher, without their consent. A class-action lawsuit filed in March 2026 by journalist Julia Angwin forced the feature’s removal and a public apology from CEO Mehrotra, illustrating how companies moving fast to reinvent themselves can outrun the ethical processes that established brands depend on.

Deloitte’s 2026 State of AI in the Enterprise report found that while worker access to AI rose 50% in 2025 and two-thirds of organizations reported productivity gains, only 34% are doing what the report characterized as “reimagining the business” with AI rather than simply adding features to existing workflows.

What Superhuman Platform ultimately illustrates is that the challenge of adapting to new technology is not limited to learning how to use it, but extends to rethinking what a product is meant to be in the first place. Just as individuals are forced to continuously upskill to keep pace with shifting tools, companies built on those tools are often pushed into far more consequential decisions, where survival depends on redefining their role in a changing ecosystem rather than simply improving what already exists.

Adapting is no longer enough

Grammarly did not just add AI features, Superhuman did not just refine its interface, and Coda did not just expand integrations. Each was forced to confront the same underlying reality: the value they once provided could be absorbed by larger platforms or replicated at near-zero cost. The response, in this case, was to rebuild around a new premise altogether, moving from standalone tools to a system that attempts to orchestrate work itself.

Whether that bet succeeds remains uncertain, but the direction is telling. In an era where technology evolves faster than the structures built around it, adaptation is no longer about incremental improvement. It is about recognizing when the ground has shifted entirely and having the willingness to rebuild before it becomes obvious to everyone else.

Headlines You Actually Need

• Hypersonic arms race: Defense startup Hermeus raised $350M and hit a $1B valuation to build autonomous hypersonic fighters, applying SpaceX-style rapid prototyping to aircraft development.

• Your chores, their training data: Startups are paying contractors around the world $5-$20/hour to film themselves doing household tasks, building the egocentric datasets that humanoid robots need to learn how to function in real homes.

• Musk's real priority: Tesla's former president revealed that Musk privately defined success at the company as getting down to one day a week at the company so he could focus on rockets.

Weekend To-Do

• Superhuman Go: Test the browser extension across your actual workflow and see how far an agentic assistant gets before it needs you.

• Coda: If you're still running docs and spreadsheets in separate tools, spend an hour building one Coda doc that replaces both.

• Microsoft Copilot vs. Gemini: Pick one task you do weekly and run it through both. Map where each breaks down. That gap is exactly what Superhuman Go is betting on.

Friday Poll

The Toolkit

• Regie: AI-powered sales copilot to draft, personalize, and refine outbound messaging.

• Replit: In-browser AI coding environment for writing, debugging, and running apps fast.

• Sourcegraph: A code intelligence layer that lets AI search, understand, and refactor huge codebases.

Rate This Edition