The standalone AI meeting notetaker had a good run. But the runway just got a lot shorter.
Zoom unveiled agentic AI capabilities in February 2026, including My Notes and Personal Workflows, shifting from reactive support to proactive intelligence. In March, Zoom expanded its enterprise agentic AI platform with workflow orchestration that turns meetings, calls, and customer interactions into completed business outcomes. Zoom’s AI note-taker now works across Zoom meetings and third-party platforms like Microsoft Teams and Google Meet, bundled free with paid Workplace plans.
Meanwhile, Salesforce just announced 30 new AI features for Slack, with Slackbot now capable of transcribing meetings and summarizing them. Unlike standalone meeting tools, Slackbot acts as a personal meeting assistant with full enterprise context, routing work across Agentforce and any connected app the moment a call ends.
This is an existential moment for Otter, Fireflies, Fathom, tl;dv, and every standalone notetaker that built a business on transcription, summaries, and action items. When those capabilities come bundled free with the platforms where meetings already happen, the standalone value proposition collapses. Transcription is no longer a product. It’s a feature.
Then there’s Granola, which just complicated the narrative. The company secured $125 million in Series C funding, vaulting its valuation to $1.5 billion, up from $250 million less than a year ago. Revenue grew 250% in the prior quarter.
That’s not a company bracing for impact. That’s a company betting it can outrun the platform wave. Granola acknowledges that AI meeting notes are becoming a commodity and is pivoting to a broader enterprise AI platform, introducing new APIs to integrate meeting context into AI workflows. Its bot-free, on-device approach gives it a privacy edge that Zoom and Slack can’t easily replicate, since both are cloud-native. But $1.5 billion is a big number for a company still racing to prove that meeting intelligence can become something more than a feature swallowed by its host platform.
And then there’s the company everyone in this conversation keeps underestimating: Dialpad (and yes, I’m proudly a shareholder, for transparency).
Friend and founder, Craig Walker, is a serial entrepreneur who founded GrandCentral, which was acquired by Google in 2007 and became Google Voice (Comunicano was their agency for that and I was a shareholder). Before that, he led Dialpad Communications, which Yahoo acquired in 2005 and became Yahoo Voice. The man has built the plumbing of modern voice communications twice over. He founded the current Dialpad in 2011 and, in 2018, acquired TalkIQ, a real-time speech-recognition and AI company, which formed the foundation of Dialpad’s AI products. That was years before “AI-native” became a pitch deck buzzword.
Here’s what makes Dialpad structurally different from everyone else in this conversation.
After years of continuous improvement and billions of minutes of voice calls analyzed, Dialpad’s proprietary transcription model surpassed Google’s enhanced telephony model for both keyword and general accuracy. DialpadGPT, the company’s proprietary LLM, makes it the first large language model designed specifically for enterprise business communications. That’s not a bolted-on AI feature. That’s a platform where intelligence was the architecture from the start.
At Enterprise Connect 2026, Dialpad debuted new agentic AI capabilities, including the ability to analyze historical conversation data to pinpoint customer experience gaps, no-code AI agent creation for voice and digital channels, and closed-loop analytics that tie conversations to measurable business outcomes.
Ninety-seven percent of Dialpad contact center customers are already using AI, with over 775 million AI Recaps and 450 million AI CSAT scores delivered to date. Those aren’t pilot numbers. That’s production at scale.
As CEO, Walker put it at Enterprise Connect: “Enterprises aren’t struggling with AI ambition. They’re struggling with AI execution.”
That framing matters. Zoom and Slack are adding AI to collaboration. Granola is racing from notetaking toward enterprise workflows. But Dialpad started from the conversation itself, built its own speech models, trained its own LLM on real business dialogue, and is now deploying autonomous agents that don’t just summarize what happened but take action on it. Dialpad’s multi-LLM architecture combines specialized models for speech recognition, intent detection, and task execution, designed to continuously learn from previous interactions.
The lesson here is one the tech industry relearns every cycle: when the platform absorbs the feature, the feature company either evolves or evaporates. Vertical specialists like Grain, whose AI can identify buying signals and track competitors’ mentions, represent the kind of depth that platform tools won’t quickly replicate. Granola’s $125 million bet says meeting context can become the connective tissue of enterprise AI. Both are defensible plays, but plays with a clock on them. The key is: can they defuse the ticking AI time bomb looming over them, either by exiting or by finding a way to stay relevant?
Dialpad’s position is different and arguably stronger. It doesn’t need to outrun the platform wave because it is a platform, one where AI isn’t a layer added on top but the foundation upon which everything else rests. In a world where Zoom offers transcription for free, and Slack’s Slackbot can summarize your call before you’ve closed your laptop, the only durable advantage is owning the intelligence layer, not renting it from the meeting. That’s Dialpad’s advantage.
The standalone notetakers built a category. Zoom and Slack are commoditizing it. Dialpad is the quiet signal that the real game was never about the notes. It was always about what happens during and after the conversation ends.