"Individual AI Productivity" and "Organizational AI Productivity" are Completely Different—HubSpot CEO Yamini Rangan's Growth Strategy in the AI Era at GROW ANZ 2026
In April 2026, at HubSpot's annual event "GROW ANZ 2026" held in Sydney, CEO Yamini Rangan took the stage. Having joined HubSpot in January 2020 and transformed almost everything within the company in just three and a half years since the arrival of ChatGPT (November 2022), she shared practical wisdom on growth strategies in the AI era.
The content, delivered in a conversation format, was not an abstract vision but a record of the concrete transformation HubSpot itself has undertaken. This article organizes the core of her talk—the historical patterns of platform shifts, the practice of the AI customer journey, the limits of vibe coding, and leadership in organizational transformation—for investors and business professionals.
1. The 15-20 Year Cycle of "Platform Shifts"—History Repeats Itself
1-1. Over-investment → Gatekeepers → New Winners and Losers
At the beginning of her talk, Rangan illustrated the historical patterns of the technology industry. "Every 15 to 20 years, a platform shift occurs. New technology emerges, people start building, often over-building, and then the question arises of who controls that technology, and winners and losers are decided. That is all there is to it."
Taking the internet era as an example, in the early 2000s, 97% of the fiber laid was unused (known as "dark fiber"). After the over-investment, Google became the gatekeeper of search, and new business models like Uber and DoorDash emerged as companies that could not have existed without the combination of the internet and mobile.
Rangan pointed out, "The same thing is happening with generative AI. There is clearly over-investment in data center capacity, which will create new winners and losers who did not move fast enough."
1-2. This Cycle is "Much Faster"
However, there is an important difference in this platform shift. "What took 8 to 10 years in previous cycles is now happening in 3 to 5 years. And everyone has a megaphone, does a podcast, and shares opinions on X or LinkedIn. So, navigating what we have seen as platform shifts feels even more chaotic," Rangan stated.
2. The Shift from "Individual AI Productivity" to "Organizational AI Productivity"
2-1. Everyone is Using AI, but Organizational Transformation is Halfway There
During the talk, Rangan asked the audience, "Who uses AI every day?"—almost everyone raised their hands. "Who has achieved significant growth or outcomes as an organization through AI?"—about half raised their hands.
Rangan positioned this gap as a core challenge. "Everyone is using AI. Everyone is using at least multiple LLMs or tools on a daily basis. However, we are all at the stage of exploring the transition from individual AI productivity to organizational AI productivity."
2-2. The Difference Between "Individual" and "Organizational" is "Context"
Individual AI productivity means that when you wake up in the morning and write an email, an LLM that understands your style becomes a better thinking partner. On the other hand, Rangan explains that organizational AI productivity requires much more context.
"How does the best campaign manager choose the optimal campaign? How does the best sales representative decide which deal to focus on? What should the best support representative prioritize? That requires the context of growth."
3. The AI Customer Journey—Concrete Use Cases Recommended by HubSpot
3-1. AEO (AI Engine Optimization)
The first thing Rangan mentioned was "AEO." Based on the reality that about 60% of Google search results do not lead to clicks, it is a new marketing method to optimize how information appears in LLM responses.
HubSpot has been experimenting in this area for two and a half years, and a dedicated team has systematized "how to appear in the answers provided by LLMs" as a science.
3-2. Data Agents & Prospecting Agents
Next, she mentioned the identification of ICPs (Ideal Customer Profiles) and the automation of outreach.
Data Agent: Automatically enriches company and contact information in the CRM with the latest data, expanding the TAM (Total Addressable Market) matched to the ICP.
Prospecting Agent: Identifies appropriate intent signals and automatically generates personalized emails tailored to the company and its challenges.
Rangan shared from her own experience: "When I started in sales, I was responsible for 500 accounts, and it was really hard to find information by researching 10-Ks and 10-Qs for each company. Now, AI makes that easy."
3-3. Sales Assistant
In the sales process, AI takes over tasks where there was "never enough time," such as taking notes during calls, entering data into the CRM, and drafting follow-up emails. The bot listens to the conversation, extracts action items, and even prepares emails for sending.
3-4. Customer Agent (70% Ticket Resolution Rate)
Rangan described the Customer Agent as "one of the first use cases where we found product-market fit." "Currently, over 10,000 customers are using it, and the ticket resolution rate is 70%."
However, she explicitly stated that this scaling is done with a "balance that does not sacrifice customer satisfaction."
3-5. The Golden Rule of Implementation: "Start with Your Own Challenges"
Rangan emphasized this repeatedly: "What is the company's goal? What is the team's goal? AI is a new platform shift, but you must start with your company's goals and what you want to rebuild. Not the other way around." The approach is to first identify the biggest challenges, choose one or two areas to work on deeply, and then expand to the next use case once you have gained scale and confidence.
4. The Limits of Vibe Coding: "With 3,000 engineers, not a single production app is vibe-coded"
4-1. Writing and Maintaining Are Different Stories
Rangan spoke candidly about vibe coding: "At HubSpot, we have 3,000 engineers working on agentic coding, but we haven't built a single core application using vibe coding."
The reason is clear: "Writing code is only a part of it. Maintaining that code, integrating it with 20 to 50 other applications, and keeping up with today's best practices—if the AEO changes, you have to reflect that in the solution. There is a big difference between the ability to have a scalable system and the ability to write code."
4-2. The Difference Between AI Output and AI Outcome
This was the distinction Rangan emphasized most strongly: "AI output and AI outcome are completely different. Writing a blog post or an email is an output. But whether that output actually generates growth is a different matter."
To generate growth outcomes, you need context: past email history, which competitor value propositions won and which lost, and how the brand should be expressed. Only by providing this context is output converted into growth outcomes.
"Personalized emails that know the company, know the brand voice, and know how to beat the competition get much higher open rates than generic emails and can win more deals."
5. HubSpot's Own Transformation—"There is nothing we haven't changed"
5-1. The Three-Stage Evolution of Engineering
Rangan explained the internal transformation at HubSpot since the arrival of ChatGPT in November 2022 in three stages.
Stage 1 (2023): Copilots and assisted coding. Started utilizing GitHub Copilot. Confirmed that product reliability was not compromised and built organizational confidence.
Stage 2 (2024): Transition to agentic coding. Model capabilities improved dramatically, leading to a transition to agentic coding using Cursor and Claude Code. Confirmed significant improvements in engineering speed and productivity metrics.
Stage 3: Building an optimized environment for HubSpot developers. Since general-purpose coding agents were not optimized for HubSpot-specific libraries and development methods, we built our own containerized environment. This allowed all agents to speak the same language, access the same data and tools, and utilize the same growth context. "100% of developers are doing agentic coding on this technology foundation."
5-2. Making our own Go-to-Market "Agent-First"
HubSpot is also transforming its own Go-to-Market activities with AI agents.
Demand Agent: Build ICP and expand accessible enriched contacts
Inbound Agent: Handles 82% of inquiries to the website. Learned pricing information, campaign information, qualification methods, and competitor information
Prospecting Agent: Significantly improved productivity for BDRs and sales representatives through personalized outreach
Sales Assistant: Provides deal risk analysis, closing time suggestions, and stage determination in a conversational format
Customer Agent: AI handles 60% of support tickets
According to Rangan, the combination of the Prospecting Agent and Sales Assistant is improving win rates and productivity.
6. Structural Changes in the Go-to-Market Organization—The Boundary Between Sales and CS Disappears
6-1. The Internet Era: The Fusion of Marketing and Sales
Rangan organized the changes in organizational structure historically. In the internet era, buyers began searching for information in advance, and the boundary between marketing and sales became blurred. Marketers needed to be more personal, and sales representatives needed to be more like marketers.
6-2. The AI Era: The Fusion of Sales and Customer Success
"What is happening now is the fusion of the boundary between sales and customer success," Rangan points out. The value of AI agents needs to be realized within minutes, and that value must be expanded immediately. Therefore, the "handoff" from sales to CS must no longer be a handoff, but a continuous process of delivering value to the customer.
At HubSpot, we integrated sales and CS into the same organization this year. The goal is to present ideas during the sales cycle and increase the speed at which value is realized in the CS phase.
6-3. The Profile of Talent to Hire: “Curiosity and an Experimental Mindset”
Rangan is clear about the criteria for talent to hire in an era where organizational boundaries are merging. “People who are curious and can lean into change, rather than fearing change itself. They must be builders who experiment and learn through technology. Curiosity, an experimental mindset, and a commitment to continuous learning—those are the hiring criteria.”
7. Key Points for Investors to Watch
7-1. The Versatility of the “AI Output vs. AI Outcome” Framework
The distinction Rangan presented between “AI output (text generation, etc.)” and “AI outcome (actual growth results)” is a framework applicable to AI investment in general. While many AI tools promote their output, platforms that can integrate organizational context (customer data, competitive analysis, brand voice) and convert it into outcomes are the ones that will possess a sustainable competitive advantage.
7-2. HubSpot’s Positioning
HubSpot has a dual narrative: demonstrating the shift to “agent-first” within its own company while providing that toolset to its customers. The figures of 10,000 companies adopting customer agents and a 70% resolution rate could serve as a foundation for monetizing AI features and driving upsells.
7-3. A Sober Perspective on Vibe Coding
Regarding the argument that “vibe coding will make SaaS platforms unnecessary,” the fact that HubSpot’s own 3,000 engineers are not using it in production environments is an important counter-evidence. From the perspectives of scalability, integration, and maintenance,