While much of the corporate world is attempting to force artificial intelligence into the daily habits of its employees through login tracking and performance reviews, HubSpot has taken a quieter, more organic path. The software-as-a-service company has achieved 100% AI adoption among its engineering team—not through executive decree, but by proving the tools actually worked.
The results of this non-mandatory approach are stark. According to company data, the integration of AI into engineering workflows has led to a 73% increase in the lines of code updated by its engineers. This milestone was the culmination of a three-part phased rollout that began in 2023, designed to let the workforce discover the utility of the tools on its own terms.
Duncan Lennox, HubSpot’s chief product and technology officer, who previously held leadership roles at Google and Amazon Web Services, believes the traditional corporate “push” is counterproductive. “We found mandates not to be effective,” Lennox said, noting that the transition required a shift in trust rather than a shift in policy.
Overcoming the reliability gap
For many developers, the hesitation to use large language models (LLMs) isn’t rooted in a fear of being replaced, but in a fear of breaking things. Lennox found that the primary barrier to HubSpot AI adoption for engineers was a concern over introducing errors, quality degradation and general system reliability.
The turning point came when adoption hit a plateau at 30%. Rather than ordering the remaining 70% to comply, Lennox and his team gathered and shared incident data. By showing a side-by-side comparison, they proved that teams using AI copilots were not negatively affecting system reliability compared to those working manually. This evidence-based approach caused usage to jump immediately to 50%.
As autonomous coding agents—tools capable of not just suggesting code but executing tasks—gained traction, adoption climbed to 80%. This growth was further accelerated by a culture of internal prestige. HubSpot’s most distinguished engineers were the first to master the tools, creating a natural competitive drive among junior developers who wanted to match the efficiency of their mentors.
A multi-model infrastructure
To sustain this momentum, HubSpot avoided locking itself into a single ecosystem. The company works closely with OpenAI, Anthropic, and Google, encouraging its teams to swap between models based on which one performs best for a specific use case. This flexibility prevents “model lock-in” and ensures the team is always using the most capable tool available.
The company also invested in its own customized infrastructure. This allows autonomous agents to read context, write code, run tests, and fix errors within a controlled environment. This strategic investment is part of a larger financial commitment. HubSpot invested more than $900 million into research and development last year, representing roughly 20% of its revenue, according to HubSpot Investor Relations.
External breakthroughs also played a role. Lennox highlighted the release of Anthropic’s upgraded Opus model in November as a catalyst, stating, “That was a meaningful step forward for coding, and that causes a spike because now you can do a bunch of things you couldn’t do before because the models have gotten better.”
The blurring of technical roles
The total adoption of AI is fundamentally altering how the company defines technical work. The traditional boundaries between front-end and back-end engineers, as well as the lines separating product managers and user experience (UX) professionals, are beginning to blur. As AI handles more of the rote syntax and boilerplate code, the role of the engineer is shifting toward architecture and orchestration.
This shift is also influencing HubSpot’s hiring strategy. While the company is still hiring engineers, it is doing so at a slower pace than in the pre-generative AI era. Lennox is specifically seeking “AI-native” engineers—those who learned to code using tools like Cursor or Claude Code. However, he maintains a strict requirement for fundamental knowledge, emphasizing the need for a grounding in computer science and software engineering principles.
The “no-mandate” playbook has since spilled over into the rest of the organization. Currently, 94% of all HubSpot employees use AI, following the same sequence: pilot the tool, make it accessible, provide training via events like hackathons, and share measurable outcomes.
Productivity vs. Market Perception
Despite the internal efficiency gains, HubSpot is navigating a volatile external environment. The broader tech industry has seen a significant correction in staffing; according to Challenger, Gray & Christmas, technology industry layoffs rose 33% in the first four months of 2026 compared to the previous year.
Wall Street remains skeptical of the long-term pricing power of SaaS providers in an AI-driven world. This tension was evident last week when HubSpot reported solid first-quarter earnings, yet saw its shares drop 20% after a revenue forecast for the current quarter came in slightly below analyst expectations.
| Metric | HubSpot AI Impact |
|---|---|
| Engineer Adoption Rate | 100% |
| General Employee Adoption | 94% |
| Code Update Volume | +73% |
| R&D Investment | $900M+ (approx. 20% of revenue) |
Disclaimer: This article contains information regarding company stock performance and financial forecasts. It is provided for informational purposes only and does not constitute investment advice.
As HubSpot continues to integrate agentic AI into its platform, the company’s next major milestone will be the results of its upcoming internal hackathons, which are designed to train workers on building their own custom AI agents to further automate business operations.
Do you think a “no-mandate” approach is sustainable for larger enterprises, or is a top-down push necessary for scale? Share your thoughts in the comments or join the conversation on our social channels.
