Should Your Engineering Team Build AI Infrastructure or Buy It?

This is part one of a three-part series on how HubSpot transformed with AI. Everything we build at HubSpot exists to help our customers grow. So when generative AI emerged, our engineering team didn't just see a productivity tool; we saw an opportunity to build better products and get more value into customers' hands sooner.

What Happened

HubSpot CEO Yamini Rangan detailed their three-phase AI transformation strategy in a comprehensive blog post. The company moved from basic coding copilots to custom AI infrastructure, achieving 100% engineer adoption and a 73% increase in code output. They built their own agent execution platform when off-the-shelf tools couldn't integrate with internal systems, proving that strategic AI investment can compound across product development.

Why This Matters for B2B Marketing Leaders

Your engineering velocity directly impacts your ability to ship features that differentiate your product in competitive markets like HR Tech and FinTech. HubSpot's 51% improvement in engineering velocity means faster time-to-market for client-facing innovations. If your competitors are still debating AI adoption while you're building custom infrastructure, you gain a sustainable advantage in product development speed and consistency.

The Starr Conspiracy's Take

HubSpot's approach validates what we've seen across B2B companies: AI transformation requires infrastructure thinking, not just tool adoption. Their three-phase model, copilots, agents, then custom platforms, mirrors the AI implementation framework we recommend to clients. The key insight is their "measure, prove, then scale" methodology. They used incident data to prove reliability before removing guardrails, which eliminated the fear that typically slows AI adoption. For marketing leaders, this translates to faster product iterations and more responsive client experience improvements.

What to Watch Next

HubSpot will likely release parts two and three of this series covering their agent-first go-to-market strategy and AI-first operations. Watch how they connect engineering AI gains to revenue outcomes, this will provide the business case template you need for your own AI infrastructure investments.

Related Questions

How long does custom AI infrastructure take to build?

HubSpot's timeline shows 18 months from basic copilots to full custom platform deployment. Most B2B companies should expect 12-24 months for meaningful infrastructure that integrates with existing systems and delivers measurable productivity gains.

What's the ROI threshold for building versus buying AI tools?

When off-the-shelf tools can't access your internal systems or maintain consistency across your product suite, building becomes necessary for competitive advantage. HubSpot's 73% productivity increase and 80% adoption rate suggest the investment pays off when AI becomes core to your product differentiation.

Should marketing teams wait for engineering to finish AI infrastructure?

No. Marketing can drive AI adoption in content creation, campaign optimization, and client insights while engineering builds product-level infrastructure. The key is ensuring both efforts align with your overall AI strategy roadmap to avoid duplicated investments.