AI Plug‑In Era Threatens Traditional SaaS Model, HubSpot Leads Hedge
Nicolas Bustamante argues the classic SaaS stack is obsolete, urging companies to become plug‑ins for AI assistants. HubSpot’s CEO Brian Halligan re‑branded the firm as a "Customer Harness Company" and opened its MCP server to all accounts, betting on two AI front doors as the market pivots to agent‑centric growth.
Why It Matters
The shift to an AI‑assistant‑first model rewrites the economics of SaaS growth. Companies that become plug‑ins can tap into the massive user bases of ChatGPT, Claude and Copilot without bearing the cost of acquiring end‑users, effectively turning the assistant into a distribution channel. This changes the calculus for net‑revenue retention, as expansion revenue will be driven by how often an assistant calls a given plug‑in rather than by traditional upsell tactics.
For investors, the emerging MCP ecosystem creates a new layer of valuation levers. Firms that own high‑quality data and robust APIs can command premium multiples, while pure‑play UI‑centric SaaS may see compression. The dual‑front‑door approach taken by HubSpot signals that even market leaders are hedging against uncertainty in the assistant hierarchy, a strategy that could become a template for other mid‑market players.
Key Points
- Nicolas Bustamante urges SaaS firms to become MCP plug‑ins rather than build native chat features
- FactSet reports MCP connector usage growing "like a hockey stick"
- Upwork CEO calls chatbots "the acquisition channels of the future"
- HubSpot re‑brands as "Customer Harness Company" and opens MCP server to all accounts
- HubSpot forecasts $3.7 billion in revenue for the current year while testing two AI front doors
Analysis
The rise of MCP as the lingua franca for AI‑driven workflows marks a pivot from product‑led to agent‑led growth, a transition that mirrors the earlier shift from on‑premise software to cloud SaaS. In the early 2010s, the value proposition moved from capital expenditure to subscription revenue; today, the value proposition is moving from subscription to usage‑based outcomes mediated by an AI assistant. Companies that can embed their core logic into an assistant’s decision tree effectively become invisible infrastructure, capturing revenue through per‑call or per‑transaction fees rather than seat licenses.
Historically, platform wars have been decided by network effects. OpenAI’s app platform, backed by heavyweight partners, is establishing a similar moat: the more plug‑ins an assistant can call, the more valuable the assistant becomes, and vice versa. This creates a virtuous cycle that could marginalize legacy SaaS that lack clean APIs or that have entrenched UI‑centric sales motions. The data‑rich verticals—financial data, market intelligence, and HR platforms—are already ahead, as evidenced by FactSet’s "hockey‑stick" growth and S&P Global’s 500‑plus MCP customers. Their early mover advantage may translate into durable moats as enterprises lock in data pipelines.
For founders, the strategic imperative is clear: prioritize API hygiene, invest in MCP compliance, and rethink pricing models toward outcome‑based or pay‑per‑call structures. For investors, diligence should shift toward assessing a company’s integration readiness and the stickiness of its data assets within the assistant ecosystem. The next inflection point will likely be the emergence of a standards body that governs data provenance and security across MCP, which could either democratize access for smaller players or consolidate power among the current data incumbents. Companies that navigate this transition successfully will define the next generation of SaaS moats.
