<- all tokdocs

Should You Vibe Code Your Own CRM Instead of Paying for HubSpot?

By , founder of Mavric and Apso

View on TikTok ->

A founder on TikTok argues that replacing a paid CRM like HubSpot with one you vibe code yourself is a mistake, because the engineering time to build and maintain it costs more than the license. That opportunity-cost argument is sound. His timeline is off in places, and the credentials he cites could not be confirmed from any public source.

The video comes from Joe Koenig (@joe.koenig), who says he sold his software company at the end of 2024 and has since wondered what today's AI could have done for it. The one thing he says he would not do is build a lot of tools internally.

Is it worth building your own CRM with AI?

Koenig's worked example is his own HubSpot bill. "We used HubSpot. And I think at the end, we were paying HubSpot about 30 grand a year." He says he could have saved that by building a CRM, "But I think the opportunity cost would have been absolutely massive, I definitely would have had to pull some of my dev team away from what they were doing, in order to not just build the initial product, but then ultimately maintain it add new features, updates, those kinds of things."

The structure of that argument holds. A license is an operating expense that ends when you cancel. An internal CRM is a permanent claim on engineering time that grows with every integration, migration and compliance requirement.

The ownership problem is the strongest part of the case. Simon Willison, quoted in the Wikipedia article on vibe coding, calls vibe coding a production codebase risky, because "Most of the work we do as software engineers involves evolving existing systems." A CRM fits that description. It holds customer records and the business depends on it for revenue.

Koenig's closing advice is "it's really important to not lose focus on what your company does."

How much does HubSpot cost?

His $30,000 figure is a private contract and a recollection he hedges with "I think." It is consistent with HubSpot's published list prices:

Product Tier List price (billed annually) Onboarding fee
Marketing Hub Professional $800/month, 3 core seats, 2,000 contacts $3,000
Marketing Hub Enterprise from $3,600/month, 5 core seats, 10,000 contacts $7,000
Sales Hub Professional $90/seat/month ($100 billed monthly) $1,500
Sales Hub Enterprise $150/seat/month $3,500

Marketing Hub Professional comes to $9,600 a year. Sales Hub Professional is $1,080 per seat per year. Marketing Hub Professional plus about nineteen sales seats lands near $30,000, and so do about twenty-eight Sales Hub Professional seats on their own.

These are current prices, not what he would have paid in 2024. HubSpot's seat-based packaging has changed since then, so the math shows $30,000 is a normal HubSpot bill for a company that size, not that it was his bill.

Are companies really replacing SaaS tools with their own builds?

They are, according to survey data. Retool's 2026 build-versus-buy report, published February 17, 2026, surveyed 817 people. It found that 35% of teams have already replaced at least one SaaS tool with a custom build, and 78% expect to build more of their own tools in 2026.

The categories teams are replacing:

  • Workflow automations: 35%
  • Internal admin tools: 33%
  • BI tools: 29%
  • CRM and form builders: 25%
  • Project management: 23%
  • Customer support: 21%

CRM is fourth on the list. One caveat: the respondents are Retool's own customers and builders, a group already inclined to build internal tools, so the percentages likely run high. The report also lists maintenance burden at 26% as a barrier to more AI automation, which is Koenig's point made by the people doing the building.

Was late 2024 "the very beginning" of AI?

Koenig says his exit "was kind of like right at the very beginning of the AI stage." For AI in general, that is off. On June 21, 2024, six months before his stated exit, Anthropic's Claude 3.5 Sonnet announcement said the model "can independently write, edit, and execute code with sophisticated reasoning and troubleshooting capabilities." In an internal agentic coding evaluation it "solved 64% of problems, outperforming Claude 3 Opus which solved 38%."

For the capability his argument depends on, the framing is closer to right. Claude Code launched as a "limited research preview" on February 24, 2025, about two months after his exit. Anthropic described it as "an active collaborator that can search and read code, edit files, write and run tests, commit and push code to GitHub, and use command line tools." That is the kind of tool that makes vibe coding a CRM realistic.

The term itself is also newer than his exit. Andrej Karpathy introduced "vibe coding" in February 2025. Merriam-Webster listed it as slang and trending in March 2025, and Collins English Dictionary named it Word of the Year for 2025.

Does AI make developers faster?

One study often cited against the build-it-yourself case should not be used that way now. METR's randomized trial, published July 10, 2025, had 16 experienced open-source developers work on 246 real issues. With AI tools available, they took 19% longer, though afterward they estimated AI had made them about 20% faster.

The post now carries a banner that reads "These results are out of date" and links to results current as of early 2026. The authors say the older findings no longer reflect how current AI models affect developer productivity.

About the credentials in the video

Koenig closes with "I'm Joe, I grew my company from zero to 10 million ARR, exit at the end of 24." The video never names the company, and no public filing, press release or acquirer announcement confirms the figures. The advice stands or falls on its own logic.

Bottom line

The opportunity-cost case against vibe coding your own CRM holds up: building it pulls engineers off the product, and maintaining it never ends. His $30,000 HubSpot bill is in line with list prices for a company his size. Survey data confirms many teams are building instead of buying, with CRM among the top categories. Late 2024 was not the start of AI, but it was just before the agentic coding tools that make this choice real.

Sources

Published October 26, 2026. Based on a TikTok by @joe.koenig. Full video notes (markdown)