Vibe Management: How AI Assistants Are Taking Over Routine Managerial Work

What 'Vibe Management' Means for the Way Companies Are Run

A business owner describes a typical Monday that no longer starts with two hours of manual analysis: an AI assistant has already checked the metrics, reviewed task execution, spotted a drop in lead generation and sent three action options to the person responsible, leaving only a short report for the owner. The author calls this approach 'Vibe Management' — running a company through dialogue with AI assistants rather than through endless manual checks. It is, the article argues, the management equivalent of 'vibe coding,' where non-programmers build software by describing the desired result in plain language.

Vibe coding, the article says, has already shown that people without technical education can create functional programs by talking to AI. The same shift, the author claims, is now happening in management: task setting, metric analysis, financial control and data preparation for decisions can all be delegated to AI assistants. The key caveat is that the user must already be an expert in the field to judge the quality of what the AI produces — an average copywriter will not get excellent text from a neural network, and the same logic applies to management.

To support the trend, the article cites Gartner's projection that AI agents will be embedded in 40% of corporate applications by the end of 2026, up from less than 5% a year earlier, and McKinsey data showing that 88% of companies already use AI in at least one business function while fewer than 10% have scaled AI agents to tangible results. The author's conclusion is that most companies are trying AI but few are getting value from it, making speed of adoption the real competitive advantage.

Two worked examples are offered. In the author's own company, co-founder Valentin Vasilevski pushed AI-assisted coding early, so that platform features that once took months or years reportedly now take weeks, with employees and even clients building modules themselves. Separately, Marcus Rush, founder of real estate agency Rush Home, used Claude and Zapier MCP to build an AI agent named Russ that reviews a database of more than 11,000 contacts each morning, recalculates lead scoring across dozens of parameters and prepares follow-up plans for the agency's eight agents — without manual effort from the team.

Why Companies Pay for AI but Get Nothing Back

From Vibe Coding to Vibe Management

The article draws a direct analogy: just as vibe coding separated software output from coding skill, Vibe Management separates routine management work from a manager's time. The author's central claim is that repetitive tasks — reviewing metrics, preparing reports, checking numbers — can be handed to an AI agent, leaving the human to make decisions. Judgement about what counts as good work still belongs to the human, which is why the author insists that AI amplifies a competent manager rather than replacing one.

Why Subscription-First AI Adoption Fails

The piece is sharply critical of companies that buy subscriptions to premium models and declare 'we have implemented AI' without changing how work is actually delivered. The reported outcome, in the author's telling, is that the tools end up writing polite letters and generating images while business operations remain unchanged. The missing ingredient, the author argues, is a motivated leader: a founder or top manager who personally uses the tools, talks about results and becomes the example others follow. Valentin Vasilevski is cited as the person who played that role in the author's company, introducing AI-assisted coding systemically and pulling the whole team — and clients — into the building process.

Two Case Studies With Very Different Methods

The Rush Home example is the most concrete illustration. Marcus Rush, who the article says is not a programmer, assembled an agent called Russ with Claude and Zapier MCP; Russ re-scores leads from an 11,000-plus contact database every morning and updates the CRM with no manual input. The author presents this as the purest form of Vibe Management because no code was written and no developers were hired. The author's own company, by contrast, went deeper into engineering, with a co-founder who adopted AI-assisted coding at scale and changed how the platform was built and extended.

The Statistical Case for Acting Now

The Gartner and McKinsey figures, if accurate, point to a narrow window rather than a permanent state. AI agents are projected to move from a niche to standard corporate infrastructure in a short period, yet current success rates remain low. The author's framing — that this is a once-a-decade opportunity comparable to the internet or the personal computer — is an opinion, but the underlying data suggest adoption is about to accelerate and the gap between companies that try AI and companies that scale it is still wide.

Five Steps the Founder Used to Make AI Adoption Stick

This is a first-person, single-company perspective, not a controlled study — but it offers business owners and managers a practical sequence directly tied to the examples and figures it cites:

• Start with the founder, not the tool. The author says the transformation in his company was driven by co-founder Valentin Vasilevski personally adopting AI-assisted coding before anyone else; without that visible example, the author argues, adoption stalls. Make the leader the first power user before rolling tools out widely.
• Pay for the best models and remove friction. The article explicitly warns against saving money on premium AI subscriptions, comparing it to economising on tools for a master craftsman, and cites companies whose paid subscriptions produced nothing because employees had no reason to use them. Give teams access to the strongest available models and integrate them into existing workflows.
• Demand the prompt, not just the result. In the author's company, tasks involving analysis and document preparation are not considered complete unless the worker also submits the prompt used to produce the result — described as a record of how the person thinks, 'like a record of someone else's chess game.'
• Create tool-specific practice, not general AI news. The company runs internal channels focused on concrete tools — HeyGen for video, ElevenLabs for audio — and assigns a specific person to test new tools before they enter production.
• Run business-scoped hackathons with real prizes. The first internal hackathon defined the functional structure and data-collection logic used in the company's accelerator programme; hackathons are now regular, open to all employees and rewarded with tangible prizes rather than certificates.
• Treat the next 18 months as the window. Gartner projects AI agents will be embedded in 40% of corporate applications by end-2026, versus under 5% a year earlier, while the McKinsey data cited in the article show fewer than 10% of companies have scaled agents to tangible results — the author's case for moving before the gap closes.