Investor Serhiy Tokarev Explains Where AI Is Already Creating Measurable Business Value

Investor and AI HOUSE founder Serhiy Tokarev said AI is already delivering measurable business value in specific workflows, citing an internal Roosh assistant built on Claude that ranks startups and raised initial screening productivity roughly tenfold, and an NDA agent that cut contract review time from about 1.5 hours to 15–20 minutes. Tokarev cited Deloitte figures that 47% of corporate users have made decisions based on incorrect model outputs and 77% of business leaders are concerned about AI risks, arguing human oversight must remain. He said a unified LP platform built with Claude Code consolidates materials from more than 60 portfolio companies, and that real advantage comes from evaluation systems, industry-specific data and deep integration rather than access to advanced models.

AI agents are already conducting audits, managing operations, and negotiating contracts. As a result, it may seem that the corporate world has entered a new era. Artificial intelligence is reshaping business, and this is not just a passing trend. A divide has emerged between organisations that are gradually adopting AI and those that still see it as a plug-and-play solution. Serhiy Tokarev https://www.therecursive.com/enterprise-ai-doesnt-need-better-models-it-needs-better-workflows/ , an investor and founder of the AI HOUSE, explained why it is becoming increasingly difficult for businesses to operate without AI. How AI Helps Analyse Data without Taking Full Control Artificial intelligence becomes the first analyst https://officechai.com/ai/ai-can-do-ipo-filing-work-in-minutes-earlier-used-to-take-6-employees-2-weeks-goldman-sachs-ceo/ when a startup is being assessed. At Roosh, an internal assistant built on Claude ranks a long list by team excellence and market opportunity and returns a short rationale for each company. Analysts begin deeper research with the top 10 instead of reviewing 100 startups one by one; productivity at the initial screening stage has increased roughly tenfold. At the same time, the model cannot determine where exactly to invest: a human must ultimately make the decision and take responsibility. Serhiy Tokarev emphasises that artificial intelligence is only one component of the entire process. An NDA agent follows a handbook of acceptable terms, changes to request and issues to escalate, then prepares https://www.forbes.com/councils/forbestechcouncil/2026/07/28/how-ai-is-reshaping-the-role-of-the-data-analyst/ an initial review and redline for a lawyer to check. Processing time fell from about 1.5 hours to 15–20 minutes. The agent initially rewrote some acceptable clauses too broadly; more examples of valid wording improved its judgment. “Even when it becomes smarter and more accurate, it will still need human oversight. According to Deloitte, 47% of corporate users have made decisions based on incorrect model outputs. Moreover, 77% of business leaders are concerned about the risks associated with AI. This is why full control cannot be handed over to an algorithm,” notes the founder of AI HOUSE. A model’s ability to read information and recommend actions within a domain differs from its ability to take autonomous action in that same domain. Autonomy must remain within clearly defined operational boundaries. Human oversight does not hinder the use of AI. In fact, it is what makes the technology useful for business. Processes Win, Not Models Access to an advanced AI model should not be seen as a lasting competitive advantage. Real value comes from evaluation systems, industry-specific data, and deep integration. A unified LP platform built with Claude Code https://economictimes.indiatimes.com/ai/ai-insights/claude-code-users-to-get-17-less-weekly-usage-from-september-14-as-anthropic-ends-50-boost/articleshow/133647006.cms?from=mdr keeps fund materials and updates from more than 60 portfolio companies in one current context, reducing the manual work of assembling and reconciling information for prospective LP conversations. “AI is not meant to replace teams in our companies. It enhances their capabilities. Some models examine and interpret information, while others synthesise research in medical technology, fintech, and game development,” says Serhiy Tokarev. It is not about whether AI will replace people https://officechai.com/ai/jobs-that-wont-be-replaced-by-ai/ . The key question is where the model is already creating measurable business value, and where the market is still selling the future instead of the present.