Today, the biggest gap between companies lies in their ability to rebuild workflows. According to BCG research, only 5% of businesses have embedded AI across all functions, while another 35% are scaling the technology but acknowledge they could be moving faster. Most firms, however, see almost no effect on costs or revenue.
According to Serhii Tokarev, an investor and founder of the Tokarev Foundation, AI itself is not to blame. Some organisations are genuinely transforming the way they work with AI. Others are simply buying “ready-made autonomy” rather than gradually changing their processes.
It is a common mistake to evaluate an AI system based on a single task or function in isolation. At Roosh, an internal assistant built on Claude ranks potential startup investments by team excellence and market opportunity and gives a short rationale for each. Analysts begin deeper research with the top 10 rather than checking 100 startups one by one; productivity in initial screening has increased roughly tenfold. The workflow changed, not just the speed of one task.
Where AI Turns Workflows into Measurable Results
Businesses should not expect artificial intelligence to immediately take over entire areas such as finance, legal support, or customer support. Roosh instead uses an NDA agent for the first review: a handbook defines acceptable terms, changes to request and issues to escalate. A lawyer checks and refines the proposed redline. Processing time fell from about 1.5 hours to 15–20 minutes.
“AI is well suited to areas such as invoice processing, large-scale data analysis, and contract review. The model takes on complex preparatory work, while a person evaluates a short, structured outcome,” says the founder of Tokarev Foundation.
When evaluating ROI, organisations cannot simply compare the cost of AI tokens with an employee’s salary. What matters is the cost of an approved result that meets the company’s requirements. It usually includes integrations, monitoring, repeated requests, model costs, and rework when something goes wrong. Some companies do not see productivity growth, even though they have started spending more on AI than on employees.
From AI Analysis to Action: Why Control Still Matters
AI can serve as the initial analyst when assessing a business. It reviews financing history, reconstructs the relevant competitive landscape, finds important data, and highlights information that requires further examination. By saving time on gathering basic facts, this technology allows a person to focus on the key question: whether the company has a chance to create long-term value.
“I never ask a model whether I should invest in a startup. The technology can analyse everything much faster than an entire team, but a human will still make the final call and take responsibility,” says Serhii Tokarev.
In his view, AI can move from analysis to action only when a clear scope is set for it. Each new level of autonomy must be backed by reliability, checks, and measurable quality.
“AI is not replacing operations or investment teams in our portfolio companies. It complements them and broadens their capabilities. What matters is not a single model, but how all these systems work together,” adds Serhii Tokarev.
What are five simple exercises to do at home?
View all responses
