AI Agent Economy: How a Trend Became a Marketing Tool 🛫

Agents have become the hottest topic in enterprise software, and on this wave, ordinary bots are increasingly being passed off as "agents." Gartner already predicts that more than 40% of such projects will be shelved by the end of 2027. Let's figure out where the real money is and where it's just noise.

⏺ How big is the market
The AI agent market is estimated at about $7.6 billion for 2025 and around $11 billion in 2026, with analyst forecasts diverging significantly beyond that. What these numbers mean in practice can be seen from the market leader. According to Salesforce's CEO, its AI agent platform Agentforce has reached an annual revenue run rate of over $1 billion, attracted 18,000 customers in 121 countries, and its agent autonomously handles 84% of support inquiries.

⏺ Why the economics don't add up
Companies expect an average ROI of about 171% from agents, but only 39% actually see a measurable impact on profit (data from Q2 2026). MIT paints an even harsher picture: 95% of corporate pilots yielded no measurable return. Often the issue isn't even the models, but that they were never integrated into real workflows. We wrote more about this here.

⏺ Where agents actually deliver value
Tangible returns from agents come where there is a clear task and a well-defined metric. In the enterprise, this is back-office: customer support, HR, routine operations where time savings are immediately visible. In industry, it's working with large volumes of data, where an agent can iterate through options faster than a human.

Demand will only grow: Gartner expects that by 2028, agentic features will appear in one-third of all enterprise applications, compared to less than 1% in 2024.

In the cards — what agent washing is and how to distinguish a real agent from an ordinary bot being passed off as an agent 🤫

🌐 Website | 🎰LinkedIn