Anyone in Serbia who works in business development and bid preparation knows what this job looks like from the inside. When a call arrives from an international organisation, a United Nations agency, the European Union, the World Bank or the EBRD, the first thing you see is often not an opportunity. It is a mountain of documentation.
A hundred-page technical specification. Forms. Certificates that have to be obtained from state authorities. References, expert CVs, and a financial offer costed down to the last dinar. And a deadline that runs down without mercy.
Bureaucracy can also surprise you in places where you do not expect it. Some organisations still require the complete documentation to be delivered on paper as well. Once, after all the documents had already been collected, we spent an entire day copying, collating and scanning. We thought it was an hour or two of work. It took the whole day. That is one place where artificial intelligence, unfortunately, does not help.
Contracting authorities are turning to technology too
What is encouraging is that the organisations issuing the tenders are increasingly willing to admit that the process is too heavy. At a public procurement conference in 2026, the World Bank spoke about how digital tools, data and artificial intelligence can make procurement more efficient and more transparent. According to the conference report, the Project IQ tool was presented, along with the example of a medical equipment procurement in which a process that previously took around eleven months was reduced to a matter of days.
The European Union is moving in the same direction.
On 9 September 2025, the European Parliament adopted a resolution calling for public procurement rules to be simplified, noting that the current framework runs to 476 articles, or 907 pages of regulation.
On the side of those who bid, the change is moving even faster. According to an October 2025 analysis by the consultancy McKinsey, artificial intelligence can make the procurement function between 25 and 40 percent more efficient. In a global survey of more than 250 chief procurement officers from 40 countries, published in August 2025, Deloitte found that the companies furthest along in digitalisation achieve an average return on their investment in generative artificial intelligence 3.2 times higher than those still at the beginning.
Where artificial intelligence genuinely helps
What does this mean in practice for a team in Belgrade, Novi Sad or Niš?
First, the dull part of the work gets faster. Instead of someone spending days reading tender documentation and building a requirements table by hand, artificial intelligence can extract the mandatory conditions, deadlines and evaluation criteria in a few minutes, and flag the items that could lead to disqualification. CVs and references are adapted to the required format more quickly. If the tool is fed the company's own earlier winning bids and its knowledge base, the first draft of a proposal no longer starts from an empty document. It already exists, and a person then refines it instead of writing everything from scratch. Vendors of such tools cite time savings of 50 to 80 percent on drafting.
That is not a small thing. When several calls arrive at once and the deadlines are short, the biggest problem is not only time. It is that the same people who run the projects are pulled off implementation to write proposals.
Research by QorusDocs, published in February 2026 on a sample of around 300 professionals, shows that most organisations fail to respond to 10 to 20 percent of the tenders they receive, precisely because they lack the time and the people.
If artificial intelligence takes over the mechanical part of the work, a team can bid on more tenders without stopping work on current projects.
Two traps: confidentiality and blandness
But this brings us to the part that is discussed less often.
The first and most important issue is confidentiality. The client always comes first, and many tools retain what you enter and may use that data to train themselves. Feeding a client's sensitive data or confidential sections of a bid into such a tool is not acceleration. It is a risk that can damage both the business and your reputation. The rule should therefore be simple: use only tools that explicitly guarantee they do not store your data and do not use it to train models, and do not share the most sensitive information at all.
The second trap is subtler. In mid-2026, experts at Lohfeld, a consultancy specialising in proposal writing, pointed out that speed in itself is no advantage once every competitor is producing drafts equally fast. Artificial intelligence gives you structure and fluent language, but it does not give you what sets you apart: insight into the client, an understanding of the context, and the arguments that raise a bid's score.
If you leave the machine to write the entire proposal, you will get a technically correct but impersonal text. A text that makes no reference to an earlier conversation with the client, does not show that you understand what genuinely matters to them, and gives no compelling reason to choose you.
The human remains the line to the client
This is why a person has to stay in the loop. Some companies will assume the opposite: that artificial intelligence will soon do everything. It will not.
It handles repetitive, mechanical work very well, and there it is a valuable help. But the relationship with the client, the sense of what stands behind a call, strategic judgement and the particular tone that every bid must have for each individual client remain human work.
A good business development manager is not the one who generates text fastest, but the one who remains a living line to the client, and who reads every bid before it goes out.
Artificial intelligence is an excellent tool. But, like every tool, it remains only as good as the hand that holds it.
