In Russia, they created a multi-agent system “Radar” to automate procurement processes

Источник изображения: Эксепто

Excepta developer of multi-agent AI systems for automating corporate processes, presented a new AI solution – the system “Radar” for purchasing analytics and market intelligence. The development helps turn any supply request into a ready-made market map with manufacturers, options for product solutions and their analogues. This allows you to reduce preparation for purchasing from several weeks of manual search to 50 minutes and receive a ready-made market sheet in a few clicks.

Radar helps automate one of the most time-consuming processes in procurement: market analysis, searching for suppliers, collating responses from contractors and checking characteristics. Based on artificial intelligence algorithms, the solution builds a complete knowledge map about any procurement category and provides data on suppliers, products, lines and models, analogues and other replacement options from trusted sources in a convenient format. At the same time, the system independently finds and analyzes the resources of not only Russian, but also foreign manufacturers, including the Chinese segment. This allows us to cover the entire spectrum of the market, taking into account the growing popularity of Chinese suppliers in categories such as circuitry, industrial equipment and components. To start the process, it is not necessary to have a complete technical specification, since the service independently compares and ranks options according to criteria and answers questions on terminology, GOST and TU in the chat.

  Image source: Expo

Image source: Expo

Unlike AI assistants, which answer specific point questions and are not always able to work with context, Radar closes an entire stage of the business process: it builds a market map, connecting information about equipment with its manufacturers, suppliers and service organizations. The system brings together contractors’ information from multiple sources into a single table and prepares materials for making purchasing decisions. The system also helps to select a replacement taking into account operating conditions: the user can describe the task or download a technological map with the necessary characteristics.

Inside the platform, there is not just one language model, but a multi-agent circuit. Several AI assistants perform different parts of the task as one system according to regulations – each has its own role, rights and limits, and all steps are recorded in a log. The AI ​​model may change in this case, since the value of the system is not in access to a specific neural network, but in the discipline and context around it.

Three placement options are available: subscription, customer isolated and hybrid. According to the company’s calculations, on approximately every third competitive lot there will be an offer that is 6–7% more profitable, which gives about 2% of the net effect from the entire competitive volume, which is about 2 billion rubles. per year for every 100 billion rubles. competitive procurement.

“Procurement is a fairly resource-intensive and time-consuming process within any business, but sometimes decisions worth millions are made based on several tenders and a manual spreadsheet. In the procurement process, the effect of introducing AI solutions in the coming years will be most noticeable, since the result is calculated in money and tested on real lots, and not in sensations. At the same time, during the development of Radar, we fundamentally left the decision on the human side: the system prepares a justification with links to each source, but the purchasing decision is still made by the manager,” commented Roman Venediktov, General Director Except.

If you notice an error, select it with the mouse and press CTRL+ENTER. | Can you write better? We always welcome new authors.

Advertisement | JSC “Selektel” INN: 7810962785 erid: F7NfYUJCUneVfSP6LveS

Source: