Artem Permyakov, Directum: We will move towards the maximum intellectualization and robotization of processes

Artem Permyakov: In fact, in any spheres there are two key directions of development of AI: routine transactions where it is necessary to replace the person, and the transactions connected with the analysis of large volume of data. Such tasks are in many industries.

On production of AI it is already now applied to forecasting of breakdowns of the equipment. Using technologies of Internet of Things data collection from sensors is carried out. Then intelligent solutions analyze equipment indicators which preceded this or that fault, and predict possible problems.

Robotization of production, processes of quality control and purchase planning is also relevant. In the market there are already precedents of such solutions — the LG company is going to start the "smart" plant where all processes will be performed in a robotic type.

Other sphere is banks for which verification of client files and automation of technical support is relevant. Here robotization with elements of artificial intelligence is used, and chat-bots together with AI allow to exclude the person at communication with clients on the first line.

One more direction — automation of warehouse work. A fine example — robotic warehouses of Amazon which allow to think over logistics and to organize deliveries of combined loads using intelligent systems.

Many resources in public institutions are spent for processing of citizens' appeals: their classification, registration, readdressing on the end user. The artificial intelligence could be used for cost reduction of the state during the work with incoming requests.

Besides, AI can be connected to HR tasks, both internal, and external.

Internal tasks are an analytics:

• burning out of employees;
• given by which it is possible to judge the worker is going to leave the company or not;
• emotional strength, for example, the analysis of conflict messages between employees.

In other words, intelligent tools can collect data in the company, analyze, than employees are engaged, to reveal deviations.

External tasks of HR are, for example, an analysis of the summary. For this purpose of AI analyzes qualities and competences of a big circle of employees and creates an ideal portrait of the candidate for a specific position. Then chat-bots compare the entering summaries to the received profile and issue a preliminary estimate of the candidate.

The artificial intelligence can be applied in information security: carry out the analysis of leak of documents, fix what documents people address more often, for what purpose.

In any company it is possible to use AI for carrying out mass researches on the basis of a large number of data which the person is simply not able to process.

Artem Permyakov: Today AI works only in the hi-tech and advanced companies, such as banks and airlines. Production companies lag behind so far. There are, of course, exceptions which are ground on an innovation in production — for example, Kalashnikov Concern. But in general penetration of AI not deep. Mostly these tools are used for accomplishment of routine transactions, but under control of the person. Completely it cannot be excluded from process.

Artem Permyakov: In many solutions there is a risk of errors — any of producers does not give an absolute guarantee of accuracy of the tool as it depends on many factors. While decisions are made by the person, responsibility will be behind him. And when we try to exclude him, for example, in the analysis of agreements, the risk increases, and many are not ready yet for it to go.

Unfortunately, it is a vicious circle because technical defects are eliminated with the advent of practice and because people are afraid to apply them, there is not enough practice. Solutions perfectly work in vitro and when come up against real situations, make mistakes.

Artem Permyakov: As I already spoke, the key factor is responsibility. Responsibility for the made decision can be conferred on the specific person. And when using AI it is unclear from whom to ask result. People are not ready to transfer responsibility for solutions to artificial intelligence yet.

Besides, heads of the companies are afraid to make the business too transparent. Not only for, but also for external inspection bodies. This fear of implementation of new technologies creates problems too.

Artem Permyakov: We have a solution Directum Ario which at this stage is focused on accomplishment of routine transactions of document handling. We implement cases of recognition of data from the entering flow, we facilitate work of the clerk and in general any person who brings documents in a system. Such processing includes:

• Intellectual classification. Directum Ario automatically distributes the entering flow on document types, magazines of registration and responsible.
• Automatic filling of details. On the basis of classification the service creates the filled ECM system card. After that the secretary only checks correctness of filling of details.

These scenarios of use of AI are already tested in large corporations. In October we completed testing of the solution in "UEC Aircraft engine" and several more companies. The majority of pilot projects showed the accuracy of extraction of details at the level of 85% and above.

Artem Permyakov: We see the interest of the market in similar solutions. More we receive requests for preprocessing of the entering flow: registration and mass entering of documents in a system.

Also one large oil and gas company is interested in a possibility of the analysis of agreements. It is necessary for the customer that AI could compare the document to standard templates and take out recommendations about change of the agreement. It means creation based on ESM of expert system which not only stores content, but also provides its intellectual processing. For creation of recommendations such system can be connected to the analysis of the external regulatory base or use a set of standard forms in ECM. Thanks to processing of these data a system "understands" whether necessary conditions are complied in the agreement or not.

We assume that in the long term our clients will become interested also in opportunities of independent completion of intelligent solutions. Now setup is possible at the level of rules. So far we do it by own efforts, but we are going to develop the tool and to provide to customers the interface for development of the rules.

Artem Permyakov: Right now we work on automatic creation of draft resolutions according to documents, we develop chat-bots, we are going to connect the analysis of a natural language for communication with a system.

Together with it we are guided by development of search with elements of artificial intelligence which will help to create corporate Mini-Google. We want that the user by search of the solution of a specific objective, for example, at a request "issue a holiday", received the hint in a search line in the form of the necessary master of actions at once: "Execution of the application for a holiday".

Smart search should be able to carry out morphological analysis of a line. For example, we state a request: "all supporting documents from Lukoil". The intelligent system understands that under the words "all supporting documents" invoices, acts and delivery notes meant. "From Lukoil" it will decrypt as "all list of subsidiaries of Lukoil with which we conduct work".

Of course, intellectual search is not only AI and morphological analysis, but also change of the layout, correction of typos and facet filtering of results.

One more direction is a creation of expert systems which will prompt to users to whom to redirect the document who can help with certain to questions with whom to organize a meeting. Such solutions will be even more effective if not to tie use of AI only on Directum data analysis, and to include information from a system environment.

Further we will move towards the maximum intellectualization and robotization of processes, i.e. to develop the concept Robotic Process Automation (RPA) and Intelligent Process Automation (IPA). It both automation routine transactions and data analysis which can be used for acceptance of management decisions.

Robotization can help to close completely routine transactions in many processes of a back office. For example, the robot accountant will be able to analyze documents, to take information and to make postings — to perform a full stroke of accounting transactions. Robots lawyers will be able to analyze agreements in terms of possible risks, comparing them to standard forms. After such preliminary check the initiator at the time of start of the agreement will already know what controversial points should be corrected to reduce time for approval.

In the long term Directum Ario can become a separate product which will perform as a superstructure for interaction with any systems. We want that the solution grew in the intelligent tool for accomplishment of routine transactions and data analysis not only in ESM, but also in corporate systems in general.