Could artificial intelligence encourage the decentralisation of business management tools?

The integration of artificial intelligence into businesses could gradually change how their management tools are designed and organised. One possible development would be to create intelligent functions tailored to specific needs, then connect them to systems already in use. Businesses would thus have greater freedom to define their digital environment, without necessarily entrusting their entire operation to a single platform.

This prospect could open the way to a form of decentralisation of management tools. Here, the term would mean greater autonomy in choosing, designing and developing solutions, rather than dispersing information. A business could retain a centralised view of its activity while relying on several specialised components selected according to its own requirements.

Artificial intelligence could contribute to this development by facilitating certain interactions between users, data and software. The extent of this transformation would, however, remain dependent on the reliability of integrations, their cost and businesses’ ability to retain control over them.

A digital organisation shaped more closely by the business

In this scenario, the design of a management tool would start more directly from how the organisation operates. Its procedures, constraints and team responsibilities would form the basis for defining features.

A service company might, for example, want to connect incoming sales enquiries to quote preparation, then to the tracking of service delivery. A workshop might prioritise coordinating jobs and equipment availability. For a restaurant, the priority could be purchasing or reconciling sales with stock.

These needs would lead to different environments, even among businesses in the same sector. Some functions could come from existing software; others would be custom-built. AI would be integrated at the stages where its use would offer an identifiable benefit.

Such an approach could also allow gradual development. The business would begin with a clearly defined problem, assess the usefulness of the solution introduced, then decide whether to extend it. It would therefore not necessarily have to undertake a complete overhaul of its management system to benefit from a new feature.

Applications directly linked to operations

Purchasing management would be one possible initial application. In a restaurant, a tool could cross-reference recorded sales, recipe sheets and recorded stock to prepare order proposals. The manager could adjust the assumptions to account for reservations, a planned closure or a menu change. Proposed quantities would remain open to checking before anything was sent to suppliers.

In a garage, an integration could focus on job scheduling. The system would examine the estimated duration of work, available skills and announced delivery times for parts. When an issue threatened an appointment, it could identify the affected cases and prepare a proposal for rescheduling. The final decision would rest with the workshop manager, particularly where the situation required a commercial or human judgement call.

For a service company, the intelligent tool could become involved as soon as a customer request arrives. It would help structure its content, identify missing information and select the corresponding services. The quote could then be generated automatically using pricing rules defined by the business. AI would interpret the need, while price calculations would remain governed by explicit mechanisms.

After acceptance, the information gathered could feed into the delivery file, customer area and project calendar. This continuity could reduce repeated data entry, provided that exchanges between the different tools were properly organised.

Administrative management would offer other possibilities. An assistant could prepare a list of invoices approaching their due date, match payments to customer files or draft payment reminders. Special situations, such as a dispute or a payment agreement, could be excluded from automated processing and submitted for individual review.

A document tool could also facilitate access to internal knowledge. In response to a question, it would search authorised documents for relevant procedures or technical information. The answer would include the references used so that the employee could verify its content. In the absence of sufficient evidence, the system should be able to indicate this limitation rather than present an uncertain answer as an established fact.

Finally, a business leader could have an interface for querying company data without having to prepare every search personally. The tool would examine, for example, projects at risk of exceeding their budget or cases experiencing delays. Results should distinguish recorded information from generated estimates while providing access to the evidence underlying the analysis.

These applications would not be intended for systematic deployment. Their relevance would depend on the difficulties encountered, the quality of available data and the benefits actually observed.

*IRSOY is already ready to explore requests of this kind: designing tailored intelligent tools and integrating them with your business software.*

Present your project →

A shared environment made up of several solutions

The proposed decentralisation could be accompanied by a shared working interface. Employees would access different functions from the same workspace, even though those functions would rely on separate systems.

A business could thus retain its accounting software, reservation service and document storage, then commission the connections needed for certain cross-system uses. A request made in the shared interface would make it possible to search several information sources or prepare an action in the appropriate tool.

AI could facilitate these interactions without having to handle every operation. Calculations, permission rules or the conditions that trigger an action could remain entrusted to mechanisms whose behaviour would be defined in advance.

Such an arrangement could make it possible to replace a component without rebuilding the entire environment. This flexibility would, however, require data exchanges, connections and the responsibilities of each system to have been planned from the design stage.

Decentralisation would then concern primarily the business’s ability to choose and develop its tools. It would not require multiplying interfaces or fragmenting activity monitoring.

Autonomy that would still need to be demonstrated

Developing custom tools would not, on its own, guarantee greater independence. A business could have a solution fully adapted to its operations while remaining heavily dependent on the provider responsible for maintaining it.

Similarly, several separate tools could rely on a common infrastructure or AI model provider. The apparent diversity of applications would therefore not necessarily amount to a diversification of dependencies.

From this perspective, evaluating a project should go beyond its features alone. The ability to retrieve data, transfer maintenance to another team or replace a service would deserve attention. Arrangements for business continuity if an intelligent function became unavailable should also be planned.

Permissions should be defined according to use. A document assistant would not necessarily need access to financial information or personnel files. A tool tasked with preparing reminders would not automatically be authorised to send them. The degree of autonomy granted to each function could be proportionate to the consequences of its actions.

The economic question would remain just as important. For a common need, an existing solution could still be preferable to custom development. Conversely, a particular procedure or repetitive manual operations could justify a tailored integration. This decision should be based on an assessment of design, operating and maintenance costs, as well as the usefulness observed by teams.

Gradual development, starting from precise needs

A first step could be to examine a limited process: handling a sales enquiry, preparing purchases or tracking a job. The business would determine the information needed, the actions permitted and the situations that should remain subject to a human decision.

A trial would then allow results to be assessed under real working conditions. Time spent on checks, possible errors and ease of use would matter as much as the tasks automated. Depending on these observations, the tool could be expanded, corrected or abandoned.

This progression would allow the business to develop its organisation without immediately committing its entire activity to a new system. It could also lead to keeping certain procedures as they are when changing them would not offer sufficient benefit.

Over time, businesses could thus gain greater freedom to adapt their tools as their needs change. Decentralisation would then take the form of an increased ability to assemble their own management environment, with the option to modify one part without systematically reconsidering the whole. AI could contribute to this development, provided that technical and contractual choices effectively preserve that freedom.