The use of technologies based on AI systems has significant consequences in terms of outsourcing agreements, which must be duly taken into account by suppliers and customers at the negotiation stage and during their whole business relationship. Suppliers and customers should assess from a new perspective a number of key contractual clauses, from which the good and profitable success of outsourcing and the reduction of the risk of future litigation very much depend.

The automation of corporate processes and functions, either core or otherwise, such as ICT infrastructure, invoice management or CRM is certainly nothing new, both within the companies and when they decide to resort to external independent suppliers. What it is new is instead the approach and the technologies employed: if in the very recent past the trend was to try to improve such processes moving from the analysis of the processes themselves, by weighing their limits and inefficiencies, the privileged observation point has now become the analysis of the data. The data – if properly interpreted and processed by AI algorithms – is able to provide valuable hints on how to direct corporate processes with the aim of increasing their efficiency and reducing their costs.

Massive amounts of data can be endlessly produced through very different sources (Big Data) thanks to the significant technological progress in terms of computational power of the information systems and storage capacity. The use of advanced analytics and machine learning technologies (the so-called, “Artificial Intelligence” or “AI”) on vast amounts of data makes it now possible to obtain extraordinary advantages also when corporate functions and processes are outsourced, outstripping the capacities of traditional Robotic Process Automation (RPA) systems. AI algorithms are in fact capable of “learning” and modifying themselves with a certain degree of autonomy depending on the data they are required to process, revealing therefore new patterns and trends.

#1 Technological evolution and benchmarking clauses

The rapid pace of innovation of the AI technologies and the increasing autonomy of algorithms may lead to developments that are useful and, at the same time, unexpected. Suppliers and clients must therefore envisage adequate contractual mechanisms that enable them to set out in advance how to address the introduction, if any, of relevant improvements in the AI technologies under their agreement or unexpected developments particularly valuable for the client, starting from the adjustment of the consideration owed to the supplier, through the so-called “benchmarking clauses”. The issue is worth of being carefully considered given that outsourcing agreements have usually a medium-long term duration and stem from complicated and lengthy negotiations. In addition, the technological lock-in that in most cases exist between supplier and client prevents the latter from acquiring, during the contractual relationship, innovative technologies from other suppliers. This is why modifying the agreement’s terms and conditions a posteriori may be the most burdensome and less appropriate choice. Whenever the parties are able to derive a mutual advantage from the improvement of the technologies employed, they will also have the possibility to set up a mutual innovation fund with shared governance, setting in advance administrative and control rules.

#2 Liability and limitations

The increasing autonomy of AI systems in processing data has to be considered also in terms of contractual liability and relevant limitations. To this end, see for further details one of our latest posts on this blog.

First of all, based on the principle of transparency, supplier should duly inform customer on the capacity and limits of the AI systems used, allowing customer to make an aware decision and reduce the risk, for supplier, to be held liable for pre-contractual liability (i.e., the fact of having conducted the negotiations in bad faith, without having disclosed material information).

Supplier shall also negotiate with customer appropriate limitations of liability to the extent permitted by law, taking into account, inter alia, the degree of autonomy of the AI systems used, the fact that they process data the source of which may go beyond supplier’s control, and the need of human intervention for the correct functioning of the system.

Caps setting out a compensation limit by supplier for any damages occurred in the performance of the agreement, shall be proportionate to the specific event from which the compensation claim arises.

#3 Know-how transfer, insourcing and supplier replacement

During the negotiations, client and supplier should also carefully consider the transfer of know-how from supplier to customer at the expiry/termination of their contractual relationship, that usually makes it possible for customer either to (re)insource the previously outsourced process and carry out it independently, or outsource it to a new and different supplier.

This may be achieved by including in the agreement the so-called “termination assistance” clauses, under which supplier undertakes to assist customer in the termination phase of the agreement and to transfer to the latter all the know-how necessary to enable it to continue the service independently or through third parties. When it comes to AI technologies, it is essential to provide for specific training courses addressed to the employees and collaborators of customer held by supplier’s specialized personnel, including, inter alia, algorithmic “training”, explanation, and maintenance.

In view of a future insourcing or supplier replacement, the client shall obviously be more minded to rely on suppliers which exploit third-party, licensed AI technologies, rather than suppliers using proprietary technologies, to avoid situations of technological lock-in which might prove difficult to overcome.

Don’t miss the second part of our hints & tips, next week on this blog! #4 Intellectual Property, #5 Data protection, and #6 Cyber risk and Insurance.