• Written by Azhar Aliyar, Director

Can we trust the answers we get from our AI system about procurement, or are we just creating a second source of truth that will diverge from our systems of record? This question is hard to answer because it requires a deep understanding of how the AI system integrates with our existing systems, and how it handles the complexities of procurement data.

Read paths, write paths and why the second needs more care

The integration of AI systems with procurement data involves both read paths and write paths. Read paths are used to extract data from our systems of record, such as the purchase order table or the supplier contract database. Write paths, on the other hand, are used to update these systems with new information, such as changes to contract terms or updates to invoice history. While read paths are relatively straightforward, write paths require more care because they can potentially introduce errors or inconsistencies into our systems of record.

For example, if the AI system updates the supplier contract database with new terms, but these terms are not properly validated or verified, they may not match the actual terms agreed upon with the supplier. This can lead to discrepancies between the AI system's understanding of the contract and the actual contract, which can have serious consequences for procurement decisions.

Reconciling definitions that differ between systems

One of the major challenges in integrating AI systems with procurement data is reconciling definitions that differ between systems. For instance, the definition of a "supplier" may vary between the purchase order table and the supplier contract database. The AI system must be able to reconcile these differences and create a unified understanding of the data, which can be a complex and time-consuming process.

This requires a deep understanding of the data models and definitions used in each system, as well as the ability to map these definitions to a common framework. If the AI system is unable to reconcile these differences, it may produce inconsistent or inaccurate results, which can undermine trust in the system.

The procurement data this depends on

The integration of AI systems with procurement data depends on a range of data sources, including purchase orders, supplier contracts and terms, invoice history, and category structure. The purchase order table, for example, provides a record of all purchases made by the organization, including the supplier, date, and amount. The supplier contract database, on the other hand, contains information about the terms and conditions of each contract, including pricing, delivery schedules, and payment terms.

The invoice history provides a record of all invoices received from suppliers, including the date, amount, and payment status. The category structure, which is used to organize and analyze procurement data, provides a framework for understanding the different types of purchases made by the organization and how they relate to each other. By integrating these data sources, the AI system can provide a comprehensive understanding of procurement activity and identify opportunities for cost savings and process improvements.

What it changes about

The integration of AI systems with procurement data can have a significant impact on procurement decisions and outcomes. For example, by analyzing spend data, the AI system can identify areas where spend is fragmented across multiple suppliers, and provide recommendations for consolidating spend with a single supplier. This can help to reduce costs and improve efficiency.

The AI system can also analyze contract terms and identify areas where they diverge from what is actually paid. This can help to identify opportunities for cost savings and improve the overall effectiveness of procurement processes. By providing a more complete and accurate understanding of procurement data, the AI system can help to inform procurement decisions and drive better outcomes.

Measured with

The effectiveness of the AI system in integrating with procurement data can be measured using a range of metrics, including spend under management, price variance to contract, and supplier concentration. Spend under management, for example, measures the percentage of total spend that is managed through the AI system. Price variance to contract measures the difference between the price paid for a product or service and the price specified in the contract.

Supplier concentration, which measures the percentage of total spend that is concentrated with a single supplier, can help to identify areas where the organization is overly dependent on a single supplier. By tracking these metrics, the organization can gain a better understanding of the effectiveness of the AI system and identify areas for improvement. For more information on how to implement these metrics, visit our operations page.

What this does not do

The integration of AI systems with procurement data is not a replacement for human judgment and expertise. While the AI system can provide valuable insights and recommendations, it is still important for procurement professionals to review and validate the results. Additionally, the AI system is not a substitute for proper procurement processes and procedures, such as reaching out to our team for guidance on implementing best practices.

Furthermore, the AI system is not a magic solution that can solve all procurement problems. It is a tool that can help to inform procurement decisions and drive better outcomes, but it is still dependent on the quality of the data and the effectiveness of the integration. If the data is incomplete or inaccurate, or if the integration is not properly implemented, the AI system will not be able to provide accurate or reliable results.