How AI is being used in tendering

More than two-thirds (68%) of the contractors who responded to RLB’s survey said they are using AI for support in writing qualitative responses to tenders. A further 8% of respondents use AI to generate first drafts.

The more detailed and quantitative the work, the less likely contractors are to use AI. For example, a total of 43% use AI in some way to identify risks in tender documents, while just 21% are using it for programme optimisation. Only 13% are using AI for pricing and subcontractor evaluation.

Paul Beeston, RLB partner and head of industry and service insight, offered this advice to those overseeing procurement: “The mainstream use of AI in tender responses suggests procurement teams should now assume that some bid content may be machine-assisted, particularly in qualitative answers. That does not make the content inherently weaker, but it does increase the importance of designing tender questions that test project-specific understanding, delivery capability and evidence rather than generic narrative.

“In practical terms, procurement decisions should place more weight on proof points, examples and interview or clarification stages where bidders can demonstrate how their written responses connect to the actual team, methodology and risks of the project. As AI use expands, evaluation needs to focus less on polish alone and more on credibility, relevance and verifiable substance.”

Furthermore, Beeston warned: “For clients, greater use of AI creates both an opportunity and a risk in the outcomes they secure. Used well, it may help bidders respond more efficiently and consistently, potentially improving the quality and completeness of submissions. Used poorly, it can widen the gap between what is written in the tender and what the project team can genuinely deliver in practice.

“The result is that client outcomes will depend increasingly on the quality of due diligence applied after responses are received. Where clients test claims properly and probe delivery evidence, they are more likely to secure credible commitments. Where they do not, there is a greater risk of selecting bids that read well, but translate less effectively into delivery performance.”