4 minutes
Companies are hard-coding AI-led efficiency discounts of 10 per cent into contracts where consulting firms use the technology to deliver work faster and with fewer staff.
The move highlights how artificial intelligence is decoupling the traditional “time and motion” billing model as AI allows projects to be completed in months with a fraction of the advisers once needed.
Big four consulting firm PwC is trying to offset the loss of traditional fees by spruiking a shift to “value-based” billing to maintain margins while citing the cost of AI tools and training.
Adam Barty, the managing director of boutique advisory firm Revium, says asking clients to pay full freight despite lower production costs may work for global firms in the short term.
“This is one of the battles we have internally,” Barty said. “Could we charge what we charged before? I think there’s a lot of people who are still charging what they charged before. I think you probably could, if you’ve got a brand name behind you ... I just don’t think there’s any justification for charging the same to do work using AI. It’s just illogical.”
Revium, which has about 40 professionals, builds IT systems and provides AI training and strategy advice for private and public sector clients. The firm mainly uses Anthropic’s Claude and generates “maybe 30 to 40 per cent” of its code using AI, a proportion expected to grow to “more like 70 to 80” per cent by next year.
He described this as a “like a rug pull” because it cuts fees for IT work as fewer professionals produce higher quality work, even as they heavily supervise the outputs for quality.
The firm is also conscious of not using AI for work that requires human judgment and experience.
“We’re using [AI] to help with project management,” Barty said. “We’re using it to help with writing specifications. We’re using it to help with stakeholder interviews. But the two areas where we’re still very human-focused are oversight and strategy.”
Clients are seizing efficiency gains. Barty cited a 2025 project building a bespoke IT system worth more than $1 million.
“We had one very specific example where we got to the final phase of the contract and we just had to go through a tick-box exercise to say whether we’d use AI,” he said. “And if we ticked that box to say we used AI, then it stated that we had to apply a 10 per cent discount to the price of the contract immediately as a result of using it, because of the efficiencies.”
Other clients are also asking for discounts. It’s “not as overt as contract terms, but [the] flow is usually ‘are you using AI?’, if yes, then it moves straight to pushing for reduced cost.”
In March, the firm pitched to redevelop a website for a listed construction company that Revium first delivered six years ago at a 40 per cent discount to the original work.
“The quote we sent through was a little over half the cost of what the original build was,” he said. “And this is a complete rebuild. So it’s basically doing the same sort of project again, but six years later our price was 60 per cent of the original cost to do it the last time.”
The client’s board initially “pushed back” on the quote “worried that they weren’t getting a high-quality output like the first time because the price was so much lower. But the reality is this is what AI allows you to do.”
He cited a third example where the firm won work to develop a three-year AI strategy for a private logistics company, beating out a big four firm.
“They wanted to have a heap of stakeholder interviews,” Barty said. “[But the] budgets didn’t really stretch to the number they wanted. So we did all of the C-suite [interviews] face-to-face with the usual process. But then we did about another 20 people, where we had an AI business analyst do the stakeholder interviews.“
The AI agent conducted a 20- to 30-minute interview, asking questions, follow-ups and prompting for more detail as required. It then summarised the interviews for input into the strategy.
The cost was the same “as about three to four face-to-face” interviews.
“The feedback from all of the people involved was really positive,” Barty said. “They could do it on their own time when they wanted to and they had no issues with talking to an AI that sounds very much like a human.”