Microsoft Projects $1.6 Million in Annual Savings with Relativity aiR

How did they do it?

• Sought a fast, accurate, defensible generative AI solution to improve review efficiency and effectiveness.
• Tested Relativity aiR for Review and aiR for Privilege against previously reviewed data sets to quantify aiR’s efficiency and speed over current workflows.
• Built a model that projected annual cost savings of $1.6 million from leveraging aiR vs. traditional human review.

When Microsoft’s litigation group set out to optimize the efficiency and consistency of its document review process, they weren’t just looking for incremental improvements – they envisioned a faster, more accurate solution with a transformative impact.

Targeting both quantifiable gains and qualitative improvements for their team on the ground, Microsoft saw an opportunity to harness the power of generative AI to amplify their human expertise and achieve these outcomes.

To bring this vision to life, Microsoft partnered with Lighthouse, a leader in e-discovery and information governance solutions. Together, they ran Relativity aiR for Review and aiR for Privilege on previously reviewed data sets to validate aiR’s performance against the prior manual workflows – establishing a quantifiable point of comparison between aiR-powered and manual review. The team then used this data to build a model to project annual cost savings gained by using aiR over traditional review.

The results were staggering. When compared against Microsoft’s spend on manual review, aiR's speed and efficiency yielded projected annual cost savings of $1.6 million, and a 45-60 percent reduction in overall review costs. Plus, aiR’s automation of slow, manual processes meant more time for Microsoft’s team – and their outside counsel – to focus on more strategic, higher-value work.

Augmenting First-Level Review with AI

For E.J. Bastien, senior director, discovery programs in Microsoft’s litigation group, the objectives of embracing aiR were clear.

“We wanted to get through projects faster, increase consistency of decisions, and save money,” E.J. said. “When outside counsel is looking at data, we want to make sure it’s the right data.”

Historically, achieving those goals meant leaning heavily on contract attorneys and outside counsel – an expensive and often slow approach.

Microsoft wanted to elevate this work rather than replace it. They sought a tool that accelerated review while keeping humans in the loop to ensure defensible oversight.

Relativity aiR was the clear solution.

Showcasing the Power of aiR

To measure how AI could transform their document and privilege review workflows, Microsoft put aiR into action with helpful guidance from Lighthouse.

“When introducing new technology, we walk our clients through each and every phase so they not only see what’s happening on the surface, but also what’s happening behind the scenes,” said Cassie Blum, senior director, review consulting and solutions, at Lighthouse. “Product adoption is more than knowing how to push buttons – it’s about understanding how the technology fits within each client’s unique program.”

Working together, the teams created a plan for Relativity aiR. They started by running the solution on previously reviewed data sets, comparing AI predictions directly to human decisions. The results were compelling: aiR’s predictions matched human review decisions 92% of the time. And in many instances, it did something even more valuable: it flagged relevant material that human reviewers had missed.

“We assumed the humans would be right more often, but aiR has convinced us otherwise,” said E.J. He also noted that aiR can clearly explain its reasoning, helping the team validate its decisions – a key element of defensibility.

While running aiR on these data sets, Microsoft also found that documents that aiR for Review predicted as relevant were able to move more quickly through the review cycle than they had during the initial, manual review, sometimes skipping traditional first-level review altogether. This acceleration would reduce the time outside counsel would need to spend on document review – allowing them to focus on the highest-value material and drastically cutting costs compared to other review methods.

The same merits proved true for aiR for Privilege. Side-by-side comparisons revealed instances where the prior human reviewers had missed privileged content. aiR for Privilege not only caught these items but also provided valuable reasoning and context to support more nuanced calls.

"Privilege logs are time-consuming and expensive for Microsoft. aiR for Privilege’s predictions and considerations are incredibly helpful, especially when working through those tougher, more nuanced privilege calls."
Kayleigh Klinzman, Program Manager, Microsoft Litigation Group

Measurable Outcomes, Material Success

When Microsoft’s team sat down to compare outcomes, aiR’s speed, accuracy, and agility drastically outpaced their previous workflows. aiR flew through data sets faster and more accurately than human reviewers. With no breaks. No hesitation. Just results.

Then came the numbers. To quantify aiR’s impact, Microsoft’s litigation team built a model to compare traditional review costs with their aiR-powered process.

• Projected annual savings of $1.6 million
• 45-60% reduction in overall review costs
• Accuracy exceeding typical human review – while adding transparency and defensibility

The numbers confirmed: aiR transformed first-pass and privilege review, turning time-consuming, labor-intensive work into a faster, more cost-effective operation. And with aiR for Review and aiR for Privilege integrated right into RelativityOne, these savings and efficiencies have become a standard part of Microsoft’s workflow – not a one-off win.

The success was clear: Relativity aiR is a transformative powerhouse for Microsoft’s legal team.

Building the Future of Legal AI

Looking ahead, Microsoft’s approach to AI is grounded in two principles: defensibility and continuous learning.

“Defensibility is about not ceding everything to AI – it’s about keeping humans in the loop and documenting the process,” E.J. said. “As long as you’re thoughtful, defensibility actually becomes easier.”

Additionally, Microsoft’s prioritization of ongoing education keeps their team at the cutting edge of AI technology, yielding smarter strategy and more efficient and creative workflows.

By embracing Relativity aiR, Microsoft and Lighthouse have created a blueprint for how legal teams can deploy generative AI responsibly. They’ve shown that with the right guardrails and oversight, AI doesn’t just make legal review faster – it makes it better.

And with less strain on resources and more time to focus on strategy and innovation, aiR is helping Microsoft build a bolder, more efficient future for legal excellence.