This blog was last updated on September 28, 2026

Sustainability at Sovos has never been a side program or a compliance checkbox. It’s embedded in how we operate and grow as responsible global corporate citizens. 

As AI became a strategic imperative for every technology company over the last few years, that meant something specific for us: responsible AI has become integrated into the discipline we’ve been building year-over-year. And I’m proud to say it shows up in how we empower our people, how we cultivate trust, and how we grow responsibly. 

Of course, this year was about more than AI. Our employee engagement survey drew 86% participation, well above the 75% industry benchmark. Employee participation in community impact tripled year-over-year. We also improved our carbon footprint data quality by 62% through deeper vendor engagement, and our EcoVadis sustainability score improved by 35%. But our work on responsible AI demonstrates the same discipline embedded in something new. 

Environmental Impact of AI: Building a Baseline Our Vendors Couldn’t Give Us 

AI’s environmental cost is a hot topic across the industry, and most companies using these tools have no clear picture of the impact of their own usage, because most vendors don’t disclose the underlying energy or carbon data themselves. 

The number is small, but we now have a credible baseline in place before our use and footprint grows further. That baseline is already shaping how we think about efficiency, and as agentic and higher-usage tools become more common, that’s the kind of data we’ll need to manage our footprint deliberately instead of guessing. 

Governance Built to Back That Up 

A baseline like that doesn’t happen without discipline behind it. Our AI Policy governs every implementation we ship or use internally: high-impact and customer-facing AI gets full review before deployment, lower-risk tools move faster, and oversight sits with a cross-functional team spanning security, legal, privacy, and technology. The policy is reviewed at least twice a year, because AI capability and regulation move too fast for a policy to be written once and left alone. 

That discipline shows up in the numbers: we achieved a 100% Responsible AI Readiness score on the Hg Sustainability Assessment. That score shows policy, practices, and culture actually working together. 

This matters particularly as AI reshapes how we build products and develop our teams. Our role is to ensure that as we accelerate adoption, we do so in a way that earns confidence rather than erodes it. We hold ourselves to that standard elsewhere: we process more than 70 billion transactions a year, with 99.9%+ accuracy across every regulatory filing. We’re applying that same bar to how we govern AI. 

Responsible AI in Practices: People Making This Work 

Policy and baselines only matter if they reach how people actually use the tools, every day, across a workforce of 2,500+ people in 25+ countries. That reach shows up in how connected people feel to where Sovos is headed. Our employee engagement is up at a time of declining interest across the industry in general. AI enablement was one of the things behind that shift, not separate from it. 

This year’s Responsible Prompting Guide reached over 1,000 employees through live training, with 100% of the workforce completing formal attestation. Enablement reached every level: company-wide Copilot use grew more than 40%, 72% of employees said their manager actively models responsible AI use, and Engineering, our most mature example, saw throughput rise 62% alongside a 90% AI confidence rate. 

Rachel Roelich-Wells, a Senior Services Partnerships Representative based in the UK, is a good example of what that looks like for one person. AI initially felt like one more thing to juggle, until an International Women’s Day panel on AI inspired her to jump, and she carried that into our Copilot Prompt Challenge, staying candid about where the tool overstated things, not just where it helped. Employee confidence using AI grew 7 percentage points between October 2025 and April 2026. Rachel is one of the reasons that number moved. 

Multiply that across 2,500 people figuring out the same thing in their own way, and you start to see what “responsible AI” looks like in practice, at scale. 

Where We’re Still Working it Out 

AI is moving fast enough that no approach to it stays right for long. Ours won’t either. How we measure it, how we govern it, how we account for its environmental impact, all of it will keep changing as technology does. Transparency about that is part of what a credible sustainability program looks like. 

To us, how we close those gaps matters more than how quickly we close them. At Sovos, sustainable impact comes from embedding the work into the business so thoroughly that it becomes how the business operates. 

Trust is a Standard We Hold Ourselves to Every Year  

On the trust front, we maintained a standard I consider non-negotiable: 100% of our employees completed annual training in code of conduct, cybersecurity, data privacy, and – for the first time formalized at this scale – responsible AI. We achieved a 100% Responsible AI Readiness score on the Hg Capital (our primary investor) Sustainability Assessment.  

This matters to me particularly as AI reshapes how we build products and develop our teams. Responsible AI governance isn’t a one-time policy document. It will evolve alongside the technology itself, and our role is to ensure that as we accelerate adoption, we do so in a way that earns confidence rather than erodes it. Our customers process billions of transactions through our platforms each year. We experienced 84.9 million in the Black Friday/Cyber Monday window alone, with 100% system uptime. That kind of reliability is trust made operational. The same standard should apply to how we govern AI.  

Going Deeper on Environmental Accountability  

Our environmental work this year was less about expanding scope and more about improving credibility. We improved our carbon footprint data quality by 62%, not by collecting more data, but by working more closely with vendors to move from cost-based estimates to activity-based measurement.  

Print is a good example of what that looks like in practice. The millions of tax forms we print and mail on behalf of customers each year make it one of our largest vendor-emissions categories. Working alongside our print management team and our carbon accounting partner, we engaged with our largest vendors to track actual paper usage and weight rather than relying on financial proxies. It’s slower, more detailed work, but it builds a more honest foundation for decarbonization.  

We also made our first attempt at something I think will matter increasingly: estimating the carbon footprint of our AI tools. In partnership with Sustainable AI Group, we placed FY26 AI-related emissions from our three most-deployed tools at 46.73 tCO2e – less than 0.3% of our total footprint today. The number is small. The methodology matters more, because AI’s share of corporate emissions will grow, and organizations that get ahead of measurement now will be better positioned to manage and reduce it.  

Externally, the trajectory is encouraging as our Hg Sustainability Diagnostic score improved 14% year-over-year, ranking us first among all US-headquartered companies in Hg’s portfolio.  

Honest About What We Haven’t Yet Solved  

I want to be direct about where we’re still maturing, because I think that honesty is part of what a credible sustainability program looks like.  

Our approach to measuring the carbon footprint of digital infrastructure – including AI – is still evolving. Our framework for accounting for climate-related financial risk is still becoming a fuller, more embedded part of how we plan. These aren’t gaps we’re papering over; they’re the next layer of work, and we’ve formalized our commitment to SBTi with science-based emissions reduction targets to be submitted by April 2027.  

To us, how we close those gaps matters more than how quickly we close them. Because sustainable impact doesn’t come from fast announcements, it comes from embedding the work into the business so thoroughly that it becomes how the business operates.  

That’s what we’re building toward. And I’m proud of how much of it is already real.  

The full Sovos FY26 Sustainability Report is available at here.