How Gravity AI Helps Small Teams Make a Big Impact

Please introduce yourselves.

Saleh: I'm Saleh ElHattab, Co-Founder and CEO of Gravity. We built Gravity because we kept seeing the same problem: companies consistently suffer through a painful, time-intensive compliance chore and conclude the exercise with a static report rather than operational wins and cost savings. 

Behind every ton of emissions is a dollar spent, and we wanted to change the script so that the measurement exercise was no longer onerous and yielded positive business outcomes—we help companies save time and money.

Ted: I'm Ted Kornish, Co-Founder and CTO. When Saleh and I started to dig into the problem, it was obvious that data collection was the bottleneck and that teams spent almost all of their time on reporting rather than taking action to reduce emissions and energy costs. My job is to make sure our software handles the heavy lifting, rather than our customers' teams.

How is AI helping Gravity cut through that data complexity and bring order to fragmented data from utilities, fuel and suppliers?

Saleh: The data problem in this space is genuinely complex: thousands of utility providers, hundreds of fuel suppliers, invoices in every format imaginable, and a multitude of facilities, assets and suppliers consuming that energy. 

What AI lets us do is read, normalise and act on that data at a scale that would otherwise be impossible or too expensive. We've consistently harnessed advancements in foundation models to automate manual tasks and package up expertise that typically required hiring large teams or consultancies. The result is compliance and tangible action.

Ted: The more interesting shift is what's happening with AI agents. What we're seeing in the broader AI landscape is that the scale of what an agent can handle effectively doubles every seven months, and that’s showing up directly in what we build. 

The question for sustainability teams isn't whether to use AI for a specific task —it’s how to delegate progressively larger workflows to agents that can process data, do complex analysis, and produce outputs that used to require a full team. That's the direction we're building toward, and it's moving faster than most people expect.

How do you see automation changing the day-to-day reality for the many sustainability teams still relying on spreadsheets and PDFs?

Saleh: The spreadsheet and PDF situation is really a symptom of a bigger problem. Teams are spending most of their time collecting data instead of doing anything with it. 

When you automate ingestion, you don't just save hours; you fundamentally change what the job looks like. Suddenly the sustainability manager isn't chasing down last month's utility bills. They're flagging inefficiencies, focusing on action and having conversations with the COO and CFO about real projects and real savings. That's a much better use of everyone's time —and an enduring benefit.

What does continuous, audit-ready reporting look like in practice and how does AI enable that level of standardisation and reliability?

Ted: Audit-ready reporting means your data has a clear chain of custody. You know where every number came from, how it was processed and you can trace it back to its source. AI makes that possible at scale by automating the ingestion and normalisation steps in a consistent, documented way, rather than relying on someone doing it manually in a spreadsheet each quarter. 

Critically, every change AI makes to data is logged so customers can always see what was modified, when and on what basis, which is what makes it defensible in an audit. In practice, that means your data is always current, traceable and you're not scrambling at year-end to pull everything together.

How close are we to having “virtual colleagues” that can run analyses and even draft full sustainability reports?

Ted: Closer than most people realise. We're already past the point where AI is only useful for discrete tasks: a chatbot here, a feature there. The shift now is toward agents that take on entire workflows end to end, and at Gravity we're already running agents that automatically ingest and reconcile data across hundreds of sites, which is exactly the kind of multi-step work that points toward what comes next. 

A sustainability team asking 'draft our annual emissions report based on this year's data' is a reasonable prompt today, not a future aspiration. Teams need to ask software providers about the breadth of tasks their AI can handle: the new standard is that AI can see and act on everything in your software, not just implement specific features.

For sustainability teams building out their digital strategy, what kinds of workflows should they start thinking about delegating to AI agents now?

Saleh: The question every team should be asking is: “what are we doing repeatedly that doesn't actually require human judgment?” Data collection is an obvious one, pulling utility bills, reconciling invoices, updating the emissions inventory. But it goes further. Variance analysis, flagging anomalies, drafting the first pass of a regulatory disclosure: all of that is delegatable today. The teams that pull ahead are the ones treating AI as a core capability, not a one-off feature. What can we hand off, and how do we keep expanding that list?

What needs to change to truly support AI-first, agent-driven workflows?

Ted: Legacy sustainability platforms were built around workflows designed for human beings. You log in, navigate screens, fill in forms and click submit. That works fine if the primary user is always a person. But agents don't operate that way: they need to read data programmatically, make decisions and write results back, often across long multi-step tasks. 

Most systems built before a few years ago simply weren't designed with that in mind, because they couldn't have been. You can bolt AI features onto those systems, but if the underlying architecture was built for human hand-holding, you're working against the grain of your own infrastructure.

How is Gravity architectured differently from older systems?

Ted: We made two deliberate architectural decisions early on that turned out to be critical. First, we built the entire system to be API-first, which means every capability is accessible programmatically rather than only through a browser. Second, we use a database architecture that lets us “dry run” changes to customer data to see what would happen if we actually made the change. 

What that means in practice is that an agent reasoning across a customer's full emissions picture, across sites, sources and time periods, has everything it needs in one place. It can execute complex sets of changes—the kind a senior consultant might implement - until it gets the task exactly correct. That foundation is what makes Gravity a genuinely good environment for agents to operate in, rather than just a platform with AI features added as an afterthought.

How does AI unlock operational efficiency and even cost savings, rather than just compliance benefits?

Saleh: The core insight driving everything at Gravity is that there's no such thing as sustainability data in isolation. It's operational and financial data. The electricity, fuel, goods and services you track for an emissions report are the exact same data that tells you where you're spending money. 

There's roughly a trillion dollars in inefficient energy spend every year, and a lot of it is hiding in plain sight in the data companies already have. AI makes it practical to surface those inefficiencies at scale, across hundreds of sites and dozens of energy sources, without an army of sustainability or energy efficiency consultants. 

We go further, using AI to model the cash flow of the opportunities that shake out of that exercise and generate a vendor-matched project that translates the exercise into improvements to your bottom line - power factor corrections with capacitors, demand response contracts with your utility, energy generation and storage systems.

How can AI provide meaningful leverage in helping small teams scale their impact without adding headcount?

Saleh: This question hits on the promise of recent AI advancements: you are no longer constrained by the size of your team or the number of hours you have in a day. A small team can replicate the expertise and scale of the most advanced sustainability and energy optimisation organisations. 

We've seen this promise delivered across our customer base, with many 1- or 2-person teams managing massive global operations, like Aadhar Kulshrestha and Meredith Smith at TTI, Inc. or Ian Pope at McCarthy Building Companies.

As AI adoption accelerates, how does Gravity think about the technology’s own sustainability footprint?

Ted: It's a fair question and one we take seriously. The compute costs of running AI models are real, and we're deliberate about where we deploy them, focusing on workflows where the value delivered meaningfully outweighs the energy cost of the models. 

The broader point is that the emissions reductions we enable across our customer base far outweigh the compute required to run our models—but that’s not a license to be careless. We're building efficiency into how we use AI, the same way we help our customers build efficiency into their energy use.