Sustainability — Otto Complete
Our Responsibility
Work done well means work done responsibly.
Otto Complete is building a new category of work — AI employees that run real business processes at scale. That opportunity comes with a duty to think carefully about the energy we consume, the people our technology affects, and the judgment we encode into the systems we ship.
A brief statement
We are not a company that will ever claim to have sustainability solved. What we will do is measure what matters, report on it honestly, and keep closing the gap between what we operate today and what responsible scale looks like tomorrow.
Three pillars guide our work.
We organize our sustainability program around the areas where an AI-first company can make the largest difference — and where, if unchecked, it can cause the most harm.
01 — Environmental
Efficient compute, honest accounting.
AI workloads have a real energy cost. We do not pretend otherwise. What we can do is make the compute we consume as clean and as efficient as the state of the art permits, and be transparent about the footprint of the work our agents perform on behalf of clients.
Cloud — We deploy on providers with published renewable-energy roadmaps and region-level carbon data, and we prefer regions with lower grid intensity where latency allows.
Right-sizing — Not every task needs a frontier model. Our orchestration layer routes work to the smallest capable model, which cuts both cost and energy per task.
Office — Our Mississauga operations run on a lease with building-level energy reporting, and we maintain a remote-first posture to reduce commute miles.
Measurement — We are building internal reporting on compute-hours, token volumes, and estimated emissions per agent deployment, with a target of publishing aggregate figures annually.
02 — Social
Taking the workforce question seriously.
We sell software that performs work previously done by people. Treating that as anything other than a significant societal issue would be dishonest. We approach it with care, not marketing language.
Augment first — Our deployment methodology starts by identifying tasks — not roles — that are well-suited to automation, and by designing workflows where AI handles repetitive volume and humans retain judgment.
Transition support — For clients undertaking role-level change, we provide change-management resources and recommend staged rollouts, reskilling pathways, and honest communication with affected teams.
Access — Enterprise-grade automation has historically been available only to the largest employers. Our pricing model is designed to put the same capability within reach of mid-market and small organizations.
Our own team — We hire in Canada, pay at or above local market, and invest in the professional development of everyone who builds Otto.
03 — Governance
Safe systems, clear accountability.
Autonomous agents acting inside client systems can cause real harm if they are poorly built or poorly supervised. Our governance model is designed to make that outcome hard, not easy.
Data handling — Client data is processed under contract-level protections, stored in the client's chosen region where feasible, and retained only as long as operationally required.
Human oversight — Sensitive actions — financial disbursements, customer communications, permanent deletions — require configurable human approval, on by default.
Audit trail — Every agent action is logged in a tamper-evident record that clients can review, export, and use in their own compliance processes.
Bias & fairness — Agents that affect people — screening, scoring, prioritizing — are subject to pre-deployment review and periodic re-evaluation for disparate impact.
"Isn't AI bad for the environment?"
It's the first question we get, and it deserves a direct answer rather than a dodge.
Training frontier models is energy-intensive. Inference — the running cost of the agents we deploy — is substantially less so, and continues to fall as model efficiency improves. But inference at scale is still consumption, and we account for it.
The more interesting question is what the work replaces. A finance agent that closes a book in six hours instead of six days typically displaces travel, overnight server runs, printed reports, and weeks of back-and-forth email. The net picture is rarely obvious in either direction, and we'd rather measure it than guess at it.
Our commitment is simple: publish what we know, acknowledge what we don't, and improve the ratio.
What we are committing to.
Concrete milestones, with dates. Anything less is a press release.
By end of 2026 — Baseline emissions report Publish our first annual sustainability report covering Scope 1 and 2 emissions from corporate operations and an estimated Scope 3 figure for cloud compute attributable to production agents.
By end of 2026 — Renewable-matched compute Migrate all production workloads to cloud regions with a published commitment to 100% renewable matching, and disclose the share of workloads meeting that criterion.
By mid-2027 — Workforce-impact framework Publish the methodology we use with clients to assess workforce impact before deployment, including reskilling guidance and redeployment patterns we've seen work.
Ongoing — Independent review Submit our sustainability claims to third-party review before publication, starting with our first annual report. We would rather be corrected than believed without evidence.
The principles our agents are built under.
Transparency — Clients always know what an agent is doing, why, and on whose behalf. No hidden actions. No opaque reasoning where a decision matters.
Reversibility — Where possible, agent actions are reversible or subject to hold periods. Irreversible actions require explicit human confirmation.
Scope — Agents operate within clearly defined permissions and do not escalate their own access. Scope creep is a bug, not a feature.
Humility — When an agent is uncertain, it escalates to a human rather than guessing. Confidence calibration is part of our quality standard.
Keep us honest
This page will change, and we hope you'll tell us when it should.
Sustainability at an AI company is not a finished topic. If you're a client, partner, employee, or observer with a concern about how we operate — or a suggestion for how we could operate better — we want to hear from you.