Consent is a feature, not a setting.
The agent does nothing it has not told you it is about to do. You can pause, redirect, or reverse at any point. What the agent does is logged plainly and readable in your own language.
EasyAccess is a voice-first agent that operates a computer on your behalf. Ask it to open an app, read a page aloud, draft a document, fill a form — it hears you, acts, and narrates what it is doing. Built for people for whom the mouse and keyboard are friction, not a tool.
The cockpit
Every EasyAccess session opens in a small cockpit window. You see your own voice, the agent's transcript, and a running account of what it is doing — each step visible before it happens, each step logged after. Nothing silent, nothing invisible.
Can you pull up the AODA compliance checklist and fill out the organisation section for me?
I'll open the checklist in your browser, then fill the organisation fields with what I have on record. I'll pause before anything I'm not sure about.
Activity
What it does
EasyAccess deliberately does less than a full-stack AI assistant and more than a voice shortcut. It is a hand on the keyboard — one that listens, waits, and explains itself.
You talk to EasyAccess the way you'd talk to a capable assistant sitting next to you. No command syntax, no voice-menu tree. English to start, with French coming alongside pilot deployments.
Opens apps, clicks links, fills fields, reads pages aloud, drafts short documents. It works across the system the way a person does — not inside a single walled garden.
You always hear what the agent is about to do before it does it, and what it did after. No silent actions. You can interrupt or correct at any point.
A position
Consent is a feature, not a setting. The agent does nothing it has not told you it is about to do — and afterwards, nothing it cannot explain in plain language.
EasyAccess principle № 01
Principles
Assistive technology lives or dies on trust. These are the stances we are prepared to defend in front of a pilot partner or a regulator.
The agent does nothing it has not told you it is about to do. You can pause, redirect, or reverse at any point. What the agent does is logged plainly and readable in your own language.
No celebratory sound effects, no confetti, no emoji. Interfaces for assistive use should be quiet. The accessibility audience tells us this, repeatedly; we listen.
Canadian pilot deployments get both official languages on day one. Additional languages follow real deployments, not a marketing page.
The agent is not a safety-critical device. It will not file your taxes, approve a purchase, or act on a regulated instrument. Where it stops, we tell you in advance, not after.
What a pilot looks like
We publish the pilot arc in advance because nobody should sign a six-month contract wondering what month three looks like. Here is the rough shape; every deployment adjusts the edges around it.
We meet your team, map the accessibility obligations on your plate, and decide which five to twenty users go first.
Device provisioning, voice calibration, and in-person or remote training with the first cohort of users.
Daily use begins. A named contact on our end is available for questions and adjustments. Weekly check-ins.
Pattern of use stabilises. We tune the command vocabulary and agent behaviours to what your team actually asks.
Optional: broaden the pilot to an additional cohort. Second-language or additional role coverage reviewed here.
Findings report with usage data, qualitative interviews, and a plain-language summary. Annual contract or exit.
Writing
We write about AODA, the Accessible Canada Act, and where voice-first technology helps (and where it does not). Working notes from a team building accessible software, with no hype and no inspiration framing.
Pilot
We are running a small number of six-month pilots through 2026 with Canadian nonprofits, municipalities, and service providers with AODA obligations. Pilots include onboarding, in-person training, and a named contact for the duration.