A navigation companion for blind Australians — helping you move through the world

The problem

Blind and low-vision Australians have excellent tools for getting around — the long cane, the guide dog, and Orientation & Mobility (O&M) training. Around those tools there is a persistent, unfilled gap.

Today’s AI vision apps answer “what’s in this photo?” — a static frame held at arm’s length. Almost none help with the moving problem: walking somewhere, tracking the ground ahead, warning about a step or a kerb edge as you move.

The dedicated devices that do help with movement cost roughly A$1,500–A$11,800 — out of reach for most people. There are roughly 575,000 blind or low-vision Australians (Vision Australia 2024 — figure will be re-verified before each external use), and a guide dog reaches only a small fraction. The single most consequential daily task — getting from A to B on foot — is the least well served by technology.

Our thesis: the iPhone most blind people already carry can do far more than it does today. Reach and affordability are the whole game, and “no extra hardware required” is a structural advantage no A$10k device can match.

What we’re building — and where we are honestly

The companion is an on-device iPhone app that sees and describes (scenes, text, objects, colours, familiar faces) and, in development, helps you get from A to B with careful awareness of the path in front of you.

How we build — safety discipline, not shortcuts

For a blind-navigation tool, how you build is as important as what you build. The rules below apply right now, in the development build — not later, when people are walking.

• One internal and one external adversarial safety review, completed before any blind person walked with the navigation layer. Both examined the real built code, not a spec. Both surfaced genuine blocking issues — finding them on the bench, not on a footpath, is the process working.
• Staged validation under O&M supervision, with HREC ethics approval, before any unsupervised solo use. The sequence is fixed: closed course with a mobility specialist first; public footpaths only after pre-registered safety metrics are met and signed off.
• A device-free automated test harness. Perception behaviour is codified in a growing automated test suite (700+ checks as of July 2026). Tests run without a phone attached — so the safety specification is machine-verified, not reliant on a human remembering to check it.
• Hedged-speech design, by engineering. The system is built never to over-claim sensor confidence. Every spoken cue hedges uncertainty by design — the language engine is engineered to step back when it is unsure, so a user is never misled about what the system can and cannot see.

These governance choices were made before any funding was secured, and before any blind person walked with the system. They are not compliance theatre — they are the minimum bar a system you walk behind must clear.

Why we believe it’s real

• A working, tested perception engine — not a pitch deck. The real-time perception core is built in Swift and passes its automated unit-test suite. This is meaningful evidence that the project is genuine engineering at MVP stage, not a concept.
• Adversarial safety reviews completed before any field walk. One internal and one external review — both examined the real built code and both surfaced genuine blocking issues. Detail in the ‘How we build’ section above.
• Grounded in the guide-dog training curriculum. We mapped trained guide-dog behaviours against what the app does and where it falls short, and modelled intelligent disobedience — the trained refusal to follow a command into danger. Assist, never lead blindly.
• Blind-founded, with real Australian engineering pedigree. The founder, Joe Webber, is a blind developer with 16+ years of production delivery across government, finance and payments — including a Reserve Bank of Australia integration that processed over $1 billion and the accessibility-mandated National Redress Scheme survivor portal. The person setting the safety rules is the person who will walk behind them. LinkedIn →
• The approach is validated in published accessibility research. Real-time street-level AI description for blind navigation has been demonstrated in peer-reviewed accessibility research (CHI 2026). We are building on an approach that the research community has independently validated — not a theoretical concept.

The honest competitive position: Dedicated devices that help blind people with movement exist and ship today — at roughly A$1,500–A$11,800. Nobody ships this as software on the phone a blind person already owns, without extra hardware. Converting that open niche into a validated, affordable product for 575,000 Australians is exactly what this funding buys.

Why it matters — the impact case

Reach and affordability are the whole game. A guide dog reaches fewer than roughly 10,000 handlers nationally, against ∼575,000 blind or low-vision Australians. Dedicated mobility hardware costs A$1,500–A$11,800. Our companion runs on the iPhone the person already owns. For a mission-driven funder trying to reach the large majority a guide dog never reaches, this is the difference between a niche device and genuine accessibility.

An Australian-owned, public-good asset. The ambition is not just a product — it’s an Australian dataset and capability for blind navigation: validated locally in Adelaide, then Australia-wide, built for our footpaths, our crossings and our conditions. A durable public good, not a foreign-owned black box.

Community-owned, co-designed, safety-first. The goal is to build this with the blind and low-vision community, with O&M specialists setting the validation bar, and with blind organisations as partners — not customers. Our longer-term vision is a mixed community — sighted and blind people together, in the spirit of Be My Eyes — where the platform’s accessibility strengths benefit everyone.

The ask — what funding does

Funding does three things: (1) keep the founder building full-time, (2) buy the safety validation that can’t be done solo, and (3) complete and ship the designed layers.

We do not publish detailed line-item budgets on this page — those go in the pitch pack shared in conversation, so you can see the exact numbers with context. This page shows what the money does; the pitch pack shows the numbers.

What your funding produces — and how you’d verify it

We propose milestone-released funding: money moves only when the previous proof-point is inspectable. The table below shows what each funded stage produces and how a funder or partner can verify progress.

All figures are illustrative ranges. Detailed line-item budgets, pre-registered metric values and exact acceptance criteria go in the pitch pack shared in direct conversation — not on this public page (disclosure discipline). The structure above reflects the real milestone sequence; the numbers are illustrative until agreed with Joe before any application.

How to partner, support or advocate

We want partners, not just funders. There are four honest ways to help:

Start a conversation

If you’re interested in funding, advocating, partnering or hosting a pilot, use the form below — it’s the fastest path to the pitch pack, a call, or a live demo. We aim to reply within two business days. You can also email directly: joseph.webber@me.com.

Email for the pitch pack (pre-filled) →