Darvy

Voice-powered AI computer use, built for the blind.

Ameen Darwish  ·  Founder & CEO
Darwish.Ameen@gmail.com  ·  +972 54 552 0777
Pre-seed · 2026
The problem

Blind users are forced to think like sighted users
navigating menus.

State rehab programs spend multi-week training cycles teaching keyboard shortcuts for Outlook. Then another cycle for every new app. Existing tools don't scale to how people actually work.

  • 50M blind globally · 250M visually impaired
  • 10 years since screen readers fundamentally changed
  • $1,475 JAWS Professional license. Describes pixels, not tasks
What Darvy does

Users say what they want.
Darvy executes it end-to-end.

Read me my unread emails

Shop for gaming headphones on Amazon under $100

Schedule a call with Sarah next Tuesday afternoon

Write me an intro about VCs in 2026 and save it to a doc

Today: email, calendar, browsing, files, Word, Google Workspace.

The bet isn't equality with sighted users. It's making blind users faster at voice-driven multi-step work, while they're still clicking menus.

Market

The buying infrastructure already exists.

  • 50M blind, 250M visually impaired globally
  • $1.3B → $2.8B screen reader market by 2032
  • $1,475 JAWS Professional license, paid by employers, universities, and state vocational rehab under ADA & Section 508

Institutions have line-item budgets. We're not inventing demand. We replace a more expensive, worse incumbent.

1% of 50M blind users at $50/mo = $300M ARR. Adjacent (visually impaired, motor-impaired, elderly): same product, much larger TAM.

Competition

Three categories. None do voice-first computer use for blind users.

Legacy screen readers

JAWS (41% share, $1,475)
NVDA (free)
VoiceOver (free)

Keyboard-shortcut interfaces. Describe pixels. Don't do tasks.

Adjacent assistive AI

Be My Eyes · Aira
Seeing AI · Envision

Visual interpretation only. None target full computer use.

General computer use

Claude Computer Use
OpenAI Operator

Screenshot-based. 30K tokens. 30-60s. Not voice-first. Not built for blind users.

Why we win

The bar for assistive tech is autonomy,
not partial help.

  • Tools that promise "equality" can be humiliating when they fail. Anything less than full autonomy creates more friction than it removes.
  • JAWS treats AI as a feature add. Their core stays a screen reader. They can't cannibalize their own product without killing margin.
  • General agents weren't designed for voice. Their screenshot loop breaks the conversational rhythm blind users actually use.
  • Lived-experience moat. My brother is QA, not a focus group. Edge cases ship in days, not quarters.
Business model

$50/month per seat. Same buyer as JAWS, 1/30th the price.

B2G. Government

State vocational rehab agencies (ADA-funded). Federal: VA, DoE, Section 508 procurement.

B2I. Institutions

Universities (disability services), employers (accommodation budgets), nonprofits and rehab centers.

B2C. Direct

Free trial, no card. Self-serve seats. Institutional users convert family and colleagues.

Same budget code, approval process, and compliance review across all three channels as the JAWS line item it replaces.

Team

Lived experience and engineering
under the same roof.

Ameen Darwish. Founder, CEO, Engineer

Software engineer. Built Darvy solo since late 2025. Background in AI tooling and ML pipelines. Going full time on funding.

Majd Darwish. Product, QA, the user

Brother. Lost his vision as an adult. Tests every feature daily. His frustration is the roadmap. Iteration cycles measured in hours, not sprints.

The ask

Pre-seed.
Close the alpha-to-beta gap in 60 days.

Use of funds: founder runway, infra (LLM credits + voice infrastructure), pilot deployments at 2 to 3 rehab institutions.

See it work

Demo.

Watch the demo →
youtu.be/zHyV5yl2rrw
Ameen Darwish · Founder, Darvy
Darwish.Ameen@gmail.com  ·  +972 54 552 0777