Product proof

What we have actually shipped.

The apps below are working products. They show how we handle device capabilities, local data, offline work, and human review. For client work, we publish a result only after we have a starting measure and permission to share it.

Explore the public apps Read the trust boundaries

The products

Each one starts with a job someone needs to finish.

These examples show product capability. They are not claims that every client deployment produces the same business result. We measure that separately.

Public product

PlantCare

The need: A gardener needs a useful next step from a plant photo, without turning uncertainty into unsafe advice.
Deployment: Camera-led iPhone/iPad workflow with local model work and explicit safety boundaries.
What the product proves: The app supports a defined set of plants, shows caution states, sets reminders, and stops when the job needs an arborist.

Public product

Safari Go

The need: A traveller needs wildlife identification and field notes when the network is unreliable.
Deployment: On-device identification with installed destination content and a local journal.
What the product proves: The app says what it can identify, uses installed destination packs offline, and handles sensitive locations carefully.

Public product

Farmbit Mobile

The need: A field crew needs to complete assigned work and capture evidence before connectivity returns.
Deployment: GPS, media capture, offline queue, and synchronization across the mobile field loop.
What the product proves: Crews see assignments, map status, queued uploads, reports, and sync state in the same field workflow.

Launch-track product

VoiceAgenda

The need: A person needs to turn an intentional recording into a useful note and reviewable follow-up actions.
Deployment: Supported Apple device capabilities with local recording, analysis, and user-approved Calendar or Reminder actions.
What the product proves: The public page states the device boundary and keeps launch status separate from the four public App Store products.

Public product

WolfeWriter

The need: A fiction writer needs long-form context to stay with the project while AI suggestions remain reviewable.
Deployment: Native Mac workspace with local project state, Story Memory, and proposed changes.
What the product proves: Project context stays with the writing workspace, while suggested changes wait for the writer’s approval.

For business work

A shipped product is not a customer result.

Baseline the loop

Volume, handling time, response time, error rate, missed work, or another agreed measure.

Build the first boundary

Inputs, tools, model placement, human review, fallback, and ownership are written down.

Run it in production

The workflow is observed with real users and real exceptions, not just a successful demo path.

Publish only approved results

Customer names, screenshots, numbers, and quotations are used only with permission.

Have a workflow to test?

Start with the number the workflow needs to move.

Bring the repeated loop, current tools, and the result that would make the change worthwhile.

Start a workflow review