Human For Now.

Applications / October 2026

What I’ve built.
What it can prove.

The work now includes a twenty-chapter game, an experimental public voice assistant, evidence review tools, and experiments in persistent local agents. Some are public. Some require access. Some are still development work.

Status review: October 31, 2026 · Dates and limits belong with the claims.

Public playable prototype

Boobs’ Journey

A game for my wife, starring Ruben and Scruffer. The current release has twenty chapters ending at A Home That Holds. The latest published build also keeps the camera clear of foliage and roofs, preserves solid-object movement boundaries, and makes training explicit: offer the prism, give the carry cue, then reward the delivery. Training progress survives a reload.

Evidence boundary: Targeted automated and browser checks are recorded in the project story. Physical-device performance, speaker sound, and normal-player pacing still need testing.

Play + read the build story →
Free experimental public pilot

Ask Ruben

General conversation and questions about Human For Now, with a simpler Start talking flow after the notice and typed input as a fallback. Recent work addresses background noise canceling transcription or cutting off a spoken answer. Company questions belong in the separate private assistant.

Evidence boundary: This is a free experiment, with no paid offering. Questions, replies, and source clicks are recorded for owner analytics after a visitor accepts the notice. Visitor transcripts are not published or used as shared knowledge. Permanent personal memory belongs only to the authenticated owner; visitor exchanges never become that memory or shared website knowledge. Ten-minute sessions, question limits, and spending stops bound usage. October 31 check: the web service is healthy and its voice worker reports ready, but public status says the chat runtime is offline on the Mac. Earlier live voice verification does not prove current conversation availability. Physical iPhone behavior and competing background voices remain unverified.

Ask Ruben about this site →Use the section guide →
Private / authenticated experiment

CEO Ask Ruben

A separate private assistant for spoken questions about approved business information. Recent work improves microphone diagnostics and hands-free turn detection in background noise, while keeping a typed fallback.

Evidence boundary: Browser, recognition, and local speech checks passed for the latest detector. The current iPhone experience still needs owner confirmation. This is separate from the public website guide; internal information and personal memory remain protected.

Authorized executive access →
Private / authenticated application

Ruben’s Cube

Formerly Customer Cube, this private workspace brings customer, product, qualification, and planning evidence together. Recent releases add sourced property research, visible conflicts, sensitivity ranges, and clearer separation between measured quantities and provisional estimates. The existing address stays the same. New parent-portfolio research explores how site-level product flags roll up to a parent; recorded flags, separately billed products, and actual installed coverage remain different measures. A related Hot 100 research interface still awaits complete authenticated release verification.

Evidence boundary: Research is still in progress. A planning estimate is not verified capacity or a forecast; beds, units, equipment, and rooms retain their original meanings. Company records, values, and internal findings remain behind sign-in.

Authorized access →
Private / authenticated applications

C1 review applications

A family of review tools for comparing records, examining evidence, and preparing decisions. Recent work connects review material, dated source snapshots, activity evidence, accountable follow-up, and checks on report delivery. An automation completing is different from a verified current report reaching the intended review channel. Checks keep proposed actions distinct from execution; better reports alone do not establish better outcomes.

Evidence boundary: Prepared artifacts do not mean external records were changed. Internal strategy, records, and commercial details are not part of this public account.

Local development

Service desk and knowledge base

Work on self-service answers, internal documentation, and reusable knowledge for future AI assistance. The current work is starting with small article sets and a local application preview.

Evidence boundary: A local preview is not a public release. Source quality, permissions, and useful answers need review before broader adoption.

Related adoption lessons →
Local development / recovery unproven

Portable recovery planning

A new effort to make the AI workspace recoverable across a lost machine or service interruption. Work includes encrypted capture, environment inventories, and explicit gaps in account-access recovery.

Evidence boundary: Capturing files does not establish a successful restore. Device-bound credentials, expiring tokens, provider-side permissions, and incomplete cloud access need their own recovery tests. A full independent restoration and completed transfer are not yet verified. No keys, account details, or backup contents are published here.

Read the recovery lesson →
Experimental / incomplete acceptance

Malcolm Mac Host

An experiment in continuing a task across AI interfaces while the designated Mac retains task state, execution history, and artifacts. The scope is deliberately narrow: one host, persistent tasks, and explicit approval boundaries.

Evidence boundary: Local tests and partial browser work do not establish reliable end-to-end continuation across clients. The browser acceptance work has encountered download and settings limitations.

Read the permissions guide →
Private pilot / synthetic demo

General Ruben territory intelligence

A mobile-first territory application with lead review, follow-up workflows, and General Ruben training. The original pilot uses synthetic people, addresses, signals, and relationships. A separate local county-research workspace is documented alongside it.

Evidence boundary: Synthetic examples are not real prospects. The local research workspace is distinct from the deployed pilot; neither its source coverage nor operational readiness should be inferred from the demo. No messages are sent automatically.

Authorized pilot access →
Historical project milestone

Daily Agenda

An iPhone agenda project that brought scattered information together and exposed a privacy problem in a demo bundle before the next release.

Evidence boundary: The published story documents July milestones. Current TestFlight availability has not been reverified.

Read the history →
Public browser tools

Readiness, cost, and signal tools

The readiness questions, pilot brief builder, OpenClaw cost calculator, and Origin Lab remain available on the site.

Evidence boundary: Budgets are editable scenarios; pattern checks cannot establish authorship. Tool inputs are processed in the browser, while the site has usage telemetry and a moderated comments API.

Open the tools →