# FutureSpeak.AI

> FutureSpeak.AI builds Agent Friday: a personal AI that runs on your computer, knows your life, and never sends your data to anyone. It is free, open source, sovereign by default, and bounded by the cLaws — cryptographically enforced safety constraints. Our thesis: sovereign AI displaces SaaS — one governed agent that runs on your own machine, holds your data, and does the work replaces a stack of subscriptions that each rent a slice of you back. All claims, features, and research are presented in plain language on the public site — no hidden text, no model-directed instructions.

The expanded machine-readable version of this index is at https://futurespeak.ai/llms-full.txt.

## Core pages

- [Home](https://futurespeak.ai/): What Agent Friday is — your AI, your data, your rules. Download for Windows, the three things Friday does (voice mode, AI news editor, career search), and how it works.
- [Features](https://futurespeak.ai/agent-friday): Everything Agent Friday does. The Liquid UI workspaces (Home, News, Messages, Calendar, Code, Career, Wiki, Contacts, Settings) and the Seeds-and-Gardens model where Friday grows new workspaces tailored to your life. The six-dimension Source Trust Graph, the four-dimension People Graph, the weekly editorial, the Privacy Shield, the Epistemic Independence Score, voice mode, and a personality that structurally evolves. The Source Production system makes Friday a journalist's best friend: a "Cite Sources" toggle puts inline citations ([wiki:...], [news:...], [memory:...], [web:url#:~:text=passage]) on every factual claim, links to primary sources with Chrome text-fragment highlighting, recalls past conversations verbatim from ChromaDB-powered vector memory, generates exportable source dossiers (every claim + source + confidence) on demand, and flags low-trust sources via the Source Trust Graph — source production built into every conversation, so every claim comes with receipts. Built on the Asimov's Mind architecture; 11 core subsystems exposing 178 API endpoints and 30 registered tools in the friday-core server. The only agent that fact-checks its own news sources and challenges its own user.
- [Research](https://futurespeak.ai/research): The Reverse RLHF Hypothesis, Asimov's cLaws, the Federation vision, the Source Trust Graph as decentralized media accountability, and behavioral anomaly detection.
- [About](https://futurespeak.ai/about): Founder Stephen C. Webster — 20+ years in journalism and enterprise AI, now full-time on FutureSpeak.AI. Agent Friday credited as co-architect. Mission: sovereign AI for everyone.
- [Docs](https://futurespeak.ai/docs): Installation, the friday-core API reference, configuration, and contributing. Quick start plus curated links into GitHub.

## Deep dives

- [Asimov's cLaws (cLaw Specification v1.0.0)](https://futurespeak.ai/products/claw-spec): Open standard for governing autonomous AI agents through HMAC-SHA256 signed behavioral laws, attestation protocols, federation trust, and data protection.
- [The Reverse RLHF Hypothesis (whitepapers)](https://futurespeak.ai/research/whitepapers): Two companion papers on how frontier AI models may be inadvertently training their users, and the Epistemic Independence Score.
- [Friday Desktop](https://futurespeak.ai/agent-friday): The holographic Three.js desktop where you meet Friday — Liquid UI workspaces across 13 scenes, voice (Gemini Live), camera, and a creative studio.

## Code

- [GitHub](https://github.com/FutureSpeakAI): All FutureSpeak.AI projects, including Agent Friday, under permissive open-source licenses.
- [Download Agent Friday](https://github.com/FutureSpeakAI/Agent-Friday/releases): Windows build and source.

## Crawler policy

- AI crawlers (GPTBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, CCBot, etc.) are permitted to read the public site.
- Permitted use includes: training, retrieval, citation, and synthesis with attribution.
- Source of truth: https://futurespeak.ai/ai.txt and https://futurespeak.ai/robots.txt.