Self-Sovereign AI Operator
Own the operator core — resident daemon, dashboard/CLI, memory, push-to-talk voice, scheduled tasks, a guardrail gate, and connectors
Sovereign is the flagship: a owner-controlled AI operator core you own outright. A launch-at-login resident daemon, dashboard and CLI, grounded memory vault, scheduled tasks, push-to-talk voice with a spoken daily digest, a client/deal CRM, a safety guardrail policy gate, and six one-click connectors — all running with your own CLI agent (Claude, Codex, or any prompt-capable CLI on your PATH) or a local Ollama model as the brain. No platform rent. No vendor kill switch. Builder rights cover white-label client deployment.
A one-time perpetual license — $500. You receive the Builder license, the owner-controlled operator package, and 30 days of engineer support.
Get Sovereign Builder License — $500Prefer to spread it out? Pay $50/mo x 10 or $100/mo x 5 — same $500 total, auto-stops after the last payment.
Every chatbot subscription bundles two things into one rent check: the operator (the shell, memory, automations, dispatch) and the brain (the inference). Stop paying and you lose both — your prompts, your workflows, your history, your context. Sovereign unbundles them. You own the operator forever and bring your own brain.
ChatGPT Team, Claude Pro, Jasper, Notion AI — every one is a login that expires. The moment the invoice lapses, the workflows you built and the memory you accumulated vanish with the seat. You never owned them.
Client history, deal pipeline, private notes — all sitting in a vendor's cloud under their terms. A model deprecation, a policy change, or an outage takes your operator down with it.
Switch inference vendors and you start over. The operator and the model are welded together, so you can't keep your memory while changing what does the thinking.
You babysit a tab. Nothing runs while you sleep, nothing remembers across sessions, nothing acts on a schedule. It answers — it doesn't operate.
Sovereign arrives installable, branded, documented, and verified for the supported core. You get a local operator package you can rebrand and extend, with higher-tier surfaces labeled when they require optional setup, credentials, or future build-out.
A supervised daemon launches at login on the Developer-ID Mac download, holds grounded context across sessions, and dispatches work without you babysitting a chat window. It restarts itself on failure, logs what it does, and runs your scheduled tasks while you're logged in — so the operator is there when you wake up, not when you remember to open a tab.
Reasoning dispatches to any prompt-capable CLI on your PATH — Claude, Codex, or your own — never bundled, never locked to one vendor, never marked up. Prefer fully on-device? Point it at a local Ollama daemon instead. Your subscription (or your own hardware) is the brain; Sovereign is the operator around it. Swap brains without rebuilding, and your memory vault survives the change.
Voice supports push-to-talk: hold the hotkey, speak, and the operator transcribes locally, dispatches your brain, and speaks the answer back. Ask "what moved today?" and it reads a spoken daily digest aggregated from your audit ledger, CRM, and notes. Wake-word support ships, remains off, and does not listen until you explicitly enable it; the dashboard and CLI remain the primary control surfaces.
Client history, deal pipeline, prior decisions, audit trails — persisted in a vault on your machine and retrieved from stored facts, tags, and timestamps, not improvised on the fly. You can view, edit, and delete what it remembers. The operator answers from your reality, and refuses honestly when it has nothing grounded.
User-invokable skills, schedule entries, dashboard/CLI commands, and plan→act→verify agent structure. Run a skill on a schedule or a trigger so proactive work happens without a prompt. The architecture is agent-ready — extend it with your own capabilities.
One-click account connectors for GitHub, Linear, Slack, Gmail, Google Drive, and Google Calendar — OAuth or token, your own credentials, stored only in your Keychain. Add any MCP server for more, plus local docs/RAG. Connectors you haven't authorized stay capability-gated instead of pretending every tool is live.
Agents operate inside permission walls you define, and every autonomous action writes a receipt to an audit log you can review — proof of execution, not narrated fake work. A deterministic guardrail classifies every tool call as read-only, side-effecting, or destructive and gates it against the posture you set — it can block a side-effecting action and ask for your confirmation before it runs.
Sovereign is built in layers — a core spine that works on day one, then foundational, mid-tier, premium, and differentiating capabilities stacked on top. Items marked present ship today; the rest are the standard the operator is built against.
Sovereign is a owner-controlled operator with a verified core spine — launch-at-login daemon, dashboard/CLI, grounded memory, skills, scheduled tasks, six one-click connectors plus any MCP server, push-to-talk voice with a spoken daily digest, a client/deal CRM, a safety guardrail policy gate, and proof/verify commands — built against the standard above, with higher tiers landing over time. The operator does not write or compile its own code. Wake-word support ships, remains off, and does not listen until you explicitly enable it. You bring the brain: any prompt-capable CLI (Claude, Codex) or a local Ollama model for fully on-device runs; memory-backed chat works without one. That's the design, stated plainly.
The operator owns the shell, memory, dispatch, voice, and safety. The brain owns inference. Here's the loop, from push-to-talk to verified action.
Hold the push-to-talk hotkey and speak, or open the command palette and type. The operator transcribes the request locally and loads relevant context from your vault.
The daemon shells out to your CLI agent login — the brain you configured — passing the request plus grounded memory. Inference happens on your terms.
Agents plan, then act through your connected tools — GitHub, Linear, Slack, Gmail, Drive, Calendar, local docs/RAG, MCP servers, and your own skills — inside the permission walls you set. The guardrail gates every side-effecting or destructive step on your confirmation.
Every autonomous action writes a receipt to the audit log. The operator reports what it actually did — real results only — and updates the vault so next time it remembers.
The Builder license delivers the verified core operator package. Here's the stack you download.
Watch the operator shell boot and run — then the delivery moment you get the second checkout clears. The same UI that ships in your zip, shown on sample data.
Loading Sovereign shell…
Sovereign isn't for hobbyists tinkering with prompts. It's for people who sell intelligence, automation, or edge — and need a product they own behind the offer.
Sell a branded AI operator to clients without building from scratch. You own the deployment, set the price, and keep the relationship. Sovereign is the product behind the retainer.
Buy Trading Engine as a separate product — backtest and paper ledger published on the trading page. Ace documents how it connects with research, journaling, and content, but each product is its own purchase and current runtime is verified separately.
Lead scout, content spotlight, and probate sourcing run as separate products on the same operator — each with its own page and proof boundary. Provider, mailbox, and county-gated work feeds the memory your deployment already knows.
A subscription gives you a seat at someone else's table — pay forever, own nothing, lose it all when you stop. Sovereign is the table. You buy it once ($500), own the operator outright, bring your own brain, and keep your memory through any provider change. Ace is the private reference architecture behind this design; current runtime is verified separately rather than inferred from this page.
Sovereign is $500 — buy once. The monthly products below run on the same operator architecture but are each their own purchase. Trading starts with execution off and paper-first, and no performance is guaranteed.
ES-only signals engine — multi-timeframe consensus, direction-lock, and a self-governing alpha monitor. Live feed; execution is paper with every fill published. Signals-only; no performance guaranteed.
Content engine for your brand — email, social, voiceovers, scheduling. Autonomous channel posting is gated while content quality is raised.
A sourced business encyclopedia — a sourced, offline encyclopedia of niches, operations, failures, and tools for building a real business.
Perpetual Builder license. Rebrand, resell, deploy to clients. Monthly products (Trading, Circuit, Marketing, Academy) sold separately. Read the license →
One-time perpetual license
macOS · Apple Silicon · Developer ID, notarized — not on the App Store. Runs on your own CLI agent login. Windows and mobile downloads are coming soon.
Notarized by Apple · runs on your machine · your data stays with you. Windows and mobile downloads are coming soon.
Sovereign is a owner-controlled operator you hold on your machine — with reasoning through your own CLI agent (Claude, Codex) or a local Ollama model. $500 one-time. Real ownership of the operator, not a login that expires.
Pay once — $500. The operator is yours forever. Rebrand it, resell it, deploy it to every client in your book. No platform rent. No vendor kill switch.
Your daemon and vault run where you put them — not on our servers. Your CLI brain handles inference on your terms. Swap brains without losing your vault.
Trading $75/mo, Circuit $25/mo, Marketing $99/mo, Academy $30/mo — each its own product. Sovereign is not a bundle hub.
Subscriptions like ChatGPT Team, Claude Pro, or Jasper give you a seat at someone else's table. When you stop paying, you lose everything — prompts, workflows, history. Sovereign is the table. You own the furniture.
The $500 price covers Sovereign only. Trading ($75/mo), Circuit ($25/mo), Marketing ($99/mo), and Academy ($30/mo) are separate — not bundled add-ons.
You bring a brain for full reasoning: any prompt-capable CLI on your PATH (Claude, Codex), or a local Ollama model for fully on-device runs. Sovereign owns memory, dispatch, voice, and safety rails; the brain owns inference. Memory-backed chat works without one, but a brain unlocks advanced dispatch and deep reasoning. Swap brains anytime without losing the vault.
It's owner-controlled and on-device, not a fully offline AI in a box. The operator — daemon, memory, voice, skills, safety — runs on your machine. Inference goes to the brain you configure: any prompt-capable CLI, or a local Ollama model for fully on-device runs. Autonomy is real but gated — a deterministic guardrail classifies every tool call and can block a side-effecting action until you confirm, and every action writes a reviewable receipt.
A modern Mac (Apple Silicon recommended). The operator itself is light — 16GB of RAM is plenty when your brain is a cloud CLI like Claude or Codex. If you run a local Ollama model fully on-device, size RAM to the model: 16GB for small models, 32GB+ for larger ones. The operator runs on your own Mac; no cloud shell required.
Yes — that's the Builder license. Deploy under your own logo, persona, and colors, set your own pricing, and keep the client relationship. The operator is yours to rebrand and resell. Read the license terms.
Bring your own. Sovereign does not mark up inference — you bring your own CLI agent login, API key, or local model. Black Label Bots bills the platform license and optional monthly products only. Your token lives in your Keychain; it is never bundled or sent to us.
Deployment help, branding setup, and architecture questions from the team that built Sovereign. Monthly products (Trading, Circuit, Marketing, Academy) have their own onboarding when purchased separately.
$500 one-time — yours outright. Launch-at-login daemon, dashboard/CLI, memory, push-to-talk voice, scheduled tasks, CRM, a safety guardrail, and six one-click connectors. White-label Builder rights. Owner-controlled, installed and run on your own Mac, with no Black Label backend in the path — your data never touches a Black Label server. Runs on your own CLI (Claude, Codex) under your own login, or fully on-device on local Ollama. 30-day engineer support included.
Occasional product updates only. Unsubscribe anytime.