VIGIL is an AI workspace that thinks before it acts — a multi-advisor council, a data-grounded trade desk, and your finance, CRM, mail and document studio, all under one human-in-the-loop agent.
Convene an AI council over any decision — CFO, CTO, Legal, Product lenses argue, score, and a chairman synthesizes a verdict. Run meetings, crystallize artifacts, ground everything in your own documents.
Forecasts and a data-grounded signal debate propose trades — every order passes a single-use human approval gate, capped at 5% of NAV. Connect your own broker; keys never leave your account.
Each surface is a real tool the agent can read, reason over, and act in — never autonomously where money or messages are involved.
Multi-advisor deliberation over any topic or transcript, with weighted consensus and a chairman's verdict.
Live candles, forecasts, and a data-grounded signal debate — proposing trades behind the approval gate.
Capture → classify → reconcile your books. A P&L the CFO suite reasons over.
Contacts and a deal pipeline; value rolls up to Finance, deals route to the Council.
Inbox triage over your mailbox, with AI classification. Outbound is review-then-send.
Draft, refine, and crystallize artifacts — proposals, briefs, contracts — grounded in your Vault.
Your real documents, so the agent cites what your contracts and statements actually say.
A tamper-evident, hash-chained log of everything the agent does. Defensible by design.
Before building or recommending, it brainstorms approaches and presents a design. No jumping to output on unexamined assumptions.
It reasons from your Vault documents, your live broker, your books — citing real numbers, never fabricating.
Trades go through an approval gate; outbound mail is review-then-send. You stay in control of anything with consequence.
Single-use approval gate on every order. 5%-NAV cap. Nothing moves money on its own.
Connect your own broker; credentials stay in your account. Every record scoped to you alone.
The council surfaces the bear case, not just the bull. You can always override the verdict.
No 'done' without verification. No tax or trade claim without a cited source.
VIGIL is an AI workspace that deliberates before it acts. It combines a multi-advisor AI council, autonomous AI departments, and a human-in-the-loop crypto trade desk with your finance, CRM, mail, and documents — all under one agent that thinks first and waits for your approval.
Unlike a chatbot, VIGIL is built around human-in-the-loop control: every trade passes a single-use approval gate and outbound mail is review-then-send, so nothing with real consequence happens autonomously. It also brainstorms and presents a plan before acting, and grounds answers in your own documents and live data.
The AI council is a panel of advisor lenses — CFO, CTO, Legal, and Product — that debate a decision, score it, and have a chairman synthesize a defensible verdict. It surfaces the bear case and dissent instead of hiding it, and you can always override the verdict.
No. VIGIL proposes trades from data-grounded forecasts, but every order is human-approved and capped at 5% of NAV. You connect your own broker and your API keys never leave your account.
Yes. Every record is scoped to your account with row-level security, your broker and API keys stay in your account, and a tamper-evident, hash-chained audit log records everything the agent does.
VIGIL runs autonomous AI departments for recurring work — support triage, finance reconciliation, revenue follow-ups, lead scouting, and legal review. Each department only counts a task as done when it passes a deterministic effectiveness check, so output is verified, not assumed.
Create an account and convene your first council, or wire up your trading desk in minutes.