Talk to the data.

You have the 10-K open in one tab. An Excel sheet of ratios in another. Six earnings-call transcripts you haven't read. A thesis you're trying to test. You spent the morning on a question that should have taken ten minutes. The margin compressed in Q3 — you know that — but you don't know why. The screener said what. The terminal said when. Neither one explained the why, and the why is what makes the decision defensible. So you keep reading, and the afternoon disappears, and the question still feels open.

We built EquiVault because the dashboard era is over. Dashboards answered what when the data was scarce and the questions were predictable. Both conditions changed. SEC filings, earnings transcripts, insider trades, financial metrics — the data is no longer scarce, it is overwhelming. The questions are no longer predictable, they are specific. A row in a spreadsheet cannot tell you why iPhone mix shifted toward lower-end SKUs in China. A bar chart cannot explain why Tim Cook bought 250,000 shares two days ago. A four-rectangle hero-metric grid cannot defend an investment thesis to a partner who is paid to disagree with it. The medium that can do all of those things is conversation — sourced, timestamped, defensible — and the substrate that makes that conversation possible has finally crossed the threshold of being economically viable. So we built it.

The product rests on three primitives. Facts are the system of record: every SEC filing, every earnings transcript, every insider Form 4, every financial metric, ingested with a freshness SLA. Every fact carries a source and a timestamp. If we don't have it, we don't claim to have it. If it is stale, it wears a chip. Atoms are the renderable knowledge unit: gross margin compressed by 2.1 points, ROIC at 28.4%, services revenue grew 18% — each one hoverable, citable, pluggable. The canvas snaps them into shape so you can see margin compression, not search for it. Narration is the explanation layer: Atlas, the chat narrator, streams structured visual cues alongside prose. Words light up as they are spoken. Causal edges draw between atoms. The camera glides across the canvas. Reading and seeing happen at the same time.

Two pillars hold the product together, and we will not negotiate either one. Fresh data and fresh signals are the essence. Stale data is the silent failure mode of every research product, and we refuse to ship it without a freshness chip. If a number is two days behind the filing, you will know. UX is interpretation and explanation, not more charts. When the answer is unclear, we add words, not boxes. When the question is hard, we cite the page in the 10-K, not a vendor data feed. The canvas does not exist to make the data look pretty. It exists to make the reasoning legible.

The trust contract is how we earn the privilege of being used at all. Every atom shows its source. Every freshness chip shows its age. Every credit charge has a pre-action cost preview rendered server-side, so you see what an action will cost before you spend it. Every debit is idempotent — we will never charge you twice for one question. If you exhaust your credits, we tell you the rate and the projected charge and put the next step behind a one-click confirm. We do not silently fail to a 402, we do not render "no data" copy when the real answer is "stale" or "upgrade," and we do not bury the unit economics of a question in jargon. A typical research question costs five credits. If your prompt is one thousand characters long, it costs six. The math is on the page; the math is in the docs; the math will never be different from what we showed you.

A manifesto says what it is by saying what it refuses to be. We are not the Bloomberg Terminal: expert-coded, dark, $2,400 a seat, designed for the senior partner who has been reading L2 order books for twenty years. The terminal was the right product for its decade; the decade is over. We are not the gamified retail app: confetti on trade, dopamine on streak, the slot-machine UX that turns fiduciary work into a habit loop. Confetti undermines trust, and trust is the only currency in our category. We are not the four-rectangle SaaS dashboard, with its three navy gradients and its hero-metric grid and its training-data sameness. That is the AI-trained-on-Dribbble reflex, and the reflex is wrong. We are purple-led specifically because the safe color in finance UI is navy, and the safe choice in this case is wrong. We are dense without being noisy. We are subtle without being precious. We are confident without being cocky. When in doubt about how polished we should ship, we ask: would Stripe ship this? Would Linear? If the answer is no, we do not ship.

Glance is free. It opens with a frozen workspace — Apple Q1, Quality Compounder blueprint, pre-recorded narration — and plays out the canvas in ninety seconds. No credit card, no commitment, no spreadsheet. Pick a blueprint. Watch a real company narrate itself. Decide if it changes the way you think about research. If it does, lock founding-member pricing for twelve months before v2.0 opens to the public. If it doesn't, write us at the address below and tell us what would have. We read every reply. We answer every question. We will not be everything to everyone, but we will be the right product for the analyst who has spent enough Saturdays reading 10-Ks to know there has to be a better way.