The research terminal that answers in sentences
Ask a question about a company in plain English. Get an answer grounded in attention data nobody else sells, earnings-call transcripts, insider filings and news flow — with the working shown.
First, what this is not. This is not a replacement for a market-data terminal. No real-time quotes, no Level 2, no order routing, no dealer chat, no fixed income. If you need those, keep your terminal. This is the research layer that sits next to it — and it answers in sentences.
Why it is different
Conversation, not commands
Conventional terminals are driven by memorised codes. Here you ask in English and the system decides which of more than fifty instruments to reach for, chains them, and shows its working. There is no command set to learn and no syntax to forget.
Data built, not licensed
The attention layer underneath — how much interest a company or theme is actually attracting, week to week, across the open web, retail order flow, news and search — is measured in-house rather than bought from a vendor every other desk can also call. It is the part of the answer you cannot get somewhere else.
Research that remembers
A desk can register a hypothesis before it is tested. The specification is hashed and pinned, both sides approve it, and it runs against fixed evaluation windows under standing statistical floors, with a running correction for how many ideas have already been tried. When the answer is negative, it says so. That is the opposite of a tool that will happily produce a good-looking backtest on the twentieth attempt.
What you can ask it
Real questions, answered end to end — each one would otherwise be a screen, an export and a spreadsheet.
“What has attention on this company done over the last year, and what moved it?”
A series, a chart, and the events sitting under the spikes — in one answer, not four tabs.
“Which names are drawing unusual interest this week?”
A cross-sectional screen over the whole tracked universe, ranked, with the move behind each one.
“Has any executive been buying?”
Insider filings and congressional disclosures, with the reporting lag stated rather than hidden.
“What did management actually say about margins last quarter?”
Verbatim from the earnings call, plus how the answer changed across quarters and who asked.
“Which analyst has been pressing this theme, and where else?”
One analyst's questions tracked across every company they cover.
“Is this the real story or forty syndicated copies of it?”
News clustered into distinct stories with the first source and how far each one spread.
What it won't do
It routes these away rather than guessing, which is what makes the rest of the answers worth reading.
- ×Real-time quotes and Level 2not a market-data feed
- ×Order routing and executionyour broker, not us
- ×Dealer chat and OTC negotiationno messaging network
- ×Fixed income depthequities, and the attention layer around them
- דShould I buy it?” · “What's the price target?”context and evidence, never advice or a forecast
It is also an endpoint
The same instruments are available over the Model Context Protocol, so your own assistant can call them directly. That matters more than it sounds: a research terminal you have to remember to open competes for your attention, while an endpoint your existing tools already reach becomes part of how the desk works. Available on request alongside a seat.
How access works
- Ask. One email — tell us what you would want to ask it first.
- Get a link. A scoped instance of your own, seeded with your context. Nothing to install and nothing to sign to look.
- Go deeper under NDA. The internal research layer, the methodology annex and a workspace for pre-registered testing open up once a mutual non-disclosure agreement is in place.
AssetFlow.ai LLC. Informational only — not investment advice. AssetFlow Intelligence reports attention, filings and news context; it does not make recommendations, produce price targets or forecast returns. Access is provided at our discretion, and the internal research layer is provided to institutional users under a mutual non-disclosure agreement.