01 · Robust searching
Beyond keyword
Classifiers built per filing type read the way an analyst would — semantic, not string matching. Ask for a concept and get the passages that mean it.
529 classifiers · 245M sentences
Investment Research
Every answer arrives with its sentence.
Ask a question across every SEC filing since 1994 and get the answer with the filing, the paragraph and the date attached — so the next person who asks where it came from already has it.
- 17.9Mfiling sections indexed
- 245Mfiling sentences scored
- 529named classifiers
- 41KSEC filers scored
- <60sfiling to searchable
- 1994–point-in-time history
The problem
Keyword search finds documents. You needed an answer.
Full-text search returns a list you still have to read. AI summarisers return prose you cannot check. Both leave the analyst doing the same job: opening filings one at a time to find out whether the thing they suspect is actually in there. And then proving it.
What you get instead
- An answer, not a result listthe finding, with its source
- A citation per figurefiling, paragraph and date
- Coverage-wide questionsask across 40 companies at once
- Checkable in one clickstraight through to the filing
One question, end to end
what an analyst actually gets back
-
01 · The question
Which of these 40 companies changed their revenue recognition language?
plain language, across a coverage list
-
02 · The answer
7 of 40 changed it
each with the filing, paragraph and date attached
not a document list — a finding
-
03 · The filing
10-K · filed 2026-02-14
Accession 0000320187-26-000032
one click to the exact paragraph
-
04 · The history
Language tracked 12 quarters
first mention · repeated · restated
What's in the solution
Four things the platform does that search alone does not.
02 · Surfaced data
The figures, extracted
Terms, figures, dates and language pulled out of the document and put in a table — so you compare values rather than re-read paragraphs.
extracted, organised, integrated
03 · Peer analysis
Section by section
Benchmark one company's disclosure against its peers on the same section, and see what changed between years.
risk factors · MD&A · policies
04 · Connected context
Ownership, comp, news
The filing sits next to who owns the company, what the insiders did, and what was announced — resolved to the same entity.
NexusID
How it lands on your desk
Same platform — a different first question depending on the seat.
Investment analysts
The scenario
You cover 25–40 names. Every quarter dozens of filings land within days and you need to know which ones matter before reading a page.
SEC Pro+AI ChatAlerts
How Kaleidoscope helps
- Screen a whole coverage list in minutes for the change you care about
- Jump from any figure straight to the paragraph that produced it
- Compare a section against peers without opening five documents
- Build notes where every claim already carries its filing sentence
- Get told when the language you are watching changes again
Research directors
The scenario
You are accountable for what the team publishes. Every number in a note has to be defensible six months later, by someone who did not write it.
SEC Pro+ProvenanceReports
How Kaleidoscope helps
- Every figure in a note traces back to a filing and a paragraph
- No dependence on one analyst's memory of where a number came from
- Consistent extraction across the team rather than per-analyst method
- Export with citations attached for compliance review
- Point-in-time history, so an old note can be re-checked as-of its date
Consultants & advisers
The scenario
You are briefed on an industry you did not cover last week and need to be credible on it by Thursday.
SEC Pro+AI ChatM&A Intelligence
How Kaleidoscope helps
- Ask the naive question in plain language and get a cited answer
- Establish the peer set and how they describe the same risk
- Find the deals and the disclosures that set the precedent
- Hand the client an answer they can verify themselves
- Re-run the same question next quarter and see what moved
The products
One platform, three ways in. Most desks use the first two together.
SEC Pro+
The search itselfSearch and compare across every filing type, with the extracted figures in a table beside the language they came from.
- Semantic search across every SEC and SEDAR filing
- Extracted terms and figures, not just full text
- Peer benchmarking section by section
- Every result one click from its source paragraph
Explore SEC Pro+ →
17.9Mfiling sections indexed
Every hit cited to item, document and page — with the sentence highlighted in the filing
AI Chat
The plain-language way inConversational research over every filing, news article and 13F position, backed by the same index — with the sources attached to the answer.
- Ask in plain language across your whole coverage
- Cited sources from filings and news, never unsourced prose
- Follow up without restating the question
- Built on the filing and news index, not a general model
Explore AI Chat →
245Mfiling sentences scored
Argos — a plain-language answer with the filings behind it
Alerts
The part that watchesTell it the language, the figure or the filer you care about, and hear about it when the next filing lands rather than next quarter.
- Trigger on language changes, not just new filings
- Under a minute from EDGAR acceptance
- Route by coverage list or by filer
- Every alert links to the paragraph that fired it
Explore Alerts →
<60sfiling to alert
The live dashboard — filings, insider activity, 8-K events and leadership changes
Every product in this solution
Start with the research tool; add the rest as the workflow settles.
Full data catalog →Core platform
- SEC Pro+ Search, compare and extract across every filing type Research UI
- AI Chat Cited conversational research over the index AI
- Alerts Language and figure triggers, under a minute Workflow
- NexusID One company resolved across every source Resolution layer
Data layers that feed it
- SEC · SEDAR Filings 1994 onward, point-in-time Live
- XBRL Reported financials, as filed Live
- Ownership Institutional positions and flow Live
- News Materiality-scored events, 9-class taxonomy Live
- Transcripts What management said, alongside what they filed Live
- Agreements Material contracts, cited to the clause Live
- Mergers Announced and closed deals, terms Live
- Catalyst Earnings and scheduled events Live
- Company Profiles Private companies and their people Live
- Provenance The sentence behind every extracted value Live
Workflow & delivery
- Reports Source-linked notes, HTML · MD · JSON Output
- APIs Push results into your own stack REST · Bulk
- MCP Server Let your agent ask the same questions Integration
- Watchlists Coverage lists that re-screen themselves Workflow
| How it compares | Kaleidoscope | Keyword search | AI summarisers |
|---|---|---|---|
| What you get back | An answer with its source | A list of documents | Prose with no source |
| Checking a claim | One click to the paragraph | Read the document | Not possible |
| Across a coverage list | One question, all names | One search at a time | Context window limits |
| Extracted figures | In a table beside the language | Not extracted | Sometimes, unverifiable |
| As-of a past date | Point-in-time, frozen | Current documents only | No concept of it |
Why teams switch
It makes SEC research fun again — I stopped opening twenty filings to answer one question.
answer, not a result listcitation per figureone click to the paragraph
Research with receipts
Talk to our team about how you work today, or start a trial and explore the platform yourself.