MtKingdom
Services Work About Contact Explore our products
Case Study 02 · All work Beta

DISCLOSEiq
Public filing research.
Every claim, cited.

Filing research for listed companies. Enter a ticker — get a 3-year deep dive across annual, quarterly, current, and proxy reports, with every finding, number, and narrative shift traceable to the exact line in the original document.

discloseiq.mtkingdom.io
Research
Search & ingest
Dashboard
Findings
Financials
Timeline
Finding · Risk

New risk-factor language added in the FY25 annual report referencing AI-related regulatory exposure — not present in the FY24 filing. [C4]

Evidence · Annual report — Risk Factors

"...emerging AI-related regulation may impose additional compliance obligations and could materially affect our operations..."

Evidence coverage
0%
How It Works

From ticker to cited deep dive in minutes

No PDF scavenger hunts. No spreadsheets stitched together by hand. Enter a ticker and let the research come to you.

1

Enter a ticker

Search any listed company in our coverage. DISCLOSEiq pulls the last 3 years of public filings — annual, quarterly, current, and proxy reports — and parses them into structured sections.

2

Findings surface automatically

Financial trends, risk flags, narrative-change alerts, and material events appear as a structured memo — each with an inline citation back to the exact filing, section, and span.

3

Ask questions, get answers

Grounded Q&A over the filings. Every answer ships with clickable citations — no hallucinated claims, no made-up numbers. If the filings don't say it, the app won't either.

Features

Built for research rigor

Every surface is designed around one non-negotiable: if you can't cite it, we don't show it.

Every claim cites its source

Each finding is linked to a precise location in the source filing — down to the character range. Click a citation to see the exact excerpt, highlighted in context. No black-box conclusions; no "trust me" outputs.

  • One-click from a claim to the exact highlighted span in the filing
  • Numbers link back to their underlying XBRL fact, not a model guess
  • Uncited claims are rejected before they reach the UI
  • Reproducible analysis runs — re-open last month's memo and see the same evidence

Finding · Going concern language

"substantial doubt" phrase detected in FY24 annual report — Risk Factors [C1]

Finding · Revenue trend

Revenue −2.8% FY24 → +4.6% FY25 · sourced from XBRL [C2]

Finding · New risk factor

Sentence added to FY25 annual report not present in FY24 [C3]

Risk flags — the ones that actually matter

A curated set of signal-bearing phrases and structural cues are flagged automatically across the 3-year window, with the exact span surfaced so you can judge whether it's material or boilerplate.

  • Going-concern · impairment · restatement · material weakness
  • Cyber incidents · covenant breaches · customer concentration
  • Leadership turnover · restructuring · material agreements (from current-report timeline)
  • Boilerplate de-prioritized so real signals rise to the top
Risk flags
4
Across 3-year window
New in FY25
2
vs. prior year
Current-report events
17
Categorized & dated
Boilerplate
Filtered
Cross-filing hash match

Financials, already normalized

Issuers use dozens of different concepts for "revenue" across filings and years. DISCLOSEiq resolves those into a canonical set of metrics — so you get consistent trends, not a patchwork of labels.

  • Canonical metrics: revenue · operating income · net income · EPS · cash · operating cash flow
  • Year-over-year · 3-year CAGR · margins · anomaly z-scores
  • Segment breakdown (product/geography/business) where the issuer discloses it
  • Amended filings supersede originals — no stale numbers in the trend line
Revenue (FY25)
$401.0B
↑ 4.6% YoY
Operating margin
+30 bps
FY25 vs. FY24
Cash flow
$118.3B
Operating · FY25
Capital return
$105B
Buybacks + dividends

Grounded Q&A · no hallucinations

Ask a question about any company. Answers come with clickable citations into the filings themselves. If the source material can't support an answer, the app says so — it doesn't guess.

  • Hybrid retrieval: semantic + keyword search over 3 years of parsed filings
  • Confidence flag when the evidence is ambiguous — both candidates surfaced
  • Inline citation chips expand to the highlighted source excerpt
  • Boilerplate down-weighted so answers draw from company-specific language

Q: How has Services revenue mix evolved?

Services share rose 22% → 24% → 26% across the 3-year window. [C1] [C2]

Low-confidence answer

Two candidate passages support different conclusions — both surfaced for review rather than a confident guess.

"I can't find that in the filings."

Honest no-answer when the source material doesn't cover the question.

Two AI companions

Atlas & Sage

Same evidence. Same citations. Two different audiences. Switch between an analyst-grade research partner and a patient teacher with a single toggle.

Analyst mode

Atlas

A research partner built for analysts. Ask a question across any ingested company and get back a cited answer — or a markdown table, or a multi-ticker comparison — with an evidence trail you can open and inspect.

  • Inline [C#] citations · clickable excerpts · provider badge under each answer
  • Structured outputs: markdown tables (revenue · YoY · margin) and multi-ticker comparisons
  • Saved queries, batch runs across 2–4 tickers, one-click CSV export of any answer table
  • Memo-on-demand · command palette (⌘K) · session threads with follow-up chips
  • Refuses to answer about companies outside the evidence set — no hallucinated tickers
Learner mode

Sage

A patient teacher for beginners and students. Sage rewrites the same analyst findings in plain English at the reader's level — then checks understanding with Socratic follow-ups and short quizzes drawn straight from the filings.

  • Age-level tuning (teen · beginner · intermediate) with reading-grade guardrails
  • Conversational answers to a 5-question framework — grounded in the underlying financials
  • Socratic mode: Sage asks questions back instead of handing over answers
  • Auto-generated quizzes · vocabulary glossary · lesson launcher per ticker
  • Simplified copy is numerically grounded — Sage can't invent a figure the deterministic layer didn't compute

One evidence spine, two voices. Atlas and Sage read from the same ingested filings, the same normalized financials, the same risk-detection layer — so an answer in learner mode and an answer in analyst mode point to exactly the same cited spans.

More capabilities

Everything else it does

Smaller features that add up — the details that separate a demo from a research tool.

Narrative change detection

Section-level diffs across consecutive filings surface new sentences added to Risk Factors or management commentary — the real signal buried in thousands of boilerplate words.

Event timeline from current reports

Every current report in the 3-year window categorized (material agreement · leadership change · cyber incident · restructuring · acquisition) and plotted on a filterable timeline.

Amendment awareness

Amended filings are linked to their originals. The analysis pipeline prefers amended content — your memo reflects the company's own corrections.

PDF memo export

Export a firm-ready memo with executive summary, findings, citations, and financial trends — formatted for distribution, with citations preserved as footnotes.

Source explorer

Open any filing in a split-pane reader: normalized section list on one side, raw HTML with highlighted spans on the other. Verify any finding directly against the filed document.

Re-runnable analysis

Every analysis is stamped as a run. Re-run at any time as new filings arrive — older runs stay intact for reproducibility and audit.

Integrity & fair access

Research you can audit

Public filings are the source of truth. We treat them — and the systems that publish them — accordingly.

Evidence chain, end to end

Every finding carries a citation to a specific filing, section, and character range — or to an underlying XBRL fact. Uncited model output is filtered out before persistence.

Fair-access compliance

Downloads run through a rate-limited client with descriptive user-agent and backoff — we stay well within the fair-access ceiling for our public-filing sources and cache aggressively so we don't re-hit what we already have.

Reproducible analysis runs

Every memo is a stamped run — rerunnable, versioned, and auditable. Compare today's output to last quarter's to see exactly what changed and why.

Important. Information provided by DISCLOSEiq is for informational and educational purposes only and is derived from publicly available filings. Findings, risk flags, narrative-change alerts, and Q&A answers are algorithmic estimates — not financial, investment, legal, accounting, or other professional advice. While data-integrity and citation checks are applied, users are responsible for independently verifying all information and determining its suitability for their use. MtKingdom.io and its affiliates make no representations or warranties and disclaim all liability for decisions made based on this information.

3 yrs
Filing coverage per ticker
4
Filing types (annual · quarterly · current · proxy)
100%
Findings cited to source
Public
Listed-issuer coverage (Beta)
Get Started

Ready to research?

DISCLOSEiq is in beta. Access is invite-only while we expand coverage and refine the evidence pipeline — email support@mtkingdom.io to request an account.

Sign in