QVerisEarnings

QVeris for developers

Build source-cited earnings agents.

One workflow returns structured earnings intelligence with sources, capability status, confidence scoring, and explicit missing fields — the raw material for agents that can be audited, not just believed.

Source citedGap awareAgent ready
analysis packetPOST /analyze
tickerNVDA
sourcescalendar · results · filings · news
missingtranscript
confidencemedium
Get a QVeris key →See a live analysis

The workflow

Capability map

configured

Calendar

figures

Results

figures

Estimates

figures

Financials

context

Segments

context

Filings

context

News

context

Transcript

context

QVeris earnings research workflow

01SetupEvent, consensus, history, and pre-print debates
  1. Confirm the event, consensus, history, market context, and key pre-print debates
02Print + varianceActuals, financials, segments, and surprise checks
  1. Pull actuals, guidance, segments, margins, cash flow, news, filings, and transcript
  2. Compare actuals vs. consensus, prior quarter, prior estimate, and company guidance
03Call + thesisTranscript signals and assumption changes
  1. Extract tone, guidance language, analyst questions, and dodged management answers
  2. Identify which growth, margin, cash-flow, capex, and segment assumptions changed
04Audit + outputQuality flags, source audit, and reusable surfaces
  1. Flag one-offs, accrual/cash-flow gaps, inventory, backlog, concentration, and risk language
  2. Bind every number to sourceIds; mark missing and conflicts instead of filling gaps
  3. Return preview, flash, call intelligence, share page, and auditable API JSON

Validated payload

varianceTable
callSignals
thesisImpact
qualityFlags
sourceIds

Web UI

output

Share Page

output

API JSON

output

Research path: event setup, print variance, call read, thesis impact, quality check, source audit.

Capability map

earningsCalendarUpcoming & reported events, timing, fiscal period
estimatesConsensus revenue / EPS, growth, estimate count
resultsActuals, margins, and segment highlights
historicalEarningsUp to 8 quarters of reported and estimated EPS
financialsQuarterly income statement, balance sheet, and cash-flow fields
segmentRevenueProduct / segment revenue breakdown when provider coverage exists
quoteDelayed price, day change, and volume
newsTicker-scoped financial news with provenance
filings10-K / 10-Q / 8-K with filing dates and links
transcriptCall intelligence — degrades to unavailable, never invented
Request — POST /api/earnings/analyze
curl -X POST /api/earnings/analyze \
  -H "Content-Type: application/json" \
  -d '{
    "ticker": "NVDA",
    "mode": "auto",
    "includeSources": true,
    "includeTranscript": true
  }'
Response shape
{
  "ticker": "NVDA",
  "mode": "combined",
  "analysis": {
    "summaryBullets": ["…"],
    "confidence": {
      "label": "medium",
      "reason": "Results, estimates, filings and news are available; transcript is unavailable."
    }
  },
  "capabilityStatus": {
    "results": "available",
    "transcript": "unavailable"
  },
  "missing": ["transcript"],
  "conflicts": [],
  "sources": [
    {
      "id": "NVDA-qveris-get_earnings_results",
      "title": "QVeris earnings results",
      "provider": "QVeris",
      "capability": "get_earnings_results",
      "retrievedAt": "2026-07-08T00:00:00Z"
    }
  ]
}
TypeScript
const res = await fetch("/api/earnings/analyze", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ ticker: "NVDA", mode: "auto" }),
});
const report: AnalyzeEarningsResponse = await res.json();

// every numeric claim carries sourceIds — render or refuse
report.sources.length;        // audit trail
report.missing;               // what the agent could NOT get
report.analysis.confidence;   // label + human-readable reason
MCP / agent usage
const packet = await mcp.callTool("qveris_earnings_analyze", {
  ticker: "NVDA",
  mode: "auto",
  includeSources: true,
  includeTranscript: true,
});

return {
  summary: packet.analysis.summaryBullets,
  sources: packet.sources,
  missing: packet.missing,
  confidence: packet.analysis.confidence,
};
Prompt template
Use only the earnings packet JSON.
Required inputs: sources, missing, analysis.confidence.
If a numeric field has no sourceIds resolving to sources, say unavailable.
Mention missing capabilities and confidence.reason. No investment advice.

The audit & confidence model

Sources, always

Every numeric claim carries sourceIds resolving into a deduplicated audit trail with provider, capability, and retrieval timestamp.

Honest gaps

Capabilities degrade explicitly: missing lists what could not be fetched; transcript-derived analysis is withheld when the transcript is absent — never fabricated.

Scored confidence

confidence is high / medium / low with a human-readable reason. Conflicting sources force low confidence and surface in conflicts.

QVeris provides research infrastructure — no buy/sell/hold output, no price targets.

Ready to wire it into your agent?

Get a QVeris key →See a live analysis