Transcript Analysis

Pulls Aiera earnings call and corporate event transcripts via viaNexus and produces a structured breakdown of what management said, what analysts pushed on, and how the tone shifted.

Step 1 — Identify the Event

From the user's message, extract:

• Ticker (required)
• Event window: "latest" (default), "most recent quarter", or a specific date

Step 2 — Find the Event

Call viaNexus:fetch:

• dataset_name: events_calendar
• product: edge
• endpoint: data
• symbols: the ticker
• last: 5 (shows most recent events)

Pick the right eventId / subkey based on eventDate and eventType (prefer earnings unless the user asked for shareholder_meeting, investor_day, etc).

Step 3 — Fetch the Transcript

Call viaNexus:fetch:

• dataset_name: historical_events_transcripts
• product: edge
• endpoint: data
• symbols: the ticker
• subkey: the eventId from Step 2 (or omit and use last: 1 for most recent)

The response contains:

• title, eventDate, eventType, publicShareUrl (user-facing Aiera link)
• transcripts: array of speaker-tagged paragraphs with speakerId, startMs, text
• people: list of named speakers (management + analysts)
• linguistics.sentiment + linguistics.topics / summaries if present
• hasTranscripts: false means the event wasn't captured — fall back to an earlier event

Edge case: transcriptionStatus: "missed" → tell the user the event has no transcript and suggest the prior earnings call.

Step 4 — Structure the Transcript

Split transcripts[] into:

• Prepared remarks — speeches by management (CEO / CFO) before Q&A
• Q&A — alternating analyst questions and executive answers (use people[] to label speakers by name and affiliation)

Step 5 — Synthesize

Output format:

## [Company] ([TICKER]) — [Event Type] — [Event Date]
[Aiera link: publicShareUrl]

### 🎤 Management Highlights (Prepared Remarks)
- **[CEO Name]**: "[direct quote]" — [one-line context]
- **[CFO Name]**: "[direct quote]" — [one-line context]
- 3–5 quotes total, focused on: strategy, product momentum, capital allocation

### 📊 Guidance & Outlook
- Revenue guidance: [what was said, vs prior]
- Margin commentary: [...]
- Segment-level callouts: [...]

### 🎯 Analyst Q&A Themes
Most-asked topics (count / lead analyst):
1. **[Theme]** — asked by [Analyst, Firm]: short paraphrase of push + management's
   answer.
2. ...

### 🎭 Tone
- Overall sentiment: [positive / cautious / defensive] — cite 1 line that
  supports it.
- If `linguistics.sentiment` is populated, report the average score.

### ⚠️ Notable
- Unusual admissions, guidance cuts, new disclosures, leadership changes.

Rules:

• Quote management verbatim — do not paraphrase direct quotes.
• Attribute every quote to a named person (look up via speakerId → people[]).
• If an analyst question is vague, paraphrase; if management's answer is non-committal, flag it.
• Never speculate about tone from the text alone if linguistics.sentiment is absent — just describe word choice.
• Always include the publicShareUrl so the user can replay the audio.

Example Triggers

• "Summarize Apple's last earnings call"
• "What did Jensen say about AI demand on NVDA's Q4 call?"
• "Pull the transcript for Tesla's investor day"
• "Any guidance changes on AMZN's latest call?"
• "What are analysts pushing Meta on?"
• "Find MSFT's most recent earnings transcript"