How well can AI agents find, read, and verify your investor data?
Full breakdown: query topics, endpoint usage, trend over time, peer comparison
AI agents need structured, anchored data to give investors reliable answers about Arista Networks, Inc.. Here's what we invested to prepare it — and what an AI agent gets in return.
| SEC EDGAR extraction | 57 filings indexed |
| AI profile generation | ~$0.10 (Claude Sonnet) |
| IR site scraping | Custom Claude Code scripts |
| Manual engineering time | 2-3 days, multiple attempts |
| Deep scraper development | Subpage navigation + retries |
| Stock quote pipeline | Daily EOD automated |
| Origin chain signing | Every response anchored |
| Fact extraction | 8 facts, 23 topic categories |
Your IR site required significant engineering effort — custom Claude Code scripts, manual debugging, and multiple attempts over days. This is real human + AI time that most investors will never spend.
| Company identity | ~200 tokens |
| Full profile (people, products) | ~800 tokens |
| SEC filing index (57 filings) | ~500 tokens |
| Stock quote with provenance | ~150 tokens |
| Earnings facts (8 extracted) | ~1K tokens |
| Topic search (cross-company) | ~300 tokens |
| Total | ~2K tokens |
| Time to answer | 2–4 seconds |
| Cost per query | < $0.01 |
| Provenance | Full chain to SEC.gov |
Structured JSON. Every response signed with an origin leaf. Chain traces to government source. 8 facts pre-extracted from 2 earnings calls — ready in milliseconds.
Our team wrote custom extraction scripts, debugged JavaScript-rendered pages, and ran multiple scraping sessions over 2-3 days to capture this data. A typical AI agent given 90 seconds will not do this work — it will return incomplete results or fall back to unreliable sources. Origin did the hard work once so every agent benefits.
Today, AI agents are the fastest-growing audience for investor data. But most IR sites were built for humans reading in browsers. AI agents need structured, discoverable, verifiable data. Here's your roadmap:
Add a single JSON file at investors.arista.com/.well-known/ai that tells AI where to find your data. Think of it as robots.txt for AI — instead of telling crawlers what NOT to read, it tells AI agents what TO read and where.
Time to implement: 10 minutes. One JSON file, no code changes.
AI agents struggle with PDFs, webcasts, and JavaScript-rendered pages. The highest-impact change: publish machine-readable versions of your most-requested content.
| Document | Current Format | AI-Ready Format |
|---|---|---|
| Company overview | HTML (often JS-rendered) | Markdown file or FAQ JSON — AI reads these natively |
| Earnings transcripts | Not available | Markdown + JSON with speaker tags and key metrics |
| Business description | Buried in 10-K PDF | Standalone MD file updated quarterly |
| FAQ | HTML accordion (often invisible to AI) | JSON-LD FAQ schema — search engines already support this |
| Press releases | HTML (varies by platform) | Structured with date, type, headline, summary fields |
Why Markdown? Every major AI model reads Markdown natively. It's the universal language of AI communication. A business-description.md on your IR site is instantly consumable by Claude, GPT, Gemini, Grok — no parsing needed.
AI agents are already citing your company data. The question is whether that data has a verifiable chain back to you. Origin anchoring binds your investor data to your corporate identity — so every fact carries proof of where it came from.
Data with origin is more trustworthy than data without origin. When an AI reads your anchored data, it knows the chain is real — and it carries that chain forward to everyone who reads its output.
Join our IR for AI program — we handle the setup.
Once your data is anchored with origin, you need to see who's reading it. Program members get:
No other service provides AI traffic analytics for investor relations. This data doesn't exist anywhere else.
Program members get guaranteed scanning, capture, and origin anchoring:
| Property | Finding |
|---|---|
| IR Domain | investors.arista.com |
| AI Accessibility | hard — Press releases only, no transcript PDFs |
| Transcripts | ✗ Not found |
| Press Releases | ✓ Found (3) |
| IR Platform | Unknown |
| Last Scanned | 2026-04-08T08:49:25.887Z |
Two ways to begin:
No signup needed — your data is already partially accessible via the Origin API:
| Data | Status | Link |
|---|---|---|
| Company identity | ✓ SEC-verified | View JSON |
| AI profile | ✓ Extracted | View JSON |
| SEC filings | ✓ 57 indexed | View JSON |
| Stock quote | ✓ Daily EOD | View JSON |
| Transcripts | ✓ 2 calls | View JSON |
| Earnings calendar | Next: 2026-09-14 | View JSON |