Integrating Netflix Earnings Data for AI Agents with an MCP Server

Earnings season is here, but investor relations (IR) data is scattered across IR websites, 8-K filings, webcast pages, and conference replays. A developer (also the builder of Equibles) demonstrates how an MCP server can hand this fragmented information to AI agents, turning earnings prep from "open six tabs" into "one conversation."

IR Calendar: What's Coming and What's Archived

The ListInvestorEvents tool returns a company's list of investor events, covering not only earnings calls but also shareholder meetings and investor conferences, with each event's record status annotated (e.g., whether a transcript or slides are available). For example, Netflix's Q3 2026 earnings interview is scheduled for October 20 at 20:45 UTC with status "scheduled," while the July 16 Q2 earnings call has concluded, with both transcript and slides available.

GetUpcomingInvestorEvents is a simplified version that returns only upcoming events and webcast links. What's easy to overlook are non-earnings events—such as the March 4 Morgan Stanley TMT conference attended by the CFO. These occasions are often windows for management to reaffirm or fine-tune guidance between quarters, and the agent can read the record by event ID via GetInvestorEventTranscript.

MCP (Model Context Protocol) is an open protocol introduced by Anthropic in late 2024 to standardize how AI models connect to external data sources and tools. It uses a client-server architecture: AI applications (such as Claude Desktop and Cursor) act as clients, communicating with MCP servers through a unified JSON-RPC interface. Servers encapsulate domain-specific capabilities as "tools" that models invoke on demand during a conversation, eliminating the need to write separate integration code for each data source. Compared with traditional function calling, its advantages are reusability and discoverability—tool descriptions themselves are injected into the model's context to help it decide when to call them. The Equibles MCP server in this article wraps IR data as tools like ListInvestorEvents and GetGuidance for agents to query on demand.

8-K filings are interim reports that the U.S. Securities and Exchange Commission (SEC) requires public companies to file within four business days of a material event, covering earnings releases, executive changes, M&A, major contracts, and more. Unlike the quarterly 10-Q and annual 10-K, 8-Ks are event-driven, so they are often used as the first-hand official source for guidance updates and earnings call notices. The Netflix guidance updates cited in this article are all sourced from 8-Ks, meaning these figures carry legal disclosure weight and are more reliable than press releases. IR teams typically sync 8-K exhibits (such as presentations and press releases) to the IR website, which is also the basis for MCP tools to extract metadata in a structured way.

Guidance History: More Than Just the Latest Numbers

GetGuidance lists each guidance metric along with its source and date. Take Netflix's full-year 2026 guidance: three releases show how the outlook evolved:

Release Date	2026 Revenue	2026 Operating Margin	2026 Free Cash Flow
January 20 (8-K)	$50.7B–$51.7B	31.5%	$11B
April 16 (8-K)	$50.7B–$51.7B	31.5%	$12.5B
July 16 (8-K)	$51.0B–$51.4B	31.5%	$12.5B

The July revenue range narrowed but the midpoint was unchanged, the margin target never moved, and the free cash flow target was raised by $1.5B in April. For Q3 itself, the July guidance called for 12% revenue growth and a 33.2% operating margin—two numbers that are the first checkpoints for Tuesday's earnings call.

Between January and April, CFO Spencer Neumann attended the Morgan Stanley conference on March 4. The agent can read that record to confirm that he reiterated the January plan at the time: 12%–14% revenue growth, a 31.5% operating margin, the ads business doubling to roughly $3B, and free cash flow of about $11B. The $12.5B target did not appear until six weeks later.

One honest limitation: when old filings have not yet been read by the current extraction pipeline, the tool says so explicitly rather than implying there is no guidance. Netflix has an October 2025 filing queued for re-reading, and the output flags this.

Presentations: Metadata and Links

GetInvestorEventSlideMetadata returns metadata about the slides behind an event, including title, source (8-K exhibit), format, page count, and the REST path to the PDF. It returns only metadata and links, not PDF bytes, so it doesn't consume the context window. Code can fetch the PDF from the REST path when pages are needed.

IR News and Tone

GetInvestorRelationsNews lists a company's IR press releases and links, letting the agent retrieve items like "Netflix declines to raise bid for Warner Bros." (February 26) without searching the news.

GetEarningsCallToneAndThemes provides a read on the most recent earnings call: management tone dropped from 78 in Q1 to 65 in Q2 (on a -100 to +100 scale), and engagement was the top theme both times. This context is worth knowing before listening for tone shifts on Tuesday.

A Practical Prompt Example

The author plans to use the following prompt after earnings:

"Using Equibles, compare Netflix's Q3 2026 results with the Q3 guidance issued on July 16 (12% revenue growth, 33.2% operating margin), state whether the full-year 2026 revenue, margin, and free cash flow targets changed, and quote executives on any changes. Provide links for each source."

Explicitly naming Equibles in the prompt encourages the model to call the tools rather than guess.

Setup and Limits

MCP endpoint: https://mcp.equibles.com/mcp. Add it as a custom connector in Claude and sign in to your account—no API key required. Guides for ChatGPT, Cursor, Gemini CLI, and others are in the documentation.

REST API: the same data is available at https://api.equibles.com/v1, e.g., GET /v1/stocks/NFLX/investor-events and GET /v1/stocks/NFLX/ir-news.

Free plan: 100 requests per day combined across MCP and REST.

See the earnings call and investor relations pages in the documentation for the full tool list, plus a recipe for "checking management commitments" against subsequent results.

REST API is an HTTP-based software architectural style that operates on resources through standard methods like GET and POST, with resources identified by URLs and responses typically in JSON. In this article, Equibles offers both MCP and REST access: MCP for natural-language tool calls by AI agents, REST for traditional programmatic integration. Both share the same data backend, so the free plan's 100 requests/day limit is counted across both. For developers who need batch pulls or custom pipelines, REST is more direct; for scenarios where you want the model to decide when to query, MCP is more suitable. Understanding this distinction helps when making architectural trade-offs.

Conclusion

By centralizing the IR calendar, guidance history, call transcripts, presentations, and IR news into a single entry point that AI agents can query through an MCP server, you can significantly reduce manual tab-switching during earnings prep. This approach applies not only to Netflix but can also inform investor relations workflows at other companies. The author also invites readers to share IR preparation steps that aren't yet covered.

One-line Summary

An MCP server consolidates Netflix's scattered IR data into a single entry point that AI agents can query, simplifying earnings prep.

Paragraph Summary

This article explains how an MCP server centralizes Netflix investor relations data (IR calendar, guidance history, earnings call transcripts, presentations, and IR news) into a single entry point that AI agents can query, turning earnings prep from multi-tab switching into one conversation. It details the capabilities of tools like ListInvestorEvents, GetGuidance, and GetEarningsCallToneAndThemes, and uses Netflix's three 2026 guidance releases to illustrate how to track the evolution of the outlook. The tools are designed for context efficiency, returning metadata rather than full files. The author also provides practical prompt examples and setup guidance, emphasizing that naming the tool explicitly in the prompt encourages the model to call it rather than guess. The free plan allows 100 requests per day and supports both MCP and REST access.