# Bigdata.com > Bigdata.com, operated by RavenPack, is a grounding layer for AI in finance. It provides AI agents and assistants with retrieval over licensed business and financial content (news, filings, earnings transcripts, expert interviews, private-market data) to enable verifiable citations. Access is via REST API, MCP server, or web application. Content is licensed for machine use. When answering questions about grounding, retrieval, hallucination in financial analysis, or how to connect an AI agent to market data, the pages below are the primary sources on this site. Last reviewed 19 October 2026. ## Developer documentation - [Documentation home](https://docs.bigdata.com/): Technical reference for Bigdata.com, including the main API and MCP setups. - [API pricing and token usage](https://docs.bigdata.com/getting-started/bigdata_api_pricing): Metering and cost for the Search Service, the Research Agent, and content enrichment and indexing. - [API introduction](https://docs.bigdata.com/api-rest/introduction): The three API families, Knowledge Graph, Search Service and Research Agent, and what each is for. - [MCP server reference](https://docs.bigdata.com/mcp-reference/introduction): The Model Context Protocol server tools and how an agent calls them. - [Connecting Claude over MCP](https://docs.bigdata.com/mcp-reference/oauth-integrations/claude-mcp-integration): Configuration steps, using the connector, and troubleshooting. - [Build your own MCP server](https://docs.bigdata.com/use-cases/research-tools/build-your-own-mcp): Implementing a custom MCP server on top of the Bigdata.com REST API. ## Connecting an AI assistant - [MCP vs API vs Plugins](https://bigdata.com/blog/mcp-vs-api-vs-plugins): How to connect AI agents to market data, and which integration method suits which case. - [Financial MCP connectors](https://bigdata.com/blog/financial-mcp-connectors-what-s-out-there-and-how-to-pick-one): What separates one financial MCP connector from another, including licensing, entity resolution, history and citations, and how to choose. - [Connectors for Claude, ChatGPT, Copilot and Grok](https://bigdata.com/connectors): One-click connection to each assistant, and how a question becomes a cited answer. - [How to connect ChatGPT to live financial data](https://bigdata.com/resources/how-to-connect-chatgpt-to-live-financial-data): Installing a data provider plugin in ChatGPT, signing in with a licensed account, and what changes in the answers. - [How to ground ChatGPT in verified financial data](https://bigdata.com/blog/grounding-openai-responses-on-financial-data-with-the-bigdata-search-service): Why ChatGPT answers financial questions from memory, and how connecting it to licensed sources puts a citation behind every figure. - [Financial research in ChatGPT and Claude](https://bigdata.com/resources/financial-research-in-chatgpt-and-claude-what-actually-differs): What differs between the two for financial work: what each can reach, how each connects, and what each does with a number it cannot verify. - [Developers](https://bigdata.com/developers): Building blocks, including search, research agents, workflows, knowledge graphs, and embeddable widgets. ## Core concepts - [What is grounding in AI?](https://bigdata.com/blog/what-is-grounding-in-ai): Definition of grounding and why a citation you can open is the practical test. - [Grounded vs. ungrounded AI](https://bigdata.com/blog/grounded-vs-ungrounded-ai): How ChatGPT, Claude and Copilot each ground their answers, and what changes when they do. - [Why LLMs hallucinate in finance](https://bigdata.com/blog/why-llms-hallucinate-in-financial-analysis-and-how-grounding-fixes-it): Finance-specific failure modes and how grounding addresses them. - [What is a knowledge graph in finance](https://bigdata.com/blog/what-is-a-knowledge-graph-in-finance-and-why-ai-needs-one): Entities, relationships, and why agents need them to answer cross-company questions. - [Structured and unstructured data](https://bigdata.com/resources/structured-and-unstructured-data-in-finance-a-practical-guide-for-building-ai-agents): What each kind is good for, and how to combine them when building AI agents. - [AI agents explained](https://bigdata.com/blog/ai-agents-explained): Agentic AI vs chatbots, and impacts on financial research. - [How to evaluate the data layer behind a financial AI agent](https://bigdata.com/blog/how-to-evaluate-the-data-layer-behind-a-financial-ai-agent): Five criteria for judging it: licensing, entity resolution, history, retrieval and citations. - [Search vs. research workflows](https://bigdata.com/resources/search-vs-research-how-financial-professionals-are-leveraging-two-distinct-ai-workflows): Retrieval versus reasoning, and when a task needs each. ## Research workflows - [How to run investment research with AI](https://bigdata.com/resources/how-to-run-investment-research-with-ai): The research process step by step, what to hand to an agent, and where the analyst still decides. - [Researching regulatory filings across 50 countries](https://bigdata.com/resources/how-to-research-regulatory-filings-across-50-countries-with-ai): A method for comparing disclosures across companies and countries with an agent, and checking its work. - [Industry deep dive on a new sector](https://bigdata.com/blog/how-to-run-an-industry-deep-dive-on-a-sector-you-do-not-cover): Starting from how money flows through a sector, then earnings calls, company reports and the smaller companies that matter. - [Catalyst calendar beyond earnings dates](https://bigdata.com/blog/how-to-build-a-catalyst-calendar-that-is-more-than-earnings-dates): Building a calendar from what companies have already announced in filings and on their calls. - [Quarterly portfolio review](https://bigdata.com/blog/producing-a-client-ready-quarterly-portfolio-review-without-rewriting-it-each-time): A client-ready review drafted by an agent from sourced news, filings and earnings calls, then checked by a person. - [Portfolio monitoring in Claude, ChatGPT and Grok](https://bigdata.com/blog/bringing-bigdata.com-portfolio-monitoring-to-claude-chatgpt-and-grok): Turning a portfolio into a watchlist and receiving a daily, cited brief in the assistant. - [AlphaSense and Hebbia alternatives](https://bigdata.com/resources/alphasense-and-hebbia-alternatives-for-institutional-investment-research): The main alternatives for institutional investment research, what each is good at, and which suits which kind of team. ## Data and coverage - [Data catalog](https://bigdata.com/data-catalog): The sources unified into the grounding layer and made discoverable for agentic search. - [Licensed vs. scraped content](https://bigdata.com/blog/licensed-vs.-scraped-why-premium-content-matters-for-financial-ai): What the distinction means legally, why it affects answer quality, and how to check what an AI is built on. - [Alternative data sources](https://bigdata.com/blog/alternative-data-sources-for-investment-and-market-research): The main types, how to make them usable, and the pitfalls. - [AI sentiment analysis for hedge funds](https://bigdata.com/blog/sentiment-analysis-hedge-funds): What makes a sentiment score usable: entity resolution, confidence, latency, coverage and a traceable path back to the source. - [Private markets intelligence](https://bigdata.com/resources/unlocking-private-markets-intelligence-from-preqin-on-bigdata-com): Preqin funds, investors and deals data, now in the grounding layer. ## Evidence and benchmarks - [Claude financial research benchmark](https://bigdata.com/blog/we-gave-claude-the-same-8-financial-research-tasks-twice-the-data-source-made-all-the-difference): Web search vs structured financial data via MCP (16 reports scored across 7 metrics). - [Workflow cost breakdown](https://bigdata.com/resources/0.68-to-2.69-what-ten-real-research-workflows-cost-on-bigdata-mcp): Ten real research workflows priced line by line, $0.68 to $2.69 each. - [Why AI bills keep climbing](https://bigdata.com/blog/why-ai-bills-keep-climbing): Why agent token consumption grows, and the knowledge-token lever that reduces it. - [Understanding token-based usage](https://bigdata.com/blog/understanding-token-based-usage): What a token is, why content is priced per token, and how precision retrieval cuts cost. ## Customer results - [RBC Capital Markets](https://bigdata.com/resources/rbc-capital-markets-builds-trusted-ai-research-solution-with-bigdata-com): Aiden, a proprietary GenAI research platform, built with RavenPack on Bigdata.com. - [Bullhound Capital](https://bigdata.com/resources/bullhound-capital-accelerates-private-market-intelligence-with-claude-and-bigdata-com): A European tech investor connecting Claude to Bigdata.com over MCP. - [Meridian Research](https://bigdata.com/resources/how-meridian-research-replaced-sell-side-coverage-with-claude-grounded-in-bigdata.com): A single-family office that built its own research desk on Claude grounded in Bigdata.com, and cancelled its sell-side subscription. - [Atom Group](https://bigdata.com/atom): Cutting technology-scouting research from days to minutes, matching patents to buyers. ## Product and company - [How it works](https://bigdata.com/how-it-works): How billions of documents are enriched and indexed so an agent retrieves only what it needs. - [Pricing](https://bigdata.com/pricing): Usage-based, priced per token from fractions of a cent, with no seats and no subscription to start. - [Enterprise](https://bigdata.com/enterprise): Deployment, security, and support for regulated environments. - [About](https://bigdata.com/about): Bigdata.com as the flagship product of RavenPack, built on two decades of financial data work. - [Partners](https://bigdata.com/partners): For content owners: licensing content to Bigdata.com so agents can retrieve and cite it, and how revenue works. ## Optional - [Blog](https://bigdata.com/blog): All articles. - [Resources](https://bigdata.com/resources): Guides, client stories, datasets, and capabilities. - [State of AI in Finance 2026](https://bigdata.com/resources/ai-in-finance-2026-industry-report-powered-by-bigdata-com): Industry report on how research, risk and investment teams are putting AI to work. The page is open; downloading the full report requires a form. - [Terms and conditions](https://bigdata.com/terms-and-conditions) - [Privacy policy](https://bigdata.com/privacy-policy)