FactSet MCP review: 24 tools on one endpoint, gated by your FactSet license

By Jan Suski. I go deep on every review, checking every number against its source and digging up details other coverage misses, a habit from six years of SEO and link building for tech companies. Bylines at Surfer, Brand24, and Senuto. How the scoring works.

Published 15 October 2026

Financial data

The FactSet MCP is FactSet's Model Context Protocol (MCP) server: a remote endpoint that lets an AI assistant call FactSet's content APIs by name, so a prompt in Claude or ChatGPT turns into a query against FactSet financial data such as fundamentals, consensus estimates or fund holdings. It went into production on 16 December 2025, and the tool list has grown from nine datasets to 24 tools in ten months.

This review covers what FactSet's MCP server reaches, how a firm connects it, what governance comes with it, what it costs, and how it compares with the MCP routes S&P Global, LSEG, Morningstar and AlphaSense have built. The FactSet review covers the Workstation itself; this page stays on the AI connection.

It's written for data or analytics teams at firms that already hold a FactSet contract. Anyone without one hits the same wall at step one.

↑ FACT SHEETFinancial data FactSet's MCP server connects to

The AI-Ready Data MCP lists 24 tools on FactSet's developer portal, and its overview claims coverage of "116 countries and 86,000+ companies." Each tool maps to one FactSet content API, so an agent picks FactSet_Ownership for fund holdings & institutional ownership, or FactSet_People for executive profiles, compensation and employment history.

Nine key datasets made the launch release, which promised "seamless, unified access" to them through one server. The portal now carries 15 more tools, from debt capital structure to power plant data across major ISOs.

The tool descriptions carry the limits a buyer needs. Here's what the portal states for the tools analysts reach for first:

• FactSet_GlobalPrices: stock prices, returns, volume, corporate actions and dividends back to 2006
• FactSet_EstimatesConsensus: consensus estimates, earnings surprises, analyst ratings and company guidance
• FactSet_EntityReference: business profiles and parent relationships for 16 million+ entities
• FactSet_CalendarEvents: company events for a window of up to 90 days
• FactSet_PrivateCompany: financials, executive profiles and funding for private companies
• FactSet_SupplyChain: direct and reverse supply chain relationships
• FactSet_UnstructuredContent: vectorized CallStreet transcripts, StreetAccount News, MT Newswire and EDGAR 10-K, 10-Q and 8-K filings from 2022
• FactSet_QFL: 2,000+ quantitative factors

Unstructured content comes from FactSet's Vector Data API, wired into the server. FactSet's MCP product page says the API "handles the processing, chunking, embedding, and indexing" of financial documents, which is the RAG pipeline a firm would otherwise build and run itself.

The devil's in the details on the dataset count. The December release listed Geographic Revenue Exposure among the nine; the product page lists Entity Reference in that slot. Both exist as tools today, so the count of nine reads as a launch figure, and 24 is the number that matches what an agent sees.

↑ FACT SHEETSetting up the FactSet Model Context Protocol (MCP) server in Claude, ChatGPT and Databricks

Every client connects to the same Model Context Protocol (MCP) endpoint, https://mcp.factset.com/content/v1. The portal names 12 AI applications plus a slot for any proprietary MCP client:

• Claude and Claude Code
• ChatGPT and Codex CLI
• Microsoft Copilot and Copilot Studio
• Gemini and Google Antigravity
• Databricks, Amazon Quick, Glean and Perplexity

In Claude, Anthropic's FactSet tutorial has an organization owner open Admin settings, choose Connectors, add a custom connector and paste the URL a FactSet representative provides. The same page labels the integration early access. A firm with a signed contract can hit the ground running; the gate is the contract.

FactSet's FAQ adds support for "DCR-enabled environments and major authentication frameworks," and notes that "some limitations may apply depending on the platform." Third-party directory mcpservers.org lists a one-line install, npx add-mcp, for Claude Code, Codex and Cursor.

The developer page says FactSet's demo in Claude Code shows prompts "cutting integration time from weeks to minutes." That's the vendor's framing of the plug-and-play claim; the measurable part is that a developer writes zero connector code, because the server already exposes each FactSet API as a named tool and turns natural language into precise API calls.

↑ FACT SHEETGovernance, entitlements and audit links for reliable data

Governance is the part FactSet sells hardest, and the product page puts it in one line: data "remains fully permissioned, auditable, and compliant," with "built-in user and content entitlement controls." In practice the server checks what your firm licenses before it answers, which is why tool availability "now reflect[s] your firm type as well as your data tier," per changelog version 1.15.

Audit links are the concrete feature. Version 1.16 added them to eight tools, including Global Prices, Ownership and People, and version 1.17 added them to ten more, from Private Company to Terms & Conditions. Each answer can then point back to the FactSet record behind the number, which keeps regulatory alignment from depending on an analyst's memory.

On model training, the developer FAQ states that FactSet "does not share client, confidential, or proprietary data with public generative AI models." That covers FactSet's own AI tools. What Claude or ChatGPT does with a response sits under your firm's agreement with that AI vendor, so read both contracts.

The Claude tutorial flags two data caveats worth knowing before an agent writes a client memo: ownership data carries quarterly lags, and insider transactions arrive with filing delays.

↑ FACT SHEETPortfolio Analytics MCP and the AI-Ready Data MCP in Gemini Enterprise

FactSet runs a second server for its analytics. The Portfolio Analytics MCP entered limited release on 26 June 2026, and the launch release describes pre-calculated performance, attribution and risk outputs delivered into a client's private LLM environment.

FactSet calls those outputs "relied upon as a book of record by investment teams globally." For a compliance reviewer, that's the line that matters: the AI reads an approved attribution figure, and it never recalculates one from raw holdings.

On 25 August 2026 FactSet went live in Gemini Enterprise for Financial Services, with coverage across portfolio analytics, fixed income, derivatives, scenario analysis, collateral and hedge tools. Early pilots ran at global investment banks, commodities trading desks and fixed income teams.

FactSet's blog describes these as "custom-deployed connectors" configured inside each firm's environment, which puts every Gemini rollout on FactSet's implementation calendar.

↑ FACT SHEETWhat the FactSet MCP costs and who can get it

FactSet's pricing page carries one headline and a "Contact Us For A Personalized Proposal" button. The MCP has no separate list price; the server is a delivery channel for data a firm already licenses.

The closest number is the subscription itself. Vendr's anonymised buyer data puts the median FactSet contract at $25,160 a year, in a tracked range of $4,200 to $28,179, and notes that API access and data redistribution can carry separate licensing fees. The FactSet pricing report breaks that range down by seat and module.

Packaging decides what the agent sees. Changelog version 1.15 moved Private Company "to the Premium package," and tool descriptions now "adapt to your AI-Ready data tier." Two firms on the same endpoint can see different tool lists.

A trial goes through the FAQ's route: contact sales, write to support@factset.com, or ask your account representative. The product page's "Try for free" button leads to the same contact form.

↑ FACT SHEETWhere FactSet's AI-ready data earns its keep

The MCP serves the same datasets FactSet Workstation users rate, so their reviews describe what the agent gets. FactSet Workstation sits at 8.6 out of 10 across 73 reviews and ratings on TrustRadius and 4.2 out of 5 from 72 reviews on G2.

Aggregation is the praise that maps straight onto an AI workflow. A partner at an 11-50 person finance firm wrote on TrustRadius in September 2025:

"The main problem FactSet solves for our firm is the aggregation of data into one platform. That feature is a huge time saver compared to piece-meal data from several sources."

TrustRadius review, Partner in Finance and Accounting, September 2025

That's the copy-paste problem the MCP targets, and reliable data in one place is what makes an agent's answer checkable. Jonathan Wellum, a financial analyst at ROCKLINC Investment Partners, described the manual version in an October 2025 review:

"By having one place to go, I can keep up-to-date on every equity security and save time from jumping back and forth between a company's investor relations page and other sources during earnings season."

TrustRadius review, Jonathan Wellum, Financial Analyst, October 2025

Formula refresh is the Excel-era version of the same saving. A finance manager at a 10,001+ employee company listed "Excel Plug In with FactSet Workstation refresh, saves so much time" as the top pro in an October 2025 TrustRadius review. Microsoft Copilot through the MCP pushes that refresh into Excel and Word from a prompt.

↑ FACT SHEETWhere users say FactSet falls short

AI was the gap reviewers flagged before the MCP shipped. Wellum's con, from the same October 2025 review, reads: "AI Capabilities, where FactSet Workstation can quickly provide me with information and consolidate information it holds on its platform." The MCP is FactSet's answer to that request, routed through someone else's chatbot.

Cost and outside data come up together. The same September 2025 partner review says:

"FactSet would not be appropriate for smaller [managers] due to its high cost. Also, if you're looking to integrate outside data sources into your FactSet jobs, it can be very difficult. We've had better luck using Bloomberg for that."

TrustRadius review, Partner in Finance and Accounting, September 2025

Support is the most repeated complaint, and it's a red flag for a product that depends on FactSet staff to configure connectors. An executive at a 1-10 person firm rated the Workstation 5 out of 10 in October 2025:

"The problem is that it doesnt work at 100% all the time, and FactSet Workstation tech team sometimes doesnt know how to fix the issues."

TrustRadius review, Executive in Finance and Accounting, October 2025

The October 2026 version of the complaint is shorter. A reviewer at a 51-200 person firm wrote that "Client support needs to be drastically improved in terms of their expertise," in a TrustRadius review dated October 2026.

The add-in formulas draw fire too. A director managing equity portfolios wrote in October 2025 that "Excel add-in formulas are not constructed in an intuitive way and too often require assistance from help desk," per his TrustRadius review. A natural-language layer over the same APIs removes the formula syntax from the analyst's side of the desk.

↑ FACT SHEETWhat FactSet publishes about its own MCP

FactSet's own words carry most of the claims buyers will hear in a demo, so here they are straight from the horse's mouth. The 16 December 2025 release describes FactSet as "a global financial digital platform and enterprise solutions provider" and the server as "enabling AI systems to reason over trusted FactSet data without intermediaries, previews, or custom integrations."

It positions the product against two rival shapes: "Unlike demo MCP servers or warehouse-dependent offerings, FactSet's MCP server provides direct, governed, production access to financial intelligence." The release adds that a "native MCP server" lets models "interact directly with FactSet's curated financial datasets, bringing authoritative market intelligence into AI workflows without warehouses, exports, or manual pipelines."

John Costigan, FactSet's chief data officer, is quoted on the APIs behind it:

"Backed by APIs already trusted and actively used by hundreds of FactSet's institutional clients for mission-critical workflows, this MCP solution extends those capabilities to agentic and other enterprise deployments via its unified, AI- and cloud-ready capabilities."

John Costigan, Chief Data Officer, FactSet, 16 December 2025 release

He follows with "Initial client feedback underscores the transformative potential of this expanded access. And this is just the beginning," and promises an "aggressive roadmap" to keep "FactSet clients at the forefront of market, data, and AI-driven innovation across the financial sector and beyond."

Its target audience is spelled out: "FactSet's MCP server offers new value for CTOs, CIOs, innovation leaders building enterprise AI strategies, data or analytics teams, and corporate development professionals and strategist to easily integrate real market data into research, scenario planning, and other AI-powered workflows."

The release also cites the beta: "early client feedback has underscored its transformative potential," after 45 firms and over 800 institutional users joined.

FactSet's boilerplate opens: "FactSet (NYSE:FDS | NASDAQ:FDS) supercharges financial intelligence, offering enterprise data and information solutions."

It goes on to say its "cutting-edge digital platform seamlessly integrates proprietary financial data, client datasets, third-party sources, and flexible technology to deliver tailored solutions across the buy-side, sell-side, wealth management, private equity, and corporate sectors." It closes on "advanced data connectivity" to "enable smarter, faster decision-making" and "long-term client success."

On the developer page, FactSet writes that "With comprehensive coverage and reliable data, FactSet empowers smarter financial analysis." The release links readers to "FactSet's MCP solution and other AI offerings." Your own entitlements test each claim in week one.

↑ FACT SHEETRelease history and the business behind it

FactSet's fiscal 2026 results, filed with the SEC on 30 September 2026, put revenue at $2,476.3 million, up 6.7%, and organic ASV at $2,568.2 million, up 7.0%. The same filing says "AI solutions ASV added in fiscal 2026 more than doubled year over year," without a dollar figure.

Guidance is the number to keep an eye on for anyone whose decision-making depends on FactSet's roadmap: organic ASV growth of 5.0% to 6.5% for fiscal 2027. A vendor guiding slower growth while it ships AI products at this pace is betting the MCP holds clients who might otherwise build their own pipelines.

The FAQ's roadmap list lags the changelog. It still names Debt Capital Structure, Private Company Financials and Deep Sector Power & Utilities as possible future areas, and all three already exist as tools in version 1.17.

↑ FACT SHEETFactSet MCP against S&P Global, LSEG, Morningstar and AlphaSense

Four data vendors have built AI connectors to their licensed content. Each one only makes sense for a firm already paying that vendor, so the comparison below is about which data your agents need; the ranking of financial data providers covers the underlying subscriptions.

S&P Global (Kensho MCP into Claude)

S&P Global announced on 15 July 2025 that Kensho built an MCP server for its LLM-ready API, putting Capital IQ Financials and earnings call transcripts into Claude. It wins for a team whose models already run on Capital IQ. It loses on breadth of named clients: FactSet lists 12, and S&P's launch centred on Claude.

A TrustRadius reviewer who switched explains the pull in the other direction: "My firm was using CapitalIQ and I found the solution to be meaningfully weaker and lobbied to change," per an October 2025 review.

Not for: a desk that needs FactSet's supply chain and GeoRev tables in the same prompt.

LSEG (MCP through Copilot Studio)

LSEG launched an LSEG-managed MCP server on 13 October 2025 for agents built in Microsoft Copilot Studio, citing more than 33 petabytes of AI Ready Content, with a phased rollout starting at LSEG Financial Analytics. It wins in a Microsoft-standardised bank with FX and rates desks. It loses for a firm that wants one endpoint across Claude, ChatGPT and Databricks.

The LSEG review covers Workspace pricing and the Eikon migration. Not for: an equity research team built around FactSet screening.

Morningstar MCP server

Morningstar's developer documentation lists seven tools, including analyst research, fund holdings, X-Ray portfolio analysis and a screener, at https://mcp.morningstar.com/mcp, with a licensed Morningstar account required. It wins for wealth management and fund research, where Morningstar's analyst reports carry the decision. It loses on company-level depth: seven tools against FactSet's 24, with no supply chain or M&A tool listed.

Not for: corporate development professionals screening deal targets.

AlphaSense

AlphaSense's developer docs describe a hosted connector in beta and a sample self-hosted server, each exposing two tools, with searches priced in credits: 10 for a fast search, 25 for thinkLonger and 100 for deepResearch, which takes 12 to 15 minutes. It wins for reading broker research and transcripts. It loses on structured numbers, where FactSet's estimates and pricing tools return tables.

An investor relations VP compared the two directly in an October 2025 TrustRadius review:

"I believe that the tool's efforts towards becoming more integrated with AI are strong and that in time the Mercury offering will be just as competitive as some of the other AI offerings out there today such as AlphaSense."

TrustRadius review, Vice-President in Finance and Accounting, October 2025

See the AlphaSense MCP review, the AlphaSense review and the AlphaSense pricing report. Not for: a quant team that needs 2,000+ factors in a tool call.

Bloomberg is the benchmark FactSet gets priced against; the FactSet vs Bloomberg comparison and the Bloomberg Terminal pricing report carry those numbers. A senior equity analyst at a 10,001+ employee firm wished FactSet could "bridge some of the gaps a Bloomberg terminal can provide such as the search function and most importantly the chat function," per an October 2025 review.

Pairs well with FactSet MCP: Crunchbase for startup funding rounds, and a web-data MCP such as Bright Data MCP for sources outside licensed feeds. Other servers sit in the MCP tools list, including the DataForSEO MCP.

↑ FACT SHEETFAQ

How much does FactSet MCP cost?

FactSet doesn't list an MCP price. Access comes with a FactSet financial data license, and Vendr's anonymised buyer data puts the median FactSet contract at $25,160 a year. Tools shown depend on your firm's data tier.

Why is FactSet falling?

FactSet's own guidance shows slower growth: organic ASV rose 7.0% in fiscal 2026, and the company guided 5.0% to 6.5% for fiscal 2027. The same filing says AI solutions ASV added in fiscal 2026 more than doubled, so the MCP sits in the one line FactSet reports as accelerating.

How much does the CEO of FactSet make?

Sanoke Viswanathan's employment terms filed with the SEC set a $1,000,000 base salary, a target bonus of 200% of salary, and a one-time $22,000,000 performance option for fiscal 2026 and 2027. Make-whole awards add $13,000,000 in cash, $26,000,000 in restricted stock units and $10,000,000 in performance units. The annual equity target from fiscal 2028 is $11,000,000.

Does JP Morgan use FactSet?

Yes, as a partner. FactSet and J.P. Morgan Securities Services announced a data management and performance collaboration in December 2024, and on 29 April 2026 they launched Whole Portfolio Distribution, a FactSet managed service powered by Fusion by J.P. Morgan. Viswanathan joined FactSet from JPMorgan.

↑ FACT SHEETVerdict

The FactSet MCP is for buy-side, banking and corporate development teams that already license FactSet and want their AI tools reading the same estimates, ownership and pricing data, with entitlements and audit links attached. One URL serves 12 named clients, and 24 tools cover the datasets most research prompts touch.

It isn't for anyone outside a FactSet contract. The pricing page asks for a proposal, the Vendr median sits at $25,160 a year, and a reviewer called FactSet inappropriate for smaller managers "due to its high cost."

It might suit a mid-sized manager with caveats: confirm which tools your data tier exposes, get the connector configured before renewal talks, and test support response times, because reviewers in 2025 and 2026 both flagged FactSet support. Equity research desks get the most from it; a quant team checks the QFL tool first, and a fixed income desk checks the Gemini and Portfolio Analytics routes.

The fact sheet scores follow from those facts: 70 on time to first output because setup waits on a FactSet representative, 60 on documentation because authentication details are thin next to a public tool list and changelog.