Bring Fintent's predictive signals — M&A intent, CapRaise, Distress, and Deal-Maker scores — directly into your Claude conversations using the Model Context Protocol.
Once you connect the Fintent MCP, your AI assistant can pull signal data on demand inside any chat — for one-off lookups, weekly reviews, list scoring, or running sample skills like the Weekly Origination Skill described below.
Ad-hoc lookups — paste a list of companies and ask "What are the current Fintent scores for these?". The assistant calls the MCP and returns scores, trend flags, and the deal-maker firms researching each company.
Saved-screener reviews — ask for this week's high-scoring companies from a specific screener, filter by signal type or threshold, and summarize what changed.
Trends over time — ask how a company or a whole screener has moved over the last year. The assistant pulls up to 12 months of weekly scores and the topics behind them, so it can tell you when a signal started rather than only that it exists today.
Spreadsheets on demand — ask for a screener and its history as an Excel file. It is rendered on our side and handed back as a download link, so a few thousand companies take one request rather than a long back-and-forth.
Screener upkeep — have the assistant build a screener from a list you paste in, add companies as you find them, and clear out the ones that went quiet. It asks before removing anything.
Multi-step skills — paste in a longer workflow prompt and the assistant will run the full sequence using the MCP plus other connectors you have enabled (PitchBook, CapIQ, your CRM, web search).
• A Claude Desktop install (macOS or Windows) — download it here. The MCP setup happens in the desktop app, not the web app.
• A Fintent account — you'll sign in during the connect flow.
• Available today: Claude Desktop. ChatGPT and GitHub Copilot support is in progress — support@fintent.ai can flag your account if you'd like to be notified when those land.
Once connected, your AI assistant can use these tools:
Your assistant can also build and maintain screeners for you — “make me a screener for these twelve targets”, “drop the ones that went quiet”. Removals are flagged as destructive, so Claude asks before it runs one.
Each signal score is on a 0–100 scale. Every row also includes 3-month and 6-month averages, trend flags (new signal, increasing, decreasing), up to three topic keywords explaining the read, and — for Deal-Maker scores — the named firms (PE, IB, VC, law) doing the research.
Setup takes about a minute. Claude Desktop's connector UI handles the OAuth handshake to Fintent — no API keys to copy.
1 Open Claude Desktop and click Customize on the top left
2 Open the Connectors tab
3 Click the + button → Add custom connector
4 Name the connector Fintent and paste the MCP server URL:
5 Save. The Fintent connector now appears in your connectors list — click Connect
6 You'll be prompted to sign in to Fintent. Use the same credentials you use for the Fintent app
7 Fully quit and relaunch Claude Desktop so the new tools are picked up
The custom-connectors flow lives behind the Customize panel in Claude Desktop. If you don't see it, make sure you're on the latest desktop build (Claude Desktop → menu → Check for updates).
We're working on bringing Fintent signals to GitHub Copilot.
Let us know you're interested and we'll notify you when it's ready.
We're working on bringing Fintent signals to ChatGPT.
Let us know you're interested and we'll notify you when it's ready.
Start a new conversation and just ask — you don't need to call the tools by name. A few patterns that work well:
• "What are the current Fintent scores for datadog.com, hubspot.com, and snowflake.com?"
• "Paste my list of 50 portfolio company domains and check which ones have an M&A score above 40."
• "Show me everything Fintent has on acme.com — scores, trends, deal-maker reads, and topics."
• "Pull all companies from my Healthcare Services screener with an M&A score of 50 or higher this week."
• "In my Vertical SaaS screener, which companies trended up on M&A score vs the 3-month average?"
• "Across all my screeners, give me the top 20 distress scores with the new_signal flag."
• "How has acme.com's M&A score moved over the last 12 months? When did it start climbing?"
• "Compare the last year of CapRaise signal for acme.com and globex.com — which one is accelerating?"
• "In my Vertical SaaS screener, which companies have built the most sustained M&A signal since January rather than spiking once?"
• "Create a screener called Q3 Targets with these fifteen domains, and tell me which ones you couldn't match."
• "Add acme.com and globex.com to my Healthcare Services screener."
• "Show me everything in Q3 Targets with no signal at all in the last six months, then remove them."
• "Drop Cake from Q3 Targets." — companies come back with their names, so you can refer to them the way you actually think of them rather than by domain.
• "Download Q3 Targets with 12 months of history as an Excel file." — the workbook is built on our side and you get a link back, rather than the assistant reading the data out screen by screen.
The real leverage shows up when you combine the Fintent MCP with other connectors in the same conversation — PitchBook for financials, your CRM for relationship history, web search for recent news. The sample Weekly Origination Skill below is a full example of that pattern.
Alongside this guide, you should have received a sample skill file (don't have it? email us for a copy):
Fintent_Sample_Weekly_Origination_Prompt_v1.1.txtIt's a multi-stage weekly workflow that uses the Fintent MCP together with whichever enrichment data source you use — PitchBook, CapIQ, web research, or your CRM. The skill walks the user through an interactive intake (including which Fintent signal matters most to them — M&A, CapRaise, Distress, or Deal-Maker reads), runs gating checks, pulls and tiers the week's signals at the threshold they specified, enriches the priority targets, maps action paths (surfacing named Deal-Maker firms as supporting evidence wherever they exist), pauses to capture the user's existing relationships at those firms, and produces a Word memo.
1 Open a new Claude Desktop conversation
2 Make sure the Fintent MCP is connected (see Setup Instructions) plus whichever enrichment connectors you plan to use — PitchBook or CapIQ for financial data, your CRM connector (HubSpot / Salesforce / Affinity / MadeMarket) for relationship history, and web search for open-web research
3 Open the skill file, copy the prompt into a new chat
4 Answer the 12 intake questions one at a time. Claude runs the rest end-to-end with a brief pause to capture your relationships at priority firms
The sample skill is intentionally generic so any deal professional can use it as a starting point. Common customizations: hard-code your default primary signal and threshold to skip those intake questions, swap in your firm's brand colors and disclaimer in Stage 5, or change the deliverable from a Word doc to a Markdown summary if you prefer.
Three things to check: (1) the connector shows as Connected in Customize → Connectors; (2) you fully quit and relaunched Claude Desktop after adding it; (3) you're in a new chat (existing chats don't pick up newly added connectors mid-conversation).
Fintent only returns scores for companies that have generated signal activity. If a domain returns nothing, either Fintent has no recent signal data for that company, or the domain doesn't match a tracked company. Try the company's primary website domain, and check the Fintent app to confirm coverage.
The MCP shares the weekly unique-company quota with your other Fintent integrations. Wait for the quota window to reset, or contact your Fintent account team about raising the limit. Note that exports are never metered, so downloading a screener cannot put you over a limit.
For questions, the sample skill file, or ChatGPT / Copilot notification requests, email support@fintent.ai — we typically reply within a few hours during business days.