Agentic deal sourcing and investment intelligence platform

At a Glance

Company. BlackFin is a leading fintech-focused venture capital fund investing in early-stage companies across Europe. The fund manages a portfolio of fintech companies and runs an AI-driven deal sourcing operation built around LinkedIn signal detection.

The Problem. Venture capital is a competitive market where getting in touch with founders in the early days of their journey can be a strong value driver. BlackFin built a proprietary AI sourcing and deal qualification engine to maximize its coverage of European fintech founders. BlackFin's sourcing process surfaces thousands of LinkedIn signals per week. Each signal needs rapid qualification on three dimensions: fintech or not, creation date, geography. But the volume made manual enrichment impossible. At the same time, their CRM data was too sparse for AI agents to query reliably, and weekly market intelligence required manual news review.

Implementation. Linkup's /search API embedded across three workflows: signal qualification at sourcing, CRM enrichment, and an automated weekly fintech news briefing.

Key Results

• Thousands of weekly signals qualified automatically
• Market intelligence fully automated
• Security architecture that keeps internal knowledge isolated from web-facing agents

"As soon as you query the web, it's really better than LLM providers’ native solutions. It's better, more reliable and drastically less expensive." - Guillaume Fazekas, Investor at BlackFin.

BlackFin operates at the intersection of deal velocity and data precision. As a fintech-focused fund, every week brings thousands of signals: LinkedIn connections, company movements, funding announcements, and market trends. Any of them could represent the next investment. The question is always the same: which ones actually matter?

To answer that question at scale, BlackFin built an AI-powered sourcing and intelligence operation. At the center of it is Linkup.

The Challenge

BlackFin's sourcing engine identifies thousands of LinkedIn signals per month. Companies that appear in the connection graphs of relevant founders, operators, or investors. Each signal is a potential deal, but a raw LinkedIn profile is never enough to surface complex and precise signals.

Those data points enable the team to focus on the opportunities most aligned with its investment mandate and thesis. Without them, the signal is noise.

Manual enrichment at this volume was never going to work. Neither was relying on LinkedIn's own data, which is frequently incomplete or outdated. The team needed a way to resolve these questions automatically with enough accuracy to trust the filter.

The same data quality problem showed up in their CRM (Attio). BlackFin tracks ~10,000 companies. For years, sparse CRM records weren't a critical issue: humans querying the CRM can work around gaps. But once AI agents became the primary interface for querying deal data, gaps became failures. An AI agent querying an under-enriched CRM loses context on every missing field. The CRM had to be AI-ready.

A third problem was simpler but equally persistent: every week someone on the team spent time manually reviewing market news so the team could access market developments and insights.

Why They Chose Linkup

BlackFin's requirements turned out to be a sharp test of search quality. A misclassified signal isn't a small inefficiency, it can be a missed investment.

That made classification and intent signaling the most demanding part of the workflow. Determining whether a company is genuinely interesting and matches BlackFin's investment strategy requires reading and interpreting company descriptions. Not just matching keywords. The same applies to intent signals: separating a real founder move from background noise on LinkedIn.

For platforms where context shapes investment decisions, Linkup turns the open web into a live structured input. At the speed and reliability a leading investment firm needs.

Two additional factors mattered beyond search quality.

• The first was security architecture. BlackFin was building an internal knowledge wiki. Condensing their data, company research, and deal knowledge. All structured into a context layer for AI agents. The risk with a general-purpose agent is bidirectionality: an AI with access to both internal documents and the open web can in theory leak internal context outward. Linkup creates a clean separation. The information flow goes in one direction only. Web intelligence in. Never internal data out. For a fund handling sensitive portfolio information, that boundary matters.
• The second was domain access. Claude and ChatGPT operating natively are constantly blocked by paywalled and restricted domains, producing poor results for news and market research. Linkup bypasses those restrictions, which turned out to be the direct cause of the news workflow problems before the integration.

Key reasons:

• Classification precision at the volume BlackFin's sourcing engine requires
• One-way information architecture. Linkup separates web retrieval from internal context and prevents agent data leakage
• Access to domains that native LLM web search can't reach
• No infrastructure to maintain. Linkup handles retrieval so the team builds on top of results. Not on top of a crawler
• Native integration with Claude/n8n workflows via a simple API. Fits directly into existing Python scripts

Results

Three production workflows running with no analyst time required: automated signal qualification at sourcing, continuous CRM enrichment, and a weekly fintech intelligence briefing.

For investment firms building AI-native sourcing operations, clean data in and structured intelligence out are the two variables that determine whether AI actually accelerates deals or just adds overhead. Linkup solves both. Get started for free or talk to our team: contact@linkup.so