Enabling Real-Time Discovery Across a 600-Startup Portfolio Through AI

5 min

IN A NUTSHELL

5 +

hours time-to-discovery reduced to under 10 seconds

85 %

match accuracy, based on user feedback from enterprise teams

7 x

reduction in time for meet preparation

600 +

portfolio companies’ data structured and centralized

About Client

A global private equity and venture capital firm managing over $90 billion in assets, the client invests in innovative software businesses and helps connect them with enterprise buyers. With a rapidly growing portfolio of 600+ startups, the client sought to streamline how internal teams discover, surface, and recommend relevant companies to large enterprises.

EMPLOYEES

500+

LOCATION

Global

INDUSTRY

Private Equity

01

What Was the Challenge

With a rapidly expanding portfolio, the client’s internal teams found it increasingly difficult to identify the right startups for specific enterprise needs. Disconnected tools and scattered data made meeting prep slow and inefficient, often taking hours. They needed a centralized, intelligent platform to surface relevant companies in real time, respond to emerging business challenges on the fly, and unlock the full value of their 600-company portfolio while reducing operational overhead and improving sales conversations.

02

How We Solved the Problem

Ciklum collaborated with the client to build a scalable, cloud-based platform that redefined how teams access and activate startup insights. At its core was an AI-powered search feature that enabled real-time discovery during enterprise meetings, surfacing relevant portfolio companies in under 10 seconds. By combining data engineering, platform design, and AI, Ciklum transformed scattered startup data into a structured, searchable catalog with intelligent filters and contextual relevance.

Structuring the Portfolio

Ciklum’s data engineering team began by ingesting and organizing startup information from across the client’s 600-company portfolio. We structured key data points like industry, solutions, buyer personas, and business challenges using a Snowflake warehouse and orchestration tools such as Airflow, Astronomer, and DBT.

Building an Intelligent Catalog

The platform enabled enterprise teams to filter startups based on industry-specific business challenges such as cybersecurity, AI enablement, or cloud infrastructure. These categories were powered by a smart recommendation engine that used LLMs to map startups to challenge categories based on relevance.

Designing the AI-Powered Search

To overcome the limitations of static categories, Ciklum implemented a natural language search feature, enabling users to type in unique business queries and instantly receive tailored startup matches. This was a game-changer for real-time meetings, reducing prep time from 5 hours to just a few seconds.

Improving Security and Scalability

The solution was built on a secure, multi-tenant AWS architecture and integrated with OKTA for authentication and role-based access. Frontend and backend were developed using React and Node.js, supported by CI/CD pipelines via GitLab.