GenAI Knowledge Mining Tool for VC & PE Firms

About

NovaAI is a GenAI knowledge platform for venture capital and private equity teams. The project covered product design, UX strategy, and AI interactions within an MVP, connecting research with document creation.

Challenge

Investment knowledge sits across reports, decks, conversations, and external sources. The product needed to help analysts explore that information, compare perspectives, and develop their findings without losing the context behind them.

Outcome

An MVP design that supports two ways investment teams work: following a line of inquiry in depth and comparing information side by side. Multi-threaded conversations and comparison views give each mode room, while iterative document building lets the research develop into a structured output.

The experience gives exploratory AI interactions a place in the wider investment workflow, connecting questions and analysis to work the team can use.

Introduction

Investment teams operate in information-dense environments — constantly working with fragmented data across reports, decks, conversations, and external sources. Much of this knowledge remains unstructured and difficult to reuse, making it hard to connect insights across deals, track evolving narratives, or move efficiently from research to decision-making.

The platform was built to bring structure to this process — enabling teams to extract insights, synthesise information, and generate structured outputs without losing context. It combined conversational AI interactions with document generation workflows, supporting both exploratory thinking and more structured, output-driven use cases.

The product was designed around how analysts and partners actually work. This meant supporting two key modes of behaviour — deep, focused exploration and side-by-side comparison — allowing users to move fluidly between analysing a single narrative and evaluating multiple perspectives.Features like multi-threaded interactions, comparison views, and iterative document building were introduced to support these workflows.

A key part of the exploration involved understanding how generative AI could fit into real investment workflows — not by replacing judgement, but by helping users structure, refine, and articulate their thinking.The system allowed users to move from scattered notes to more coherent narratives, generate working drafts of reports, and collaborate on them within the platform.

We also explored early multi-agent interaction patterns, where different AI agents could support distinct modes of thinking — from analysis and synthesis to simulation and scenario exploration. These were intentionally lightweight, allowing users to shift contexts fluidly while testing how such systems might evolve.

Rather than forcing a single way of working, the system was designed to adapt to how analysts naturally think — supporting both deep, focused exploration and side-by-side comparison when evaluating multiple perspectives.

The result is a workflow that moves seamlessly from fragmented inputs to structured narratives — enabling teams to develop clearer thinking, generate working drafts, and collaborate within a shared context.