Building a Voice AI Deal from Brief to Close
Took a broad enterprise AI mandate, narrowed the sector, sourced stealth companies, and supported a closed investment
Vedansh Agarwal
Investment Banking Analyst at SG Analytics
From their time as
Investment Banking Analyst
SG Analytics • 2025
Overview
Vedansh received a broad brief from a US-based VC client: invest a defined sum into an enterprise AI company at pre-seed to Series A stage. The client had no sector preference beyond that. The first task was to make the brief investable.
The Story
Vedansh received a broad brief from a US-based VC client: invest a defined sum into an enterprise AI company at pre-seed to Series A stage. The client had no sector preference beyond that. The first task was to make the brief investable.
He mapped three candidate sectors: CRM-based AI, general enterprise chatbots, and voice AI. For each, he built a quick view of growth trajectory, competitive dynamics, and cash flow predictability. Voice AI stood out on stickiness: once a company integrates a voice interface, switching costs are high and revenue becomes recurring. He recommended voice AI to the client, and the client agreed.
Sourcing Beyond the Databases
Standard databases like PitchBook and Capital IQ were insufficient for pre-seed companies in an early-stage sector. Vedansh identified two to three LinkedIn creators, deep tech investors based in the US, who published funding announcements and new company launches before they appeared on PitchBook. He added Product Hunt as a secondary channel for emerging voice AI products.
This sourcing approach surfaced companies one to two months ahead of the databases, giving the client a timing advantage in a sector where early entry mattered for valuation.
Narrowing to Two Companies
From the sourcing pipeline, he screened companies against the client's criteria: healthcare voice AI, pre-seed to seed stage, strong founder background. He narrowed the field to two companies, one sourced from LinkedIn and one from Product Hunt. Both were at pre-product stage.
For each, he built a profile covering founder credentials, product focus, target customer, and strategic fit with the client's thesis. The recommended company was more narrowly focused on healthcare AI and had a more developed product layer compared to the alternative.
Valuation and Term Sheet
With two candidates shortlisted, Vedansh built a valuation model using comparable deal multiples from PitchBook and Capital IQ. A DCF was not appropriate at pre-seed stage given the absence of reliable revenue projections. The model used stage-matched transaction multiples to arrive at a valuation range.
The client used this analysis to structure a term sheet. After a compliance and founder-market-fit review of approximately one and a half months, the deal closed. The company accepted at a slightly higher valuation than the initial offer, which the client accepted given conviction in the sector.