Why We Invested in Echovane?

What is the Opportunity?

AI in research has enabled new research methodologies to become possible and cover multiple things:

• First, starting with large language models that can sustain coherent and contextual conversation. We anyway see that happening with voice AI in general. That technology is democratising building agents yourself which are sounding as good as a human.
• Second, the multi-modal understanding where not just conversation but video, audio, and text simultaneously can be processed. Participants' facial expressions while listening to their words can notice that those two are telling different stories and then probe accordingly.
• Third, this all can happen on a system that can do multiple interviews simultaneously without the quality of any interview or conversation declining.
• Fourth, there is durable economic value now in understanding the why of a research rather than what, which is mostly commoditized. Probing conversations, understanding motivation is the contextual nuance that organisations need.

The WTP in research is tied strictly to the cost of a mistake avoided, hence these enterprises have a large insights team that is meant to give stakeholders evidence and de-risk, validate before a launch, hence the per-study price compression might not collapse/shrink the market.

The distinction that decides who wins

Our early instinct was that AI-moderated interviews were becoming a crowded field - we mapped many players across interviews, synthetic users, survey tools and UX research. What we hadn't understood, and what the founders articulated with unusual clarity, was that almost all of that density sits in the tooling layer.

But when we dived deeper with the team, we realised CPG insights teams are not researcher ICP. They are research buyers - lean teams with very large budgets whose job is to procure insight and defend it to senior leadership, not to operate a research platform. Unlike digital businesses, they have no clickstream, no direct line to their consumer, and so they have always outsourced the work and bought the answer. This is precisely why fifteen years of excellent DIY research software under-penetrated the segment holding the biggest budgets. A tool transfers work to this buyer. An outcome removes work from them.

Echovane deliberately built for the outcome layer. A client brings a business question, For eg. Packaging changed for product X and sales are softening in one region, why? - and Echovane returns a finished, decision-grade answer in about a week rather than six to eight weeks, at a fraction of legacy pricing. Between those two points sits a system of agents: research design, participant recruitment, AI-moderated interviews running 60–120 minutes across 65+ languages, vision models capturing what people do rather than only what they say, and synthesis into a deliverable an insights director can walk into a leadership meeting with. A thin layer of human judgment sits at the two ends where trust is actually transacted — framing the question, and vetting the answer.

What convinced us

The team has kept its footprint deliberately small, choosing to solve depth before adding headcount, and has built the platform, the panel infrastructure and the go-to-market motion profitably while doing so. That showed up in the product as much as the P&L - in how carefully they've engineered around cost per interview, model efficiency and orchestration, so that quality improves without the cost curve moving with it. They've completed 50,000+ interviews, 5,000 observational studies and 3 million questions, working with the likes of Procter & Gamble, Coca-Cola, Haleon and Kantar. Every pilot they ran converted. When we saw rather larger incumbent counterparts choosing to bring them into its own client relationships, it moved from being hypothesis to an empirical observation. Every client they had worked with was super impressed by the quality of insights and with massive scale improvement across different regions/languages that they could conduct these interviews and return back for more.

The durable thing here isn't speed, though. Today the deepest knowledge about how a brand decides - its ethos, what it accepts or rejects, why the last recommendation was overruled - lives in an account director's head and walks out when they change jobs. Echovane is encoding that into a system, so the fifth study for a brand is meaningfully sharper than the first. Research shifts from a recurring expense to compounding infrastructure.

Where we think this goes

There’s a classic case of Jevon’s paradox flowing in here. Cheaper research should expand this market, not shrink it. Over 30,000 consumer products launch every year and most fail - rarely from poor execution, but usually from untested assumptions. When the cost and latency of asking collapse, the questions teams always had but never commissioned finally get asked, and research moves from an occasional procurement event into the operating rhythm of a brand. We can visualize this happening with even tier 2, tier 3 enterprises ($1-$20B) in the US wanting to research more while having not budgets as big as a Mega cap ($20B+) corporation.

Most AI-native service delivery startups we met assert “data flywheel” that lacks how are they extracting org-specific information and reading between the lines; here we can trace the specific sensor, the specific data, and the specific payoff that will lead them to get entrenched in an organisation having better clarity and headway on their research dynamic.

Smriti, Vipul and Himadri have known each other since a long time and through IIT (BHU). They've pivoted twice and chosen to build genuine infrastructure where it was hardest and we're glad to be partnering with them along with Neon fund for their first round. With this capital, we plan to go deeper across enterprise spends in research & delighted partners along their journey.