Keytext

Narjesu Buffadin founded Keytext in 2010 as a professional services firm rooted in natural language processing and AI—coming from an academic background in the field. She spent the first four to five years building the business on services, generating revenue from consulting work while laying the technical groundwork for what would become a product.

By 2015, Narjesu made the pivotal decision to pivot from services to product. She recognized the fundamental constraint of the services model: "You are using people, their knowledge to sell something... Unless you start to stack people one on the other, you can't necessarily grow very fast." The product launched at the beginning of 2016, automating the core value she'd been delivering manually—analyzing customer feedback to identify reasons for disengagement. The platform ingests unstructured text data from any source (Zendesk, Intercom, surveys, live chat, chatbots) and uses AI to surface actionable insights about why customers and employees are becoming disengaged.

Keytext grew primarily through partnerships with survey platform providers and marketing agencies. Early partnerships included companies like iPerception and LG2 (a marketing agency). These partners integrated Keytext into their offerings, giving them a 50% revenue share model. By the time of this interview (roughly 2018-2019), Keytext had 30 customers, most of them enterprise businesses. The partnership model meant Keytext didn't directly acquire these customers—the partners handled the sales cycle.

The partnership channel proved extremely effective. Over the past 12 months, Keytext grew approximately 5X year-over-year. Monthly revenue climbed from roughly $16,000 a year ago to "almost" $100,000 per month ("not far away from $100,000"), putting them on a trajectory to break $1 million in ARR. With most customers paying $2,500 per month on annual contracts, Keytext achieved zero churn—all customers renewed. Narjesu attributed this to genuine product value and active platform usage, though the host suggested the low churn might also indicate underpricing.

One limitation: calculating customer acquisition cost (CAC) proved difficult. With partners bringing in 50% of revenue through shared deals and handling most of the sales cycle, it was hard to isolate the true CAC. The company had 14 engineers and researchers out of 20 total headcount, reflecting the heavy R&D demands of an AI product.

Keytext had raised approximately $4 million in equity (Canadian dollars) and was still burning cash to fund growth, though Narjesu expected 24-36 months before reaching profitability—typical for AI-driven SaaS companies with higher R&D costs. The team was entirely based in Montreal, benefiting from the city's concentration of AI talent and the University of Montreal's AI research community. The company was at an inflection point, having validated product-market fit through partnerships and beginning to explore direct sales to brands and large businesses. Approaching $1 million in ARR with zero churn and 5X annual growth, Keytext had built a durable business in an emerging category.