Enterprise sales has always been an expensive game to play well. Winning a contract with a company like Nestlé or Puma typically meant fielding a large, trained sales team: researchers who could identify the right buyers, sellers who understood procurement cycles, and managers who could coach a pipeline for months before a single deal closed. For most early-stage startups, especially those outside the US, that cost was simply out of reach. It is one of the quieter reasons why so many promising companies stay regional even when their product is genuinely world-class. Over the past two years, that calculation has started to shift. AI agents, software systems that can monitor buyer behaviour, flag intent signals, draft outreach, and adjust targeting without constant human input, are increasingly being used not to replace sales teams but to make small ones behave like much larger ones. Instead of hiring ten more sales development reps to cover more accounts, a founder can now deploy agents that work around the clock, learn from every interaction, and hand off warmer leads to a much smaller human team.
A year ago, Archit Anand set out to build a sales team for his new startup, and decided not to hire one in the traditional sense. As founding GTM lead at Refold.ai, he chose instead to build the function around AI agents from day one.
That decision sits inside the same bigger shift. The pattern holds across most industries selling into large enterprises: procurement cycles are long, the buying committee is hard to reach, and the traditional fix has been to throw more people at the problem, more researchers, more outreach, more managers to keep the pipeline moving. What is changing is not the difficulty of enterprise sales itself, but who can afford to compete in it at all.
AI agents are starting to change that calculation. A typical setup stitches together a few specialised functions: one layer tracks signals such as a company hiring for a relevant role or raising funding, another drafts and sequences outreach, and a third learns from outcomes to hand off only the warmest conversations to a human. The effect is threefold: faster sales cycles, sharper targeting, and, in some teams, agents that coach human sellers in real time on what is and is not working. The technology has limits, agents can misread signals or reach the wrong stakeholder, and large, risk-averse buyers still expect a credible human closing a six or seven-figure deal. What has changed is how much ground a small team can cover before that human judgement is needed at all.
Archit Anand's own numbers give that shift some texture. Refold.ai raised a $6.5 million seed round from Eniac Ventures and Tidal Ventures, and the agent system he built tracks buyer signals, orchestrates outreach, and coaches the company's human sellers daily on who to prioritise. A custom AI SDR system he built reportedly lifted sales-development effectiveness by 3.5x, helped refold grow from $0.9M ARR earlier this year to $2M ARR by July and a related advisory engagement helped a client double annual recurring revenue from $750,000 to $1.5 million in a single quarter, cutting the sales cycle nearly in half.
What makes his account worth listening to is the comparison he can draw from his own career. As co-founder of Omnirio, a commerce platform he built and scaled between 2019 and 2023, Archit Anand had already run the old version of this playbook: growing the company from zero to over 500 enterprise clients across more than 30 countries, winning brands including Nestlé, Lacoste, Marks & Spencer, and Puma, using a conventional, people-heavy sales motion. Few founders get to compare both eras of enterprise sales from inside two companies of their own, and that before-and-after is what makes his read on the shift worth more than speculation.
The broader question his work points to matters beyond Refold: if a small team of agents can genuinely do the work of a much larger sales org, it changes who gets to compete for enterprise contracts at all. Startups that could never have afforded a 30-person revenue team may not need one.
