How Procol Took Procurement From WhatsApp to AI Agents

A designer from Jaipur saw grocery buyers awarding crores over WhatsApp, and realised that this could be solved via software. Eight years and a couple of iterations later, Procol runs procurement for 250+ enterprises in more than 10 countries. This is their story.

In early 2018, Gaurav Baheti was 24, and doing design gigs when Future Group came calling. They hired him to build an app for their procurement team. And as he researched the problem, he observed a deeply entrenched habit.

The buyers, people who were responsible for stocking the chain’s shelves, would ring vendors for spot rates, haggle a little, and award orders worth lakhs, or crores, on a call or a WhatsApp message. There was no bid sheet and no record of who else had quoted. One person on the ground made the decision, and the money started moving.

“I was like, how can a million dollars’ worth of trade happen just like that? There was no structure around it,” Gaurav says.

That question became Procol (Blume Fund III). Nearly eight years on, the company runs procurement for more than 250 enterprises across 10-plus countries, including the US, parts of Europe, the UAE, Saudi Arabia and Southeast Asia. Over $30 billion of spend flows through the platform, and Procol says it has saved customers more than $1 billion along the way. A million-plus suppliers are onboarded, and about 200,000 bids pass through every month. With revenue growing 5x in two years. 

It wasn’t smooth sailing; getting here took a lot of pain and effort. And the Procol of now isn’t what it started as. The startup spent its first five years as a commodity auction platform for food companies, and grew to ₹500 crore of monthly volume. It was rebuilt as an enterprise platform after all that.

Beginnings: A Schoolboy With Clients in Estonia

Gaurav grew up in Jaipur in the 1990s. His grandfather had started a small brokerage in agri commodities. His father and uncle carried it on, always trying something new on the side. ​“It was nothing glamorous,” Gaurav says. 

The family moved to Delhi when he was ten or eleven. It was a whole new atmosphere compared to the Jaipur of the 90s. New city, no friends, so the internet filled in. 

By 13, he was designing websites for people he met on online forums and Reddit. Clients talked to him on Skype and paid online, and nobody needed to know he was still in school. A quick job took three or four hours and paid $20 to $30. Most work came from the US and Canada, though some came from Estonia and Austria. 

“I didn’t even know that there was a country called Estonia before the project came my way,” he says. 

He taught himself to code off YouTube and began passing projects to other freelancers, taking a cut. His father noticed a schoolboy with an income and got worried that his son was up to something shady.

In between all the freelance projects, there was tennis too. A Federer loyalist, Gaurav played at the national level until entrance exams took over in the 11^th grade.

The years after college: Zomato, a decision, and a Ticket to New York

In 2014, while at IP University in Delhi, Gaurav applied online for an internship at Zomato. Deepinder Goyal immediately saw the potential, and made him an offer. He joined as a tech intern, wrote code for a week, before his skills were noticed, and he was asked to move to design. 

He ended up working on the restaurant page and the discovery page of Zomato’s mobile apps. This was before food delivery, when Zomato was a listings and table-booking company pushing into 20-odd countries.

Deepinder would sit with the design and tech teams and give feedback standing at their desks. What stayed with Gaurav was the hiring bar. ​“He would hire the JEE rank one. He would hire people from Facebook and Twitter,” he says. ​“No other Indian company was doing that in 2014.”

Zomato offered Gaurav a full-time role on one condition: that he drop out of college. But that wasn’t something he wanted to do. He said no, and went to Zo Rooms as its first designer, building the app, the website and the rest of its digital experience. Then Google came through. At 21, just past his third year of college, he flew to New York to join the Google Forms team as one of five designers. It was his first time outside India.

“I was not going there to study like most people,” he says. ​“I was going to work and get paid for it.”

He spent his fourth year of college holding down a full-time job at TableHero, a table-reservations startup in the US. After graduating, he moved to the US to work with them, then left a few months later to go independent again.

The next two years were freelance, and he was picky about it. Nikhil Kumar of iSPIRT, who later co-founded Setu, asked him to design the India Stack website. More work followed, including the BHIM app that NPCI launched in 2016.

“iSPIRT opened my eyes to how impactful the work can be when you work with the policymakers of the country,” Gaurav says.

Procol: Bringing the Mandi online

The Future Group project arrived at the right moment. As he worked on the procurement problem, he drew upon his own family background. Gaurav had grown up around commodity trading, with two generations of his family in the brokerage business. He understood the issues, the customer, and the ecosystem.

And the insight that the space was just waiting for software spurred him on.

In October 2018, he and Sumit Mendiratta, a colleague from Zomato, founded Procol as a commodity bidding platform. 

Then they went straight to old Delhi for research. Gaurav spent time in Khari Baoli, Asia’s largest wholesale spice market. The market sold spices, grain, and dry fruit. From the street, to an untrained eye, it simply looks like a large crowd. However, inside, buyers and sellers find each other, swap samples, and keep negotiating until a price holds. ​“Millions of dollars of transactions are happening there at any given moment,” he says.

Gaurav rebuilt that loop on a screen. A buyer posted a requirement, and suppliers studied the spec and asked for samples. The ones who qualified competed in a reverse auction that drove the price down.

The mechanics for this came from an older Indian ritual. In agricultural markets, bids were traditionally made under a handkerchief, between two hands, so nobody else saw the number. Every price stayed private, and the best one still won. Procol’s auctions keep that logic. Suppliers can’t see each other’s bids, and the buyer still closes at the best price.

The on-field research kept rewriting and expanding the product. Price turned out to be just one lever among several. Buyers and suppliers argued over quality, delivery terms and contract clauses, so Procol built a chat-based negotiation feature that let them haggle on any parameter. After a deal closed, people wanted to know where the goods were, which led to delivery tracking. A quality-check module came next, so supplier ratings had evidence behind them.

“We had to live the life of the user,” Gaurav says. ​“For every product we built, we did the same discovery over and over again.”

Blume led Procol’s $1 million seed round in mid-2019, with Rainmatter joining in. Sequoia’s Surge programme (now Peak XV) came in with its third cohort in January 2020, weeks before COVID shut the country. Procol was then handling about ₹50 crore of procurement a month. By 2022 – 23 that number was ₹500 crore.

Karthik Reddy, co-founder and Managing Partner at Blume, had met Gaurav before Procol. ​“His first pitch to me was a fantasy cricket, prediction-market kind of idea, about three years before we funded Procol,” he says. ​“That whole area was always too grey for me, and I said this is not for me. But I loved the energy, so I told him to keep us posted.” Gaurav came back with another idea, and then with Procol, by which point he had a scaled implementation running at Future Group.

“It was just one large client at that point, but we saw the spark of something interesting,” Karthik says. ​“Gaurav was a very good product builder, and Sumit was a very good technology leader. The combination felt like it would meet our requirement, which is to obsess about the problem until you find the best solution.”

Risks and iterations

Around that time, B2B commerce was picking up steam in the country. Companies like Zetwerk, Moglix, and OfBusiness were raising big rounds. Procol had real volume, and the obvious next step was to go the trade financing route. However, when they ran the numbers, they didn’t like where it was logically taking them. 

Trade financing was the centrepiece of this B2B commerce boom, but Procol was only the rails the trades ran on. To earn from those transactions, it would have to lend against them, which meant becoming an NBFC. ​“That is not what our aspiration was,” Gaurav says.

So they looked at global leaders like SAP Ariba and Coupa. Between them, they sit under trillions of dollars of B2B trade worldwide and never touch the working capital. Nothing like them had taken hold in India or the rest of Asia, and the buyers were already there: roughly 10,000 Indian companies with more than ₹2,000 crore in revenue, most still running procurement on Excel and email.

The call took a lot of soul-searching. ​“What do we want to spend our next decade working on?” Gaurav remembers asking himself. He went with his gut, and with the thing he’d been doing since he was 13, which was building good products and selling them.

“The crux of any business strategy is identifying what your strengths are and doubling down on them,” he says. ​“I’ve started to admire founders who go with their gut.”

However, the price was steep. The commodity business was shut down, and Procol cut its team by half. The software wasn’t enterprise-ready yet, and the sales team had never worked a deal cycle longer than two months. Almost everything had to be rebuilt from scratch.

“That took a lot of perseverance,” Gaurav says, ​“from the core team and from every single person who had been with us for three or four years.”

Gaurav describes the journey as a series of iterations, each one prompted by what the market told them. ​“We thought we could go down the rabbit hole of agri-commodities procurement and become the most specialised player in it, until the clients told us they wanted a more robust horizontal system,” he says. The second change came from the financing question. ​“Financial services need a very different organisational culture and machinery. So we decided to stick to our competencies and build the best software.”

Selling to the World

Procol had zero footprint outside India, and Gaurav admits he didn’t know how to sell to enterprises abroad. He started with North America, hired an inside-sales rep in India to sell into the US, and began cold outreach on LinkedIn.

The problems abroad were familiar, just further along. ​“Some of these global companies have already solved the basic problems, and are looking to solve a much more mature version,” Gaurav says. That was useful at home too, because Indian customers would eventually ask for the same things.

Similarly, one customer in Dubai opened up a new product line: Enterprises everywhere still onboard suppliers by hand. They email a registration form, collect documents, run checks, and create a vendor code in the finance system before anyone gets paid. Procol built a vendor onboarding, risk and performance system for that Dubai customer, and that became the Vendor 360 module.

Agents for the Back Office

AI changed two things for Procol, Gaurav says. Software got cheap to write, and it became possible to deliver it much faster, which matters most in enterprise software, where deals tend to die slowly in implementation.

Procol immediately made speed its edge. A typical rollout ran three to six months, and about half of that was workshops and blueprinting, where the customer would pin down scope and the processes it wanted to standardise so spend has proper governance. AI helped Procol speed up this entire process, ensuring a faster rollout. 

The second shift was agents for the end user. Procol already held years of data on what each customer bought, from whom, and when. It used that context to build autonomous procurement agents that negotiated with suppliers, discovered new ones, and raised purchase orders on their own.

“These buyers hadn’t even moved from Excel and email to a platform,” Gaurav says. ​“Now they are adopting AI agents to do the work itself.”

The ontology problem

Gaurav’s bigger bet rests on a problem he calls ontology. Every enterprise runs on its own systems, with its own labels and its own context, none of which match anyone else’s. An agent brought in from outside can’t read that data, so it can’t act on it.

“That’s why enterprises mostly have not adopted agentic software for specific functions,” he says. He counts two functions that have cracked it: legal, with Harvey and Legora, and software engineering, with Cursor and Codex. Procurement and finance, the core of the back office, are still open. ​“Everybody has built agents, but enterprises can’t adopt them at scale due to the very unique ontology everyone has.”

The obvious pushback is that enterprises could build their own. Many do try, but Gaurav reckons that 70% of the software large companies use today is built in-house, by big IT teams or by firms like Infosys and TCS. Procurement tends to break that habit. Customers show up with a homegrown system they’re tired of maintaining, or a prototype they want Procol to take into production.

“If implementation takes three to six months for someone who does it repeatedly, building it and failing will take you 24 to 36 months,” he says. What customers buy from Procol, in his telling, is the experience of having solved the same problem for more than 200 companies.

The team that stayed

Gaurav calls the core team his biggest achievement of all these years, and also the thing that took him the longest. As a first-time founder, he says he didn’t know what it took to build camaraderie around the mission. Plus each iteration made it harder. People hired for one version of Procol didn’t always fit the next. 

The past year, he says, has been the best yet. Procol has now spent around 18 months on enterprise software and AI, and the people who stayed through the pivots have grown with it. The first engineer who built the core systems is still there; so are the first salespeople who sold the software. The hiring rule is simple: experienced people for known problems, young people for new ones. And Gaurav expects revenue to at least double every year from here, possibly triple.

“Gaurav is very profit-oriented and focused on building sustainably, and we have always loved that about him,” he adds. 

The phone call at Future Group, and the handkerchief-haggling in the APMC yard were versions of the same thing: large sums changing hands on trust, with nothing written down. Procol has now spent eight years building the record. The next step, as Gaurav sees it, is letting agents do the bidding inside it.