‘KOGO’s AI agentic framework performing real-world tasks a first in India’

MapmyIndia-backed AI start-up KOGO claims to have developed India’s first AI agentic framework that can perform real-world tasks. The company’s proprietary product KOGO operating system (OS), is built on small action models and can perform various tasks regardless of the industry, says Co-founder and CEO Raj K Gopalakrishnan.

“KOGO was started six years ago as a deep tech AI company. The objective was to use AI to create path-breaking products. Our first offering disrupted the travel and mobility space. We solved problems using the Gen AI precursor - Narrative Intelligence and Computational Linguistics,” said the CEO, adding that KOGO was launched in February 2020 in India, before Australia, and Indonesia.

“As soon as the world opened after Covid, we launched our enterprise stack and got the top three largest manufacturers of two-wheelers, Hero, TVS, and Bajaj, to adopt it. Okay. We became the default AI stack for mobility and travel.”

The company realised no platforms or frameworks were available to create an AI travel expert/super agent. “There still is no framework available to create AI agents like this because talent is scarce and you need the right technology. Logically, we had to build an AI agent OS platform, and so we did. We created our AI agent builder and then built the travel AI expert. In May this year, we said we are India’s first AI agent OS or agent builder; a no-code, low-code platform you can build agents within minutes,” he added.

“When there is a monetisable business idea, technology is an enabler. Our platform is an enabler across the board. Our agent can do sales, research, operations, customer service, and understand data. The industry doesn’t matter because it can be applied across all of them,” said Gopalakrishnan, adding, “Everybody needs customer service, business intelligence to crunch data, operational support, research, and sales. We take enterprise workflows and get manual and repetitive work done. Our stated goal is very simple. We have a human-first manifesto and our tagline is human potential meets AI. KOGO AI agents can bear the grunt work.”

‘AI is expensive and difficult’

Mentioning that businesses are unaware of where to start because building AI is expensive and difficult, Gopalakrishnan pointed out the gaps in privacy and lack of real-world deployment. “Despite various frameworks like Langchain, Llama Index, and Autogen, there is no real-world deployment or real-world agents. Only large enterprises with a lot of money can deploy and get expertise in AI since there is no reliable production-grade platform for building AI agents,” he said, adding that most enterprise-grade frameworks, of which there are very few-- are thin wrappers on top of ChatGPT. 

According to KOGO’s business model, consumers get a million agent tokens for $66. Some agents may use 2,500 tokens for processes, while others will require half a million because of the complexity of others. “An agent performing multiple functions will use a lot of agentic computation because of multiple skills and capabilities and may cost 1 million tokens.”

“KOGO did a few rounds of funding with angel investors before MapMyIndia, in 2022, acquired a 26 per cent stake. In the following phase, they further invested and got another 14 per cent stake, with a final stake of 40 per cent. We started generating revenue last year, closing at about $200,000. This year, in the first month, we closed an order book of $400,000. Our ARR will be about $2.6-2.7 million by the end of this year,” said the CEO. With 27 ongoing proofs of concept (POCs), most will go live in the second and third quarters.

Out of the 27 POCs, seven are happening between the Gulf and North America. KOGO’s recent representative office in the MENA region will operate its European markets. The company has started conversations in North America.

Gopalakrishnan mentioned that globally, the AI agent industry in 2024 is worth around $8.7 billion. It is growing at 43.8 per cent, crossing $100 billion by 2029; 30 per cent of this development is in the US, followed by 26 per cent in Europe and about 18 per cent in India. The largest sectors using this innovation are medical and BFSI, followed by retail and e-commerce, and travel and mobility. “We are focusing our energies on these sectors with the agent templates we’re making,” he said.