A former Ola employee says under 10,000 people use Ola Krutrim's LLM chatbot, which supports 10 Indian languages, and that over 60% of them are random testers
Swathi Moorthy / The Economic Times : X: @ettech , @ettech , @ettech , and @ettech X: @ettech : 🚨🚨 In a setback to @bhash and the broader Indian #AI ambitions, several founders and investors told us that @Krutrim large language models (#LLMs) and cloud offerings have received a lukewarm response from the #market. @ettech : 📌 Krutrim, the AI venture backed by the #Ola group, became India's first #AI unicorn in 2024, after raising $50 million at a $1 billion valuation. But the #company has since faced product roadblocks and #leadership churn. @ettech : ✍️The AI model also suffers from high latency, which refers to #response time, deterring potential users. In tests reviewed by ET, Krutrim's AI chatbot took 41 seconds to generate a response to a single prompt. In contrast, #ChatGPT-4o and #DeepSeek responded in under 10 seconds. @ettech : ➤ Founders cited poor documentation as a key issue with @Krutrim's products. ➤ They also flagged a lack of #technical maturity. ➤ More than 20 #employees have exited the company since 2024. [image]
Context & Ripple Effects
Krutrim’s story began with the launch of a multilingual model positioned for 10 Indian languages, then expanded into developer cloud access and a chatbot app. The reported usage figures test whether that product expansion translated into sustained demand.
The weak uptake arrives after coverage that the company was seeking a smaller funding round amid muted investor interest, making product traction—not just model scale—the central issue for its next phase.
First-order effects
- Krutrim faces an immediate credibility problem with developers, enterprise buyers, and prospective investors: reported usage is low and mostly experimental rather than recurring.
- High response latency, alongside reported documentation and product-maturity gaps, makes the chatbot and cloud offering harder to position as dependable alternatives for practical workloads.
Second-order effects
- A smaller recurring-user base gives Krutrim less real-world feedback and fewer opportunities to refine its multilingual product, potentially widening the execution gap with better-used AI services.
- Fundraising discussions are likely to place greater weight on retention, product performance, and customer adoption; the company’s planned large-model effort with Lenovo may face higher pressure to demonstrate a clear use case.
Third-order effects
- The episode underscores that a locally positioned LLM needs distribution, developer usability, and reliable inference—not language coverage alone—to convert national-AI ambitions into a durable platform.
- If similar adoption gaps persist across domestic AI labs, capital may concentrate around companies with demonstrable customer usage and the infrastructure to deliver lower-latency services.
The trend: India’s AI market is shifting from symbolic model launches toward scrutiny of whether locally built systems can earn repeat use and developer adoption at production-quality performance.