During 2018, Anshuman Pandey, Naman Maheshwari, and Rohan Pooniwala were involved in work to simplify Artificial Intelligence (AI) development for Machine Learning (ML) engineers.The trio met while studying for a master's degree at SRM University in Chennai, and later moved from a hacker house in Kovalam city (a workstation for developers) to Nimblebox. Ai was built.<\/p>\n\n\n\n
Bootstrap used to offer a comprehensive toolkit designed to help startup developers quickly set up their ML projects.<\/p>\n\n\n\n
\"The first thing we wanted to do was help users get started on their ML projects quickly.Since there was a huge barrier to entry, we wanted it to be easy.We wanted to fix that, and that's why we created Nimblebox. Ai started making it in its first iteration, says Naman<\/p>\n\n\n\n
The platform has grown rapidly, serving over 16,000 developers from 200 companies in more than 30 countries. However, in 2023, they noticed a growing demand for generative AI solutions. So, they changed the focus and created Tune AI to help enterprises and developers leverage generative AI for their applications and workflows.<\/p>\n\n\n\n
\"We built machine learning models for enterprises and updated our platform to work with advanced language models.\"We're one of four companies that made Meta's Llama 2 available to the public within 24 hours of it being available to internet customers,\" he explains.<\/p>\n\n\n\n
Before this, Rohan was setting up the infrastructure for detailed studies at Harvard Medical School, Anshuman did research on Ethical AI at Carnegie Mellon University and Naman was doing research on human-computer interaction at IIIT, Delhi.<\/p>\n\n\n\n
Product offers<\/p>\n\n\n\n
The San Francisco and Chennai-based startup has two main offerings - Tune Chat and Tune Studio.<\/p>\n\n\n\n
TuneChat is an AI chat app that provides powerful models for text creation, code generation, and brainstorming.<\/p>\n\n\n\n
\"We're developing this platform for builders who incorporate General AI into their apps and products,\" says Naman.<\/p>\n\n\n\n
\"All companies will use AI in some form or the other. TuneChat is a platform for knowledge workers in the enterprise. This is for companies or enterprises where your data security and privacy are of paramount importance. So basically, we help set it up in the enterprise or behind their firewall, where their employees can use our platform to create content or help with writing, \"he adds.<\/p>\n\n\n\n
Tune Studio is a comprehensive solution for fine-tuning, deploying and managing the AI model lifecycle and enabling data security with enterprise-grade compliance.Through the \"Ultimate Creative Sandbox\" playground, Tune Studio allows users to experiment with open-source models and superior performance.<\/p>\n\n\n\n
\"Anyone who wants to incorporate Gen AI into projects or applications can use the platform using closed and open source models,\" explains Naman.<\/p>\n\n\n\n
The startup also uses it to create automated workflows for large pharmaceutical and information services enterprises with global operations.<\/p>\n\n\n\n
The Applied AI team at Tune AI works on use-case-driven projects to increase the efficiency of certain processes within the enterprise or to bring advanced technologies, such as agents, into their workflow.<\/p>\n\n\n\n
The startup claims to have reduced the manual processing time of some workflows by up to 30 times, reducing a portion of the cost for enterprise customers.<\/p>\n\n\n\n
For developers and developers<\/p>\n\n\n\n
The Tune AI platform is built on open and closed-source models such as ChatGPT, GPT- 3.5, GPT-4, and LAMA, which are more flexible and can be used at will.<\/p>\n\n\n\n
\"As of now, the number of 180,000 users has increased to 360,000. We are nearing the milestone of 3,000 new users every day, \"says Naman.<\/p>\n\n\n\n
\"TuneChat helps strategic units in the enterprise to get better insights. It can also go from document intelligence to general operations. Many of these companies have their own data and documents, which they are not able to provide from their firewall, \"he adds.<\/p>\n\n\n\n
Workers in the enterprise can use the platform to gain better insights and get accurate sources for documents.<\/p>\n\n\n\n
Builders on Tune Studio AI are used to whip up quick prototypes, which means building apps.Builders and developers today want to take advantage of this new technology to test and prototype new applications and move on to the next experiment.We are also optimizing a lot for that.Use cases can range from simple API integration and LLMs to providing responses or building an AI agent. \"He says.<\/p>\n\n\n\n
The Market and the Fund<\/p>\n\n\n\n
Tune AI has raised over $4.4 million in funding this year from Accel, Together Fund, Flipkart, Techstars, and CSVP Fund, among others.<\/p>\n\n\n\n
\"Due to rapid advances in generative AI and the growing need for efficient AI solutions, Tune AI stands out for its speed in integrating the latest open source models into its platform.With their offerings, Tune Chat for developers, Tune Studio for enterprises, Tune AI is excited to support their world-class team to optimize internal workflows, easily deploy AI models, and improve AI models, we are on a mission to make AI adoption seamless and effective for Fortune 500 companies.<\/p>\n\n\n\n
The 25-member team claims to have 360,000 users and multiple enterprise customers in the last six months.<\/p>\n\n\n\n
\"The main difference is our team and products. Our products have been in development for more than five years, and it's a domino effect that all the platform features we've built over the past half-decade are now coming together, \"says Naman.<\/p>\n\n\n\n
Some of the prominent clients include agriculture AI start-up Kissan AI AmalgamRx, a US-based healthtech firm, Juicy Chemistry, and some other clients are from information services and healthcare industries.<\/p>\n\n\n\n
In terms of pricing, in addition to a free plan with limited features, Tune AI offers a Pro version at $20 per user \/ month and a custom enterprise plan with premium support that can be customized based on budget and workflows.<\/p>\n\n\n\n
Other companies of this type in the market include San Francisco-based Scale and Imbu.<\/p>\n\n\n\n
Safety needs to be a priority as any AI system is prone to hallucinations, says Naman.<\/p>\n\n\n\n
\"Hallucinations are essentially the modern version of viruses.Using our products, developers or teams can use a number of checks and balances to reduce the risk of these systems being hallucinated.Today LLMs are trained on data that is publicly available on the Internet.While LLMs show hallucinations in giving answers from time to time, similar answers can be found by searching the web, \"he adds.<\/p>\n\n\n\n
One of the main challenges the team faced while building the product was in the area of generative artificial intelligence (general AI), where open-source models are not yet as effective as closed-source models.<\/p>\n\n\n\n
Open source models refer to software that makes the original source code available for free. Closed-source software, by contrast, keeps its source code proprietary, limiting access to developers or vendors.<\/p>\n\n\n\n
\"Gen AI is a new field.Open source models are not as good as some closed source models.Open source models need time to understand where closed source models are today, \"explains Naman.<\/p>\n\n\n\n
According to a report by EY, India has the potential to add $359 billion to $438 billion to its GDP in 2029-30 through the adoption of Gen AI, an increase of 5.9 per cent to 7.2 per cent.<\/p>\n\n\n\n
Going forward, the startup is focusing a lot on optimizing the platform for developers.<\/p>\n\n\n\n
\"We are looking to help 1 million developers use generative AI for their own projects using our platform. We are also looking at nurturing the Indian ecosystem of developers and builders by seeing first-hand how the community in the Bay Area functions, \"he adds.<\/p>","protected":false},"excerpt":{"rendered":"
2018 \u0d15\u0d3e\u0d32\u0d18\u0d1f\u0d4d\u0d1f\u0d24\u0d4d\u0d24\u0d3f\u0d7d \u0d05\u0d7b\u0d37\u0d41\u0d2e\u0d3e\u0d7b \u0d2a\u0d3e\u0d23\u0d4d\u0d21\u0d46, \u0d28\u0d2e\u0d7b \u0d2e\u0d39\u0d47\u0d36\u0d4d\u0d35\u0d30\u0d3f, \u0d30\u0d4b\u0d39\u0d7b \u0d2a\u0d42\u0d28\u0d3f\u0d35\u0d3e\u0d32 \u0d0e\u0d28\u0d4d\u0d28\u0d3f\u0d35\u0d7c \u0d2e\u0d46\u0d37\u0d40\u0d7b \u0d32\u0d47\u0d23\u0d3f\u0d02\u0d17\u0d4d (\u0d0e\u0d02\u0d0e\u0d7d) \u0d0e\u0d1e\u0d4d\u0d1a\u0d3f\u0d28\u0d40\u0d2f\u0d7c\u0d2e\u0d3e\u0d7c\u0d15\u0d4d\u0d15\u0d41\u0d33\u0d4d\u0d33 \u0d06\u0d7c\u0d1f\u0d4d\u0d1f\u0d3f\u0d2b\u0d3f\u0d37\u0d4d\u0d2f\u0d7d \u0d07\u0d7b\u0d4d\u0d31\u0d32\u0d3f\u0d1c\u0d7b\u0d38\u0d4d (\u0d0e\u0d10) \u0d35\u0d3f\u0d15\u0d38\u0d28\u0d02 \u0d32\u0d33\u0d3f\u0d24\u0d2e\u0d3e\u0d15\u0d4d\u0d15\u0d41\u0d28\u0d4d\u0d28\u0d24\u0d3f\u0d28\u0d41\u0d33\u0d4d\u0d33 \u0d2a\u0d4d\u0d30\u0d35\u0d7c\u0d24\u0d4d\u0d24\u0d28\u0d19\u0d4d\u0d19\u0d33\u0d3f\u0d7d \u0d0f\u0d7c\u0d2a\u0d4d\u0d2a\u0d46\u0d1f\u0d4d\u0d1f\u0d3f\u0d30\u0d3f\u0d15\u0d4d\u0d15\u0d41\u0d15\u0d2f\u0d3e\u0d2f\u0d3f\u0d30\u0d41\u0d28\u0d4d\u0d28\u0d41. \u0d1a\u0d46\u0d28\u0d4d\u0d28\u0d48\u0d2f\u0d3f\u0d32\u0d46 SRM \u0d2f\u0d42\u0d23\u0d3f\u0d35\u0d47\u0d34\u0d4d\u0d38\u0d3f\u0d31\u0d4d\u0d31\u0d3f\u0d2f\u0d3f\u0d7d \u0d2c\u0d3f\u0d30\u0d41\u0d26\u0d3e\u0d28\u0d28\u0d4d\u0d24\u0d30 \u0d2c\u0d3f\u0d30\u0d41\u0d26\u0d02 \u0d2a\u0d20\u0d3f\u0d15\u0d4d\u0d15\u0d41\u0d28\u0d4d\u0d28\u0d24\u0d3f\u0d28\u0d3f\u0d1f\u0d46\u0d2f\u0d3e\u0d23\u0d4d \u0d2e\u0d42\u0d35\u0d30\u0d41\u0d02 \u0d15\u0d23\u0d4d\u0d1f\u0d41\u0d2e\u0d41\u0d1f\u0d4d\u0d1f\u0d3f\u0d2f\u0d24\u0d4d, \u0d2a\u0d3f\u0d28\u0d4d\u0d28\u0d40\u0d1f\u0d4d \u0d15\u0d4b\u0d35\u0d33\u0d02 \u0d28\u0d17\u0d30\u0d24\u0d4d\u0d24\u0d3f\u0d32\u0d46 \u0d12\u0d30\u0d41 \u0d39\u0d3e\u0d15\u0d4d\u0d15\u0d7c \u0d39\u0d57\u0d38\u0d3f\u0d7d \u0d28\u0d3f\u0d28\u0d4d\u0d28\u0d4d (\u0d21\u0d46\u0d35\u0d32\u0d2a\u0d4d\u0d2a\u0d7c\u0d2e\u0d3e\u0d7c\u0d15\u0d4d\u0d15\u0d41\u0d33\u0d4d\u0d33 \u0d12\u0d30\u0d41 \u0d35\u0d7c\u0d15\u0d4d\u0d15\u0d4d\u0d38\u0d4d\u0d31\u0d4d\u0d31\u0d47\u0d37\u0d7b) Nimblebox.ai \u0d28\u0d3f\u0d7c\u0d2e\u0d4d\u0d2e\u0d3f\u0d1a\u0d4d\u0d1a\u0d41. \u0d2c\u0d42\u0d1f\u0d4d\u0d1f\u0d4d\u200c\u0d38\u0d4d\u200c\u0d1f\u0d4d\u0d30\u0d3e\u0d2a\u0d4d\u0d2a\u0d4d \u0d38\u0d4d\u0d31\u0d4d\u0d31\u0d3e\u0d7c\u0d1f\u0d4d\u0d1f\u0d2a\u0d4d\u0d2a\u0d4d \u0d21\u0d46\u0d35\u0d32\u0d2a\u0d4d\u0d2a\u0d7c\u0d2e\u0d3e\u0d30\u0d46 \u0d05\u0d35\u0d30\u0d41\u0d1f\u0d46 ML \u0d2a\u0d4d\u0d30\u0d4b\u0d1c\u0d15\u0d4d\u200c\u0d31\u0d4d\u0d31\u0d41\u0d15\u0d7e \u0d35\u0d47\u0d17\u0d24\u0d4d\u0d24\u0d3f\u0d7d \u0d38\u0d1c\u0d4d\u0d1c\u0d40\u0d15\u0d30\u0d3f\u0d15\u0d4d\u0d15\u0d3e\u0d7b \u0d38\u0d39\u0d3e\u0d2f\u0d3f\u0d15\u0d4d\u0d15\u0d41\u0d28\u0d4d\u0d28\u0d24\u0d3f\u0d28\u0d4d \u0d30\u0d42\u0d2a\u0d15\u0d7d\u0d2a\u0d4d\u0d2a\u0d28 \u0d1a\u0d46\u0d2f\u0d4d\u200c\u0d24 \u0d12\u0d30\u0d41 \u0d38\u0d2e\u0d17\u0d4d\u0d30 \u0d1f\u0d42\u0d7e\u0d15\u0d3f\u0d31\u0d4d\u0d31\u0d4d \u0d35\u0d3e\u0d17\u0d4d\u0d26\u0d3e\u0d28\u0d02 […]<\/p>","protected":false},"author":11,"featured_media":1432,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"om_disable_all_campaigns":false,"footnotes":""},"categories":[11],"tags":[],"class_list":["post-1431","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-startup-news"],"yoast_head":"\n