With a mission to build AI employees, Bengaluru-based generative AI startup Alchemyst AI has raised $300K in its pre-seed round from key investors, including Inflection Point Ventures, 100 Unicorns (formerly 9U), and EarlySeed Ventures, reaching a valuation of $2 million.
“We want to build an entire ecosystem of generative AI based digital employees, backed by our own AI infrastructure,” said Uttaran Nayak, the founder of Alchemyst AI, in an exclusive interview with AIM.
“We want to create a stagnant and simple ecosystem where these GenAI employees will work as hyper-intelligent AI assistants for humans and interact with each other to solve high-level intelligent problems in these enterprises,” added Nayak.
Inside Alchemyst’s AI Infrastructure
Nayak told AIM that they have created their own model, AlchemystC1, by combining Phi-3 vision and an Indic LLM from AI4Bharat. He said that they are not using any of the popular models like OpenAI’s GPT-4o or Google Gemini due to concerns from enterprises about their data being exposed on the web.
“In the next three to four months, we are planning to extend into 28 languages, particularly to cater to Indian customers,” he said.
Regarding the issue of hallucinations, he said, “We have fine-tuned our model to say ‘I don’t know’ rather than producing gibberish. This requires more data, as maximum reliability in instruction-following is essential from any employee across an organisation.”
Alchemyst’s chief technology officer, Anuran Roy, explained that the company has developed a dedicated mesh architecture solely for storing and auditing data over time.
Is Employer’s Data Safe with AI Employees?
Roy said that AlchemystC1 does not have direct access to employees’ emails. “We have stored the data in a separate mesh,” he said and explained that customers can define scopes, restricting LLMs from accessing certain data.
“We also have a BYOM (bring your own model), but AlchemystC1 guarantees that no data prompts leave the company. Everything is secured in compliance with GDPR, HIPAA, SOC, and other strict standards,” he added.
The company currently utilises multiple cloud service providers, including AWS, Microsoft Azure and GCP, for LLM inference.
“We use a self-hosted database for our needs. For LLM inference, we host containers on AWS and bare metal providers. As we move forward, our goal is to consolidate our infrastructure to better manage costs while scaling over time,” shared Roy.
The startup is in talks with E2E Networks to acquire more GPUs for LLM inference.
Digital Employees to the Rescue
Founded by school friends Nayak and Anuran Roy, Alchemyst AI was officially launched around the end of November 2023. The team consists of six members, all technical experts.
Nayak said that he worked with a B2B SaaS company during his third year of engineering when the idea came to him. “That was the first time I entered a whole new world of SaaS and realised that if you’re building a product for $500, you can sell it for $50,000 to all those clients,” he said.
Alchemyst’s AI digital employees currently cater to various sectors, including sales, marketing, HR, customer support, and analytics. Nayak told AIM that these digital employees will assist in lead generation, email management, and meeting scheduling, among other tasks, thereby optimising overall productivity.
The company’s flagship product, Maya, is an AI sales development representative (SDR) that automates enterprise sales processes. Maya can research prospects and compose hyper-personalised messages without human intervention, performing 50 times faster than humans.
The company plans to introduce more AI employees in the near future. Currently, Moh serves as the AI marketer, Leela is the AI HR manager, Ron is the AI customer support representative, and Sam is an AI analytics specialist.
“We want these AI employees to interact with each other and solve problems within the enterprise. They don’t need to shift their entire architecture or infrastructure; they just have to adopt a plug-and-play concept,” explained Nayak.
These AI employees can be integrated into Hubspot, Gmail, Jira, Salesforce, WhatsApp, Telegram, Zoho and ServiceNow. “We will be integrating with WhatsApp and Telegram very soon. These integrations will happen quickly, especially for sales activities on these platforms. This is particularly relevant in the Web3 space, where there are numerous use cases,” said Nayak.
Notably, Salesforce recently launched Agentforce Partner Network that brings together tech giants such as AWS, Google Cloud, IBM, and Workday to enhance the AI-powered Agent Force platform’s capabilities.
Meanwhile, OpenAI has introduced a new approach for creating and deploying multi-agent AI systems, called the Swarm framework. It simplifies the process of creating and managing multiple AI agents that can work together seamlessly to accomplish complex tasks.
Similar to Ema?
The company’s solution is similar to what former Coinbase CPO Surojit Chatterjee’s Ema is building. Recently, Ema launched the ‘Universal AI Employee’, which can take on various roles within an enterprise. The company also raised an additional $36 million in Series A funding, bringing its total capital raised to over $61 million.
The company’s Persona Builder Platform allows businesses to create and deploy custom AI Personas tailored for specific roles without extensive model training. This new tool is expected to enhance how companies integrate AI into various enterprise functions including customer support, sales, and compliance.
“Ema is essentially building what we are—one universal AI employee for enterprise teams, complete with its own AI architecture,” said Nayak, acknowledging their presence in the agentic AI market.
Building $200 Mn Company in the Next Three Years
Nayak further shared that he recently met with Apoorva Ranjan Sharma, the co-founder and director of 9Unicorns and Venture Catalyst. “He showed me the entire trajectory and roadmap of how the companies go from there. We are looking to build a $200 million company in the next three years,” he said. Moreover, the company plans to expand into the US, UAE, and Southeast Asia.
The post This Bengaluru Startup Secures $300K to Build AI Employees for Enterprises appeared first on AIM.
