Fay.
Artificial intelligence has moved rapidly from experimentation to adoption inside businesses, but a significant gap remains between what AI can demonstrate and what it can reliably do. Companies may have access to increasingly capable models, yet turning that intelligence into repeatable, production-ready business processes often continues to require technical teams, lengthy implementation cycles and fragmented systems.
It is this gap between AI's capabilities and its practical application inside businesses that entrepreneurs Harsh Nigam and Rohit Hazra are attempting to address with Fay, a new AI agent and workflow platform from PropulsionAI, Inc.
The launch of Fay marks the latest chapter in a journey that began almost two decades ago, when Nigam and Hazra first met in 2008 and began exploring technology and entrepreneurship together. Their first major entrepreneurial venture came in 2016, when they established a software consultancy that eventually grew into a 50-person organisation, delivering more than 100 products for customers globally and generating millions in revenue.
That experience exposed the founders to a recurring problem within enterprises: while businesses invest heavily in technology, many of their most important processes remain dependent on manual coordination, repetitive tasks and operational handoffs.
In 2023, the two founders made a deliberate decision to move away from consultancy work and focus entirely on artificial intelligence. That transition led to PropulsionAI, an enterprise AI platform built around helping companies develop and deploy custom AI systems. The company was subsequently accepted into UC Berkeley's Berkeley SkyDeck accelerator, giving the founders an opportunity to build within one of the world's most active technology ecosystems.
The Founders Behind Fay
Harsh Nigam and Rohit Hazra's partnership spans nearly two decades. After meeting in 2008, the two went on to build a software consultancy in 2016 that grew into a 50-person team and delivered more than 100 products for global customers. Their years of working closely with businesses gave them firsthand insight into the operational challenges enterprises face when adopting technology. In 2023, they shifted their focus entirely to AI, founding PropulsionAI and later joining UC Berkeley's Berkeley SkyDeck accelerator. With Fay, the founders are bringing that experience together to build AI systems designed not merely to assist employees, but to take on meaningful operational work.
Fay represents the next evolution of that work.
Rather than positioning AI as another interface for asking questions, Fay is designed around the idea of AI as an operational worker. The platform enables non-technical teams to describe a business process in plain language and use AI to help design, configure and deploy agents and workflows capable of carrying out that work.
"We spent years building software for companies and seeing firsthand where technology breaks down inside real organisations. The challenge was rarely whether a company had enough software. It was whether that software could actually take work off people's plates. With Fay, we wanted to build toward that outcome, an AI that doesn't just answer a question, but can take responsibility for getting something done," shared Harsh Nigam, Co-founder, PropulsionAI, Inc.
Moving Beyond the AI Demo
The enterprise AI market has largely been shaped by increasingly capable foundation models, copilots and conversational interfaces. Yet for many businesses, the harder question is what happens after the conversation.
An employee may use an AI assistant to draft an email or analyse a document, but operational work often involves multiple systems, decisions, approvals and follow-up actions. A customer issue may require information from a CRM, communication through email, an internal update on Slack and escalation through another system. A finance workflow may involve reviewing information, applying rules, requesting approvals and updating records.
These processes cannot always be solved by a single prompt.
Fay is designed around this distinction. Its platform combines autonomous AI agents with deterministic workflows, allowing businesses to use AI's reasoning capabilities where judgment is required while retaining predictable automation where processes need consistency.
At the centre of the platform is Architect, an AI builder intended to help non-technical teams create and configure agents and workflows without depending entirely on engineering resources.
The platform can connect with workplace systems such as Slack, Gmail, Jira, PagerDuty, webhooks and other business tools, allowing agents to operate within existing workflows rather than forcing companies to replace their technology stack.
Its applications span customer operations, people operations, finance, business operations, compliance, IT operations, procurement, revenue operations and supply chain - areas where repetitive operational work can consume significant amounts of employee time.
"AI adoption cannot be measured simply by how many models a company has access to or how many employees are using a chatbot. The real question is whether AI can become part of the operating system of the business," said Rohit Hazra, Co-founder, PropulsionAI, Inc. "We see Fay as a way to close that gap, giving the people closest to a workflow the ability to build an AI worker for it, without having to wait for a long engineering cycle."
An AI Platform Designed Around Proof, Not Promises
One of Fay's more distinctive approaches is its Proof Week model.
Instead of asking companies to commit to an abstract AI transformation programme, the approach starts with an actual operational problem. A workflow is identified, connected to relevant company data and developed into a working AI agent within five days.
The objective is straightforward: allow a business to see what the agent actually does, how much work it can remove from a team's workload and whether the use case generates enough value to justify continued deployment.
This approach reflects a broader shift in how enterprises are beginning to evaluate AI. The conversation is moving away from experimentation for its own sake toward measurable outcomes - time saved, processes completed, response times improved and operational capacity created.
Fay is also designed to remain model-agnostic, supporting leading models including GPT-5.6, Claude, Gemini and open-source models. This allows businesses to select models based on the requirements of a particular task rather than committing their workflows to a single model ecosystem.
For enterprise organisations, the platform also incorporates controls around security, access and deployment, including role-based access, SSO/SAML and options for VPC or on-premise deployment. PropulsionAI says customer data is not used to train its AI models.
From Building Software to Building the Future of Work
For Nigam and Hazra, Fay is not an isolated product launch. It represents the culmination of a progression that has taken them from software services to enterprise AI infrastructure and now to AI agents capable of performing operational work.
Their earlier experience remains central to that thesis.
Having spent years building products for businesses, the founders have seen the limitations of conventional software from the inside. Enterprise software has historically required people to adapt their work to the logic of the tools they use. The emerging AI paradigm, they believe, could reverse that relationship, allowing employees to describe what needs to happen while AI handles an increasing portion of the execution.
"The long-term vision is bigger than automation. We believe the next generation of enterprise software will be less about giving people another tool to operate and more about giving them intelligent systems that can operate alongside them. Fay is our attempt to build toward that future, starting with the operational work that businesses deal with every day," reflected Harsh Nigam.
That vision also explains the company's emphasis on non-technical teams. The founders believe the people who understand a workflow most deeply are often the people performing it, not necessarily the engineers responsible for building software. By allowing those teams to translate their operational knowledge into AI agents, Fay aims to make AI deployment less dependent on specialised technical resources.
For PropulsionAI, the opportunity is therefore not simply to participate in the crowded enterprise AI market, but to redefine where AI becomes useful inside an organisation.
The company's journey from a consultancy founded by two long-time friends to a San Francisco-based AI company reflects the broader evolution of enterprise technology itself. The first phase was about putting software into businesses. The next was about putting intelligence into software.
Fay is built around a more ambitious proposition: putting AI to work.
