Pranjali Awasthi, a teenage entrepreneur of Indian origin based in the United States, has reportedly propelled her startup Delv.AI to a valuation approaching ₹100 crore (approximately $12 million). The company, which focuses on research-data extraction and workflow automation powered by generative AI, is drawing attention not only for its rapid ascent but also for the young founder’s ambition to address trust and efficiency in knowledge work.

Awasthi launched Delv.AI with the goal of eliminating the drudgery of manual research. The platform uses advanced language models to parse vast datasets, academic papers, and internal documents, then surfaces key insights in seconds. Early adopters describe it as a “co-pilot for researchers,” capable of cross-referencing sources, flagging inconsistencies, and even generating structured reports. The startup’s recent valuation, though not officially confirmed by financial filings, has been widely reported in tech circles and signals intense investor interest in tools that promise to democratize high-level analysis.

The idea for Delv.AI came to Awasthi while she was still in high school. Frustrated by the amount of time she spent searching for verified information across dozens of tabs, she built a prototype that used natural language queries to aggregate and summarize content. That prototype evolved into a full-fledged platform now used by university labs, policy think tanks, and even corporate R&D departments. The company’s value proposition rests on three pillars: speed, accuracy, and workflow integration.

Delv.AI’s Core Technology

At its heart, Delv.AI is a research automation engine built on top of large language models (LLMs) and custom retrieval-augmented generation (RAG) pipelines. The system ingests documents—PDFs, web pages, internal wikis—and indexes them into a secure vector database. When a user asks a question, the engine retrieves semantically relevant chunks and synthesizes an answer complete with citations. This architecture mirrors recent advances in enterprise search but is fine-tuned for research-grade accuracy.

The platform stands out by offering a “trust score” for each claim it surfaces. Users can see whether a piece of information is backed by a single source, multiple independent studies, or only internal documents. This feature addresses one of the most significant barriers to adopting generative AI in critical research: hallucination. By tying every assertion back to a retrievable passage, Delv.AI allows researchers to verify outputs quickly, without chasing footnotes.

Awasthi’s team has also worked on domain-specific models. A policy researcher analyzing climate legislation, for example, can select a model that understands regulatory language and cites relevant statutes. The underlying infrastructure runs on a hybrid cloud, giving enterprises the option to keep sensitive data on-premises while still tapping into the power of frontier models. The company has not disclosed which LLM providers it uses, but analysts suspect a combination of open-source models (such as Llama 2 and Mistral) and commercially licensed APIs.

The Teen Founder Behind the Code

Pranjali Awasthi’s story resonates because it challenges the stereotype of the Silicon Valley founder. She began coding in middle school, teaching herself Python through online courses. By 16, she had won several hackathons with projects that used computer vision to assist the visually impaired. Delv.AI was born out of a science-fair project that caught the eye of early-stage investors. Awasthi relocated from India to the United States to pursue entrepreneurial opportunities and enrolled in an accelerator program that provided initial funding.

In interviews, she emphasizes that her age has been both an asset and a hurdle. “Investors often did a double-take when I walked into the room,” she told a podcast last year. “But once I demonstrated the product, the conversation shifted from my age to the problem we were solving.” That problem is universal: knowledge workers spend up to 19% of their time just searching for information, according to a McKinsey study. Awasthi’s pitch was that Delv.AI could cut that figure in half.

Her leadership style blends technical hands-on work with a keen focus on user experience. She regularly joins customer calls to hear pain points directly. This feedback loop has led to features like customizable dashboards that allow each research team to define its own quality criteria. The startup now employs over 30 people, split between a headquarters in San Francisco and a small engineering hub in Bangalore.

Valuation and Funding Landscape

The reported ₹100 crore valuation places Delv.AI in a competitive bracket of early-stage AI startups. While the precise funding round has not been made public, people familiar with the matter say it involved a mix of angel investors and a New York-based venture capital firm. The capital is being used to scale the sales team, strengthen the trust engine, and build integrations with popular tools like Microsoft 365, Google Workspace, and Slack.

Industry watchers note that valuations in the generative AI space have been frothy. However, Delv.AI’s focus on research—a domain where the cost of error is high—gives it a defensible moat. Unlike general-purpose chatbots, its engine is designed for rigorous citation and audit trails. That makes it attractive to law firms, pharmaceutical companies, and academic institutions that cannot tolerate black-box answers.

The startup’s business model follows a subscription tier, with a free tier for individual researchers and enterprise plans that scale by the number of documents indexed and queries run. One early enterprise customer, a biotech firm in Boston, reportedly reduced the time to complete a literature review from two weeks to three days using Delv.AI’s automated meta-analysis feature. Awasthi has hinted at a forthcoming collaboration with a major university press to embed the platform into its manuscript submission system.

Workflow Trust and the Human-in-the-Loop

“Workflow trust” is a phrase that appears frequently in Delv.AI’s messaging. Awasthi argues that AI adoption in research will stall unless users can confidently incorporate AI-generated insights into their decision-making pipelines. To that end, the platform includes a human-review mode where flagged assertions are queued for expert verification. Over time, the system learns which sources a particular organization trusts most and weights them accordingly.

This approach resonates with Windows-centric enterprises that run locked-down desktop environments. Delv.AI offers a progressive web app (PWA) that integrates with Windows Hello for biometric authentication. IT administrators can enforce group policies to control which document repositories the platform can access. For end-users, the experience is seamless: a sidebar in Microsoft Edge or a pinned tile in the Windows 11 Start menu can launch the research assistant directly.

Security is paramount because research often involves pre-publication data. Delv.AI holds SOC 2 Type II certification and is pursuing FedRAMP authorization to serve U.S. government agencies. All data encryption happens over AES-256, and the vector database is isolated per tenant. The company’s policy is not to use customer data for model training unless explicitly opted in.

Implications for the Windows Ecosystem

Windows remains the dominant operating system in enterprise and academic settings, which means any research tool that gains traction must play nicely with it. Delv.AI’s decision to build a PWA and a native Windows app (via the Windows App SDK) signals a commitment to the platform. Users can highlight text in a Word document, right-click, and select “Ask Delv” to get an instant analysis without leaving the document. Integration with OneDrive and SharePoint allows the AI to index files stored in those services.

For Windows enthusiasts, this represents another example of AI moving beyond the browser tab and into the operating system. Microsoft’s own Copilot is pushing in a similar direction, but Delv.AI’s strength lies in research-specific workflows that general assistants may not handle as deeply. There is also talk of a future integration with Power Automate, letting researchers trigger multi-step literature searches based on calendar events or email notifications.

Awasthi has not ruled out a deeper partnership with Microsoft. During a recent tech demo, she showed Delv.AI running on an Azure virtual desktop, with the AI suggesting changes to a grant proposal drafted in Word. “The goal is to make the operating system and the research assistant feel like one cohesive unit,” she said. Windows 11’s recent AI-focused updates, including the Copilot key on new keyboards, suggest the ecosystem is ripe for specialized assistants like Delv.AI.

Community Reactions and Skepticism

While the news of the valuation has been largely met with excitement—especially in Indian-American tech circles—some on discussion forums have voiced skepticism. “Another AI wrapper with a shiny demo?” one commenter on WindowsForum wrote. Others pointed out that Delv.AI’s trust-score system is only as good as its source ranking, and that bad actors could game it by flooding the system with convincing but misleading papers.

A separate thread questioned whether any AI can truly understand context enough to replace a skilled researcher. “I spend years developing expertise in my field,” a university professor posted. “I’m not convinced a teenager’s startup can replace that.” Defenders countered that Delv.AI is not meant to replace researchers but to handle the tedious cross-referencing that eats up billable hours.

Awasthi herself has engaged with critics online, responding thoughtfully to technical questions about the platform’s citation mechanism. “We don’t claim to replace human judgment,” she wrote in one reply. “We aim to give you a faster, more transparent starting point.” This candor has won her allies among early adopters who appreciate that the company isn’t overselling machine intelligence.

The Road Ahead

Delv.AI’s roadmap includes several features that could further entrench it in the Windows workflow. A plug-in for Visual Studio Code is in development, targeting data scientists who want to query research papers directly from their coding environment. There are also plans for an agent that can autonomously monitor pre-print servers and Slack channels, then push curated summaries to a team dashboard each morning.

Awasthi is careful to frame the company’s growth in terms of responsible AI. She has hired a full-time ethics advisor to audit the trust-score algorithms and ensure they don’t inadvertently penalize non-English sources. The team is working on support for multiple languages, with Hindi and Spanish slated for a beta release later this year.

Investor appetite for AI startups shows no signs of cooling, but the window for early-stage valuations may be narrowing as open-source models commoditize basic chat functionality. Delv.AI’s bet is that deep domain-specific trust mechanisms will keep it ahead. If the company can deliver on that promise, the ₹100 crore figure may soon look conservative.

For Windows users watching the AI revolution unfold, Delv.AI offers a glimpse of how specialized tools can coexist with platform-level assistants like Copilot. Rather than one monolithic AI that knows everything, the future may be a mosaic of focused assistants that users call upon for specific tasks. Awasthi’s startup is positioning itself to be the default choice when the task is “research this.”

As the generative AI landscape matures, Delv.AI’s story serves as a reminder that innovation can come from anywhere—even a teenager’s frustration with homework. The coming months will test whether the company can live up to its valuation, but one thing is clear: the conversation around AI in research has only just begun.