• Startup Pulse
    Humyn Labs raises $47 million in a Series A round, comprising $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton JohnAIC raises $1.9 million in a pre-seed funding round led by Crane Venture Partners, with participation from Neon Fund. Dreamteam Raises $53 Million to Take on Salesforce and HubSpot With AI Coworkers Deeptech Aerospace Startup Tessat Space Raises ₹1 Crore in Angel Funding Round from Xylapay Morsa raises $1.5 million in a pre-seed funding round led by Jackson Square Ventures, with participation from the a16z Scout Fund. Kutlerri.ai raises $4 million from Narahari Investments. JSW Ventures completes the first close of Fund III, with JSW Family Office as anchor investor. Desible.ai raises ₹32 crore in a funding round led by Prime Venture Partners, with participation from Invention Engine. mistEO raises ₹18 crore in a seed round led by MGF Kavachh, with Seafund and other investors participating. Humyn Labs raises $47 million in a Series A round, comprising $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton JohnAIC raises $1.9 million in a pre-seed funding round led by Crane Venture Partners, with participation from Neon Fund. Dreamteam Raises $53 Million to Take on Salesforce and HubSpot With AI Coworkers Deeptech Aerospace Startup Tessat Space Raises ₹1 Crore in Angel Funding Round from Xylapay Morsa raises $1.5 million in a pre-seed funding round led by Jackson Square Ventures, with participation from the a16z Scout Fund. Kutlerri.ai raises $4 million from Narahari Investments. JSW Ventures completes the first close of Fund III, with JSW Family Office as anchor investor. Desible.ai raises ₹32 crore in a funding round led by Prime Venture Partners, with participation from Invention Engine. mistEO raises ₹18 crore in a seed round led by MGF Kavachh, with Seafund and other investors participating.
  • Unsiloed AI Gets $7.5 Million From Top Investors to Turn Messy Business Documents Into AI Intelligence

    Unsiloed AI Co-Founders Aman Mishra and Adnan Abbas

    San Francisco: Unsiloed AI has raised $7.5 million in a seed funding round led by venture capital firm Felicis, with participation from Y Combinator, Orange Collective and Uncorrelated Ventures, to develop infrastructure that enables artificial intelligence models and agents to understand, process and reason over complex enterprise documents.

    Angel investors Aarthi Ramamurthy and Beth Turner also participated in the round.

    The San Francisco-based startup is addressing a fundamental challenge in enterprise AI adoption: making information trapped in unstructured business documents accessible to AI systems.

    While advances in large language models (LLMs) have expanded the capabilities of AI applications, enterprises continue to hold much of their operational knowledge in PDFs, spreadsheets, presentations, financial statements, legal agreements, charts and other document formats.

    These files contain information that conventional AI systems cannot always interpret reliably without specialised processing infrastructure.

    Unsiloed AI is building APIs that convert such multimodal, unstructured documents into structured Markdown and JSON formats that AI models and agents can consume directly.

    Its broader objective is to create an infrastructure layer through which AI systems can retrieve information from business documents, understand relationships between data points and use that information in downstream workflows.

    IIT Kharagpur Alumni Build Enterprise AI Infrastructure Startup

    Unsiloed AI was founded by Aman Mishra and Adnan Abbas, former classmates at the Indian Institute of Technology (IIT) Kharagpur. The company is headquartered in San Francisco and was part of Y Combinator’s Fall 2025 batch.

    The founders bring experience across financial technology, multimodal AI and autonomous systems.

    According to the company’s Y Combinator profile, Mishra previously built an ultra-low-latency trading system at a hedge fund. He was also the first engineer at a San Francisco startup developing AI copilots for financial firms.

    Abbas previously built multimodal models deployed at a Fortune 10 company and worked on autonomous navigation systems at Mercedes-Benz.

    Mishra announced the $7.5 million funding round in his post on X. In a follow-up post, he thanked Y Combinator and other investors and angels in this round.

    Mishra said the company has also brought together machine learning researchers and engineers from Google DeepMind, Cognition and quantitative trading firms.

    Unsiloed AI Targets a Critical Infrastructure Gap in Enterprise AI

    Enterprise AI adoption increasingly depends on whether AI models and agents can work with the information organisations already possess. However, much of that information is not stored in structured databases that software can query directly.

    Finance, legal services and healthcare are particularly document-intensive sectors. Financial institutions work with statements, reports and transaction records, law firms handle contracts and case documents, while healthcare organisations manage extensive records and reports.

    For AI agents expected to perform tasks across these environments, simply generating responses from general-purpose language models is not sufficient.

    They also need access to accurate information contained in enterprise documents, the ability to interpret different document structures and mechanisms to trace extracted information back to its original source.

    This is the problem Unsiloed AI is attempting to address.

    In his announcement, Mishra said finance, law and healthcare run on documents rather than databases. The company uses computer vision and dual-stream models to process complex multimodal documents and convert them into formats that AI systems can understand and reason over.

    Mishra described the technology as a “computable layer” that would allow AI agents to query documents the way software queries tables.

    The distinction is important to the company’s positioning. Rather than building another general-purpose AI model, Unsiloed AI is developing specialised infrastructure that connects existing enterprise information with the AI models and agents expected to operate on top of it.

    From Document Extraction to AI-Ready Enterprise Information

    Unsiloed AI’s APIs are designed to handle different stages of document processing, from interpreting source files to extracting information in formats that can be used by AI applications.

    According to the company’s documentation, its platform supports four core operations.

    The first is document parsing, which converts complex files into structured Markdown. This includes PDFs, presentations, spreadsheets, tables, charts and images.

    The second is structured data extraction. The API can identify and extract typed fields against a predefined schema, providing confidence scores and citations indicating where each extracted value appears in the original document.

    The third operation involves splitting files containing multiple documents into individual components, while the fourth classifies documents based on their content.

    Together, these capabilities are intended to make enterprise documents more than searchable files. They provide AI applications with structured information that can be retrieved, interpreted and used in specific business processes.

    Source-level citations and confidence scores also address an important requirement in enterprise AI deployments: the ability to verify where information originated rather than relying entirely on an AI-generated response.

    For AI agents operating in financial, legal and other regulated environments, this traceability can be particularly relevant when decisions and automated workflows depend on information extracted from business records.

    Fortune 150 Banks and NASDAQ-Listed Companies Among Customers

    Unsiloed AI says its technology is already being used by vertical AI startups, financial services companies and law firms.

    According to its Y Combinator profile, the startup processes millions of pages of complex documents every week for Fortune 150 banks, NASDAQ-listed companies and early-stage startups.

    The company also claims that its models outperform offerings from LlamaIndex, Gemini, Mistral and Unstructured.io on public benchmarks.

    Enterprise deployment requirements extend beyond document extraction accuracy. Organisations handling sensitive financial, legal and business information also need control over where data is processed and how their infrastructure is deployed.

    Unsiloed AI says its technology can operate in highly regulated environments, including fully air-gapped deployments on customers’ own infrastructure.

    This allows organisations to deploy the platform within their own infrastructure rather than depending on conventional internet-connected environments.

    The combination of multimodal document processing, structured extraction and controlled deployment is central to the company’s enterprise AI proposition.

    Unsiloed AI Experiments With Autonomous API Purchasing

    Beyond its document intelligence infrastructure, Unsiloed AI has introduced an agent storefront on its website.

    The storefront allows AI agents to access live pricing information and purchase the company’s APIs either autonomously or with human approval.

    The feature extends the company’s focus on AI agents beyond document processing, providing a mechanism through which agents can discover and independently procure the software infrastructure they need.

    Author

    • Salil Urunkar

      Salil Urunkar is a senior journalist and the editorial mind behind Sahyadri Startups. With years of experience covering Pune’s entrepreneurial rise, he’s passionate about telling the real stories of founders, disruptors, and game-changers.

    Back to top