Komprise fights Model Context Protocol bloat

Komprise launched Universal File MCP, a single interface that lets AI agents and LLMs query across enterprise file storage, NAS, cloud and unstructured data silos, the company announced. Komprise CEO Kumar Goswami said the product "lifts the veil on dark enterprise data by unifying governed access across silos," and the company cites a McKinsey report finding that 60 percent of agentic AI computing costs go to response refinement. Komprise says the tool right-sizes responses and enforces user-specific access permissions to counter MCP bloat from multiple vendor tool definitions and unbounded file listings.

Komprise fights Model Context Protocol bloat Komprise has launched Universal File MCP, a single interface for AI agents and LLMs to query across enterprise file storage, NAS, cloud, and unstructured data silos. MCP https://www.blocksandfiles.com/data-management/2022/04/02/mcp/1613060? gl=1 1qbpaz1 ga MzkxNDQyMTIwLjE3NzcwMzc0NTc. ga NSDTXHMMN0 czE3OTA3NTk2NDAkbzQyNyRnMSR0MTc5MDc1OTY0OCRqNjAkbDAkaDA. stands for Model Context Protocol and it provides secure, bi-directional API-level communication between AI large language models LLMs and agents, and external data sources or tools. MCP Servers handle requests from AI clients, process them, and return contextual responses. Komprise CEO Kumar Goswami said: “The Komprise Universal File MCP lifts the veil on dark enterprise data by unifying governed access across silos. We’re thrilled to help our customers easily leverage AI across all of their file data by asking complex questions and initiating workflows to act on the results, all while ensuring access controls remain intact.” There’s an age-old problem here. Applications needing to act on or look at an organization’s unstructured data estate typically act on multiple separate data sources. It takes time to inspect and correlate a data index across distributed silos. Do you decide to store everything in one supplier’s silos; go all-in on NetAp or Dell, etc. say? Or do you use a third party’s data estate discovery and classification system to create a single virtual index without moving data from the source silos? Komprise is in the latter camp and aims to provide cross-supplier and cross-silo access to distributed, multi-supplier data estates. What is MCP bloat? According to Komprise, as nearly every vendor publishes a MCP interface, AI is getting overloaded with multiple tool definitions, resulting in lower accuracy, slow performance and higher costs. The problem compounds with unstructured data response bloat as file and object listings are unbounded and can be millions to billions of files across disparate hybrid storage. Komprise says it right-sizes responses by sending just the right unstructured data, enriched with context and governed by user-specific access permissions. The company asserts that AI tools process too much wrong data, and, citing a McKinsey report https://www.marketscale.com/industries/software-and-technology/60-of-agentic-ai-costs-go-to-response-refinement-and-most-enterprises-are-already-over-budget , 60 percent of agentic AI computing costs are consumed entirely by “response refinement,” which is the behind-the-scenes token loops where sub-agents critique and format data before presenting a final answer. We understand this means that AI-using organizations should send a single MCP request to Komprise rather than multiple MCP requests to individual data sources, and that Komprise’s SW will right-size responses shrink them by sending just the right unstructured data, enriched with context and governed by user-specific access permissions. Its Universal File MCP provides: - Governed, Secure Access: Authenticated access only delivers responses based on the user's privileges and what they are authorized to see. It audits what data is sent to AI for governance and reporting. - Consistent Schema: Leverages the petabyte-scale Komprise Global Metadatabase https://www.blocksandfiles.com/data-management/2026/02/20/komprise-launches-kappa-to-hunt-metadata-across-enterprise-file-silos/4091645https://www.komprise.com/product/global-metadatabase/ to provide uniform structure for unstructured data across storage. AI gets a consistent way to query, regardless of who produced the data or where the data currently lives. - High-Quality, Enriched Data: Komprise AI Preparation & Process Automation KAPPA https://www.komprise.com/product/kappa-data-services/ extracts contextual metadata based on industry, enterprise, sensitivity, users, and other context to support specific AI use cases. - Noise Filters: Komprise Deep Analytics https://www.blocksandfiles.com/data-management/2021/10/12/komprise-adds-global-unstructured-data-search-and-subset-move/1610986 allows users to eradicate irrelevant, outdated, conflicting and duplicate data, resulting in high-quality inputs. - Progressive Disclosure with Transparent File Tables: Komprise Universal File MCP enables progressive loading of information to AI, beginning with metadata for summarization and the option to export results as an Apache Iceberg table using Komprise Transparent File Tables https://www.blocksandfiles.com/data-management/2026/06/23/komprise-provides-lakehouse-access-to-petabytes-of-unstructured-data/5260267 . It then loads files only when the AI needs it, further reducing response bloat. - Works on Tiered Data: Komprise Universal File MCP serves up data even as data moves, so tiered and archived data remain accessible via Komprise Transparent Move Technology https://www.blocksandfiles.com/file/2020/03/31/komprise-kedm-migrates-file-data-6-times-faster-than-the-status-quo-and-at-half-the-cost/1609301? gl=1 49zkh1 ga MzkxNDQyMTIwLjE3NzcwMzc0NTc. ga NSDTXHMMN0 czE3OTA3NTk2NDAkbzQyNyRnMSR0MTc5MDc2MDgzMSRqNjAkbDAkaDA.https://www.komprise.com/product/transparent-move-technology/ . - Ingests into AI, Lakehouses and Analytics: Komprise Universal File MCP enables follow-up actions from the prompt, such as ingesting the results into AI or a lakehouse for further processing using Komprise Intelligent AI Ingest. https://www.blocksandfiles.com/ai-ml/2025/09/23/komprise-launches-ai-focused-ingest-tool-to-clean-up-unstructured-data/1612772 A Komprise customer likes this. Stephen Clark, Director of Information Security at Children’s Medical Center, Dallas, said: “Opening up clinical and research data to AI tools means my team needs to review who could see what, case by case, putting security in the middle of every request. With Komprise Universal File MCP, the query respects the permissions a user already has and leaves us a record of what went to the model. Our researchers get their data faster, and IT is not managing a separate connector for every storage system.” Universal File MCP solves the multiple separate connector issue for Clark. Komprise Universal File MCP is available today in an early access program for customers and partners. Learn more in a blog https://www.komprise.com/blog/interview-komprise-universal-file-mcp/ .