# fileyap: Sovereign Private Intelligent Search for your files > fileyap is Sovereign Private Intelligent Search for your files. Ask a question across every data format and get the answer with every source passage highlighted, from a system that runs fully offline, inside your own infra or on the secure fileyap cloud. A Sovereign AI company for private data. Private beta: The first waitlist cohort onboards in November 2026. Source: https://fileyap.com/ (this file is generated from the same content as the page) ## Overview Sovereign Private Intelligent Search for your files. Ask a question across every data format. Get the answer with every source passage highlighted, from a system that runs fully offline, inside your own infra or on the secure fileyap cloud. Runs: On-premises, Sovereign cloud, Air-gapped, Private intelligence. ## The problem: Everything your company knows is buried in its files. In regulated, contract-heavy companies, much of the day goes on hunting for the right contract, clause or version. The tools they have weren't built to find answers, and the AI that could help isn't allowed in. - **Search finds file names, not answers.** Shared drives match words in titles, so people open dozens of files to find one clause, one figure or the latest version. - **Decisions get lost.** The reason behind a decision sits in an email thread, a call recording or an approval comment, where nobody can find it later. - **Knowledge walks out the door.** When someone leaves, what they knew about clients, contracts and past decisions leaves with them. - **Cloud AI isn't allowed in.** Client files and contracts can't go to a vendor's servers, and the private options are built for the largest banks and governments. ## What we learned Regulated companies don't want an AI platform. They want answers they can defend. They stay accountable for whatever their AI says. So every answer needs a clickable source, the AI must never see a file the user can't, and often the files can't leave the building. From conversations with security, compliance, risk and technical leaders at regulated, contract-heavy companies. ## How it works: Stop moving data. Move the question. Every copy of your data is a new risk and a new compliance problem. So fileyap never makes one. The AI comes to your files, inside your own infrastructure, and nothing leaves. 1. **Deploy where your data lives.** Install fileyap on your own servers, in a sovereign cloud region or fully air-gapped. One-click deployment, ready to use in seconds. 2. **Index everything in place.** Point it at your file shares, object storage and email archives. Every file is indexed where it already lives, into one searchable layer. 3. **The question travels. The answer comes back..** Ask in plain language, the way you'd ask a colleague. fileyap searches where the files live and replies with the exact passages it relied on. If it isn't in your files, it says so. 4. **Answers you can defend.** Answers respect the access rights you already manage, so nobody sees a file they couldn't open before. Every question, answer and source is logged for audit. ## Connectors and data fabric: Every system you run. One living context. The facts about one customer live in a dozen platforms: the contract in Google Drive, the deal in Salesforce, the tickets in Zendesk, the invoices in SAP. fileyap's connectors weave them into one data fabric that stays current, so a single question can reach across every domain. - **Connect.** Connectors read every file and record already there, then pick up each change as it happens. - **Unify.** Everything about the same customer, supplier or case is linked, and each source's permissions come with it. - **Store.** The storage connector keeps the whole fabric in the home you choose. Where the fabric is stored: fileyap premium cloud (A dedicated, encrypted tenant, run and monitored for you by fileyap.); Your enterprise cloud (Inside your own AWS, Azure or Google Cloud account, or your data centre.) ### Connectors for the platforms you already use - **Files and storage:** Google Drive, SharePoint, OneDrive, Dropbox, Box, Egnyte, Amazon S3, Azure Blob Storage, Google Cloud Storage, Network shares (SMB, NFS), SFTP - **Email and calendar:** Gmail, Outlook, Microsoft Exchange, Google Calendar - **Chat and meetings:** Slack, Microsoft Teams, Google Chat, Zoom, Google Meet - **CRM and sales:** Salesforce, HubSpot, Microsoft Dynamics 365, Pipedrive, Zoho CRM - **Support and service:** Zendesk, ServiceNow, Freshdesk, Intercom, Jira Service Management - **Knowledge and projects:** Confluence, Notion, Jira, Asana, monday.com, ClickUp, Linear - **ERP and finance:** SAP S/4HANA, Oracle NetSuite, Oracle Fusion, Microsoft Dynamics 365 Finance, Xero, QuickBooks, Sage - **HR and people:** Workday, SAP SuccessFactors, BambooHR, Personio, HiBob - **Legal and records:** iManage, NetDocuments, DocuSign, Adobe Acrobat Sign, Ironclad, OpenText, M-Files - **Data and warehouses:** Snowflake, Databricks, BigQuery, Amazon Redshift, PostgreSQL, MySQL, SQL Server, Oracle Database, MongoDB - **Engineering:** GitHub, GitLab, Bitbucket, Azure DevOps - **Industry systems:** Epic, Veeva Vault, Guidewire, Temenos, Procore, FHIR and HL7 Not listed? fileyap also connects to any database, REST API or MCP server, and we build custom connectors with you during the pilot. ## The core engine: An AI-native file system on a distributed data fabric. Search made the web findable. fileyap makes your own files answerable, for one person or a whole enterprise. Every format is understood, every change is indexed as it happens, and every search runs in milliseconds. ### AI-Native File System **Multimodal understanding.** Folders only know file names. fileyap reads what's inside every document, spreadsheet, scan, recording and email, and keeps each passage tied to its exact page, cell or timestamp, so every answer can be cited. Every format. One file system. ### Distributed Data Fabric **Real-time, distributed index.** One central index goes stale and slows down as it grows. fileyap spreads the index in shards across the fileyap premium cloud and your own servers, indexes every change from every connected system as it happens, and runs each question on every shard at once. Current, accurate and fast at any size. ## Industries: Wherever the answer is in a file you can't send out. Every regulated, document-heavy industry shares the same problem: the files that matter most can't leave. Pick one to see the kind of questions fileyap answers. ### Banking and financial services Client records, deal files and supervisory rules keep the most useful documents inside your walls. - Which credit agreements include a change-of-control clause? - Show every KYC file that is missing proof of address. - What did we tell the regulator about our liquidity buffer last quarter? Files it reads: Credit agreements, KYC files, Board minutes, Call recordings. ### Insurance Policyholder data, claims history and pricing models are too sensitive for a public AI. - Which policies exclude flood damage for coastal postcodes? - Summarise the adjuster notes on claim 48-2210. - How do this year's underwriting guidelines differ from last year's? Files it reads: Policy wordings, Claims files, Adjuster photos, Underwriting guides. ### Legal Privilege and client confidentiality come first, so matter files stay on your own systems. - Find every NDA we signed with a European supplier. - Which precedents cover liquidated damages in construction disputes? - What did the client agree about fees on last week's call? Files it reads: Contracts, Precedents, Matter emails, Call recordings. ### Professional services Engagement files and client deliverables are confidential by contract. - Which past proposals covered supply-chain risk for a retailer? - Find the final operating model we delivered to this client. - What did the steering committee sign off in March? Files it reads: Proposals, Engagement letters, Workshop recordings, Meeting notes. ### Accounting and audit Audit evidence and client ledgers have to stay where the client keeps them. - Which invoices over £10,000 have no matching purchase order? - Collect the evidence for this year's revenue recognition testing. - What changed in the lease schedule since the last audit? Files it reads: Ledgers, Invoices, Workpapers, Bank confirmations. ### Healthcare Patient records are among the most protected data there is. - Find the current version of the sepsis protocol. - Which discharge letters mention a follow-up scan? - Summarise this quarter's incident reports for ward 7. Files it reads: Clinical protocols, Discharge letters, Incident reports, Scanned referrals. ### Pharma and life sciences Trial data, lab records and regulatory submissions are tightly controlled and audited. - Which SOPs still reference the old cold-chain procedure? - Where does the submission justify the dose selection? - List the deviations reported at site 14 since March. Files it reads: SOPs, Trial master files, Lab notebooks, Submissions. ### Government and public sector Citizen data and sovereign workloads must stay under your own control. - Which planning applications cite the 2019 flood assessment? - What did the council decide on the bus-lane consultation? - Find every case file that mentions this address. Files it reads: Case files, Council minutes, Consultation responses, Scanned records. ### Defence and security Classified estates run air-gapped. fileyap runs there too. - Which maintenance reports mention the same hydraulic fault? - Summarise the lessons learned from the last three exercises. - Where is this part number referenced across the fleet manuals? Files it reads: Technical manuals, Maintenance logs, After-action reports, Imagery. ### Energy and utilities Documents about critical infrastructure are a security matter. - Which substations are overdue for inspection? - Find the safety case for the turbine retrofit. - What do our permits say about discharge limits? Files it reads: Asset records, Safety cases, Inspection photos, Permits. ### Manufacturing and engineering Designs, specifications and supplier terms are your competitive edge. - Has a similar requirement been solved in an earlier project? - Which suppliers are certified for this alloy? - Show every non-conformance report for part 7731. Files it reads: Specifications, Technical drawings, Quality records, Supplier contracts. ### Aerospace and automotive Certification evidence and engineering IP never leave the programme. - Trace this requirement to its verification evidence. - Which design changes affected the brake assembly? - Find the test reports for next year's homologation. Files it reads: Requirements, Test reports, Change requests, Test footage. ### Real estate and construction Leases, drawings and contracts are spread across firms, sites and decades. - Which leases have a break clause in the next 18 months? - Find the fire strategy for block C. - What did the contractor commit to in the variation orders? Files it reads: Leases, Drawings, Planning documents, Site videos. ### Telecommunications Network designs and subscriber data are regulated and security-sensitive. - Which sites use the same antenna model? - What did we commit to the regulator on rural coverage? - Find the contract terms for this roaming partner. Files it reads: Network designs, Site surveys, Partner contracts, Compliance reports. ### Transport and logistics Safety records and customs documents carry legal weight. - Which shipments are missing a certificate of origin? - Find the maintenance history for vehicle 2241. - What do our carrier contracts say about delay penalties? Files it reads: Customs documents, Maintenance logs, Carrier contracts, Delivery photos. ### Retail and consumer goods Supplier terms, pricing and customer data are commercially sensitive. - Which suppliers have agreed to the new packaging rules? - Find our product recall procedure. - What margin terms did we agree with this distributor? Files it reads: Supplier agreements, Product specs, Price lists, Store audits. ### Education and research Unpublished research, student records and interview data need to stay private. - Which grant reports mention this dataset? - Find every thesis that uses this method. - What does the ethics approval allow for interview recordings? Files it reads: Papers, Grant reports, Ethics approvals, Interview recordings. ### Non-profits and NGOs Beneficiary data and field reports need protecting in every country you work in. - Which field reports mention access problems in the north? - Find the safeguarding policy for partner staff. - What did we promise this donor in the grant agreement? Files it reads: Field reports, Grant agreements, Case notes, Field photos. ## Platform: Six capabilities. One perimeter. One modular Sovereign AI platform to ask, reason and act on private data. Start with search, add agents when you're ready, and bring your own models and tools. Every layer runs inside your own perimeter. ### 1. Multimodal AI-Native Filesystem (Data) Every file becomes something you can ask. Documents, spreadsheets, scans, images, audio, video and email are parsed beside your files and understood by meaning, not file name. ### 2. Agentic Retrieval (Agents) Answers that follow the evidence. Agents plan multi-step searches across keyword and vector indexes, follow references between files and cite the exact passage. ### 3. Agentic Multi-Sandbox Reasoning (Agents) Parallel reasoning, sealed apart. Each agent reasons and runs code in its own isolated sandbox, so parallel tasks never share memory, credentials or data. ### 4. Agent Orchestration Engine (Agents) Fleets of agents, one conductor. Plans, schedules and supervises autonomous agents, with human approvals, retries and a complete audit trail. ### 5. Cross-Domain Agentic Data Fabric (Data) Connect silos without centralising them. Spans departments, systems and classification levels with policy enforced at every hop. Every source stays put. ### 6. On-Premises Cognitive Isolation (Perimeter) The whole mind stays inside. Models, inference, memory and embeddings run entirely inside your perimeter. Bring your own model, or let fileyap route by cost, speed and privacy. ## For agents and developers: Your agents get a source of truth. Agents query the same layer through a deterministic query language built for them, so they get precise, repeatable results instead of guesses. Any AI tool, agent or app can connect over MCP or the API and get cited answers from your own data. - MCP server and REST API - Deterministic, repeatable queries - Same permissions as the user Next: agents that act on your files inside the same walls, like flagging contracts before they renew. ## Services: We deploy it with you. Every company deployment starts with a paid pilot on your own infrastructure. We work alongside your team from the first question to production, then hand over everything you need to run it yourselves. 1. **Scoping: Find the questions worth answering.** We map where your files live and which questions cost your team the most time, then agree what success looks like before anything is installed. You get: A map of where your files live; The questions that cost the most time; Agreed success criteria. 2. **Evaluation: Prove it on your own questions.** Your team's real questions become the test. We measure every answer and its sources against them, and tune until the results clear the bar you set. You get: A test set built from real questions; A score for every answer and source; Search tuned to your files. 3. **Deployment: Run it in your most sensitive environments.** We install and harden fileyap with your infrastructure team, wherever it has to run. Your compliance and data-residency rules decide the architecture. You get: A hardened install where you need it; A security and data-residency review; Access rights matched to your directory. 4. **Integration: Connect it to what your teams already use.** We wire fileyap into your file shares, archives and access controls, and help your developers plug in their own agents and tools. Then we hand over the knowledge. You get: Your systems connected; Your agents and tools plugged in; Runbooks and training for your team. Our engineers work alongside your team at every step. ## FAQ ### What is fileyap? fileyap is private, AI-powered search for your files. Ask a question in plain language and get an answer drawn from your own documents, spreadsheets, scans, recordings and email, with every source passage highlighted. ### Who is it for? Anyone whose answers live in files they can't send to a public AI: individuals with private files, teams that share knowledge, and regulated, document-heavy companies in industries like finance, legal, healthcare and government. ### Is fileyap available now? fileyap is in private beta. The first waitlist cohort onboards in November 2026, and we plan a full public launch in January 2027. Join the waitlist to get an invite. ### How is it different from ChatGPT, Copilot or enterprise search? Cloud AI tools run on the vendor's servers, and shared-drive search finds file names, not answers. fileyap runs inside your own infrastructure or on the secure fileyap cloud, reads every format, and shows the exact passage behind every answer. If the answer isn't in your files, it says so. ### Can I use fileyap on my own? Yes. Individuals can run fileyap on their own device or in the fileyap cloud, and start free. ### Do my files leave my device? Not if you run fileyap on your own device: your files, and the AI that reads them, stay there. If you choose the fileyap premium cloud instead, your files are kept in a dedicated, encrypted tenant. ### What can fileyap read? Documents and PDFs, spreadsheets, scans and images, audio, video and email. It also connects to the platforms your teams already use, such as Google Drive, SharePoint, Salesforce, Slack and SAP. ### How does a company get started? With a 30-minute briefing, then a paid pilot on your own infrastructure. Your team's real questions become the test, and we measure every answer and its sources against the bar you set before moving to production. ### How is fileyap priced? Companies start with a paid pilot, then an annual licence. Teams pay per seat, and individuals can start free. Request a briefing for a quote. ### Where does fileyap run? On your own servers, in a sovereign cloud region, fully air-gapped, or in the fileyap premium cloud. Models, inference, memory and embeddings all run inside that environment. ### Does it respect our access rights? Yes. Answers follow the permissions you already manage, so nobody sees a file they couldn't open before. Every question, answer and source is logged for audit. ### Can we use our own AI model? Yes. Bring your own model, or let fileyap route between models by cost, speed and privacy, including models that run entirely on your own servers. ### Can our agents and AI tools use fileyap? Yes. Any AI tool, agent or app can connect over MCP or the REST API and get cited answers from your own data, with the same permissions as the person using it. ### Is there a version of this site for LLMs and bots? Yes. Choose Machine in the theme menu to read the whole site as plain text, or point your tools at /llms.txt and /llms-full.txt. ## Get started: Your data is sovereign. Your AI should be too. Join the private beta waitlist for early access, or ask for a briefing and see it running inside a perimeter like yours. - Join the waitlist: https://fileyap.com/#waitlist - Request a briefing: https://fileyap.com/#briefing