{"id":7490,"date":"2026-08-19T19:10:11","date_gmt":"2026-08-19T19:10:11","guid":{"rendered":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/"},"modified":"2026-08-19T19:10:11","modified_gmt":"2026-08-19T19:10:11","slug":"faq","status":"publish","type":"wvn_doc","link":"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/reference\/faq\/","title":{"rendered":"Frequently asked questions"},"content":{"rendered":"

Frequently asked questions<\/h1>\n
\n

Confidentiality<\/h2>\n

Is my data leaving my infrastructure?<\/strong>
\nIn a fully local configuration, no: documents, index, conversations and
\nThe models stay at home. Each release results from an explicit choice \u2014
\nexternal model provider, hosted vector database, search engine
\nweb. The screen Settings \u2192 Privacy<\/code> gives the actual state.<\/p>\n

Does WivenLLM work without an internet connection?<\/strong>
\nYes, in an all-local configuration. A connection is only required for
\ndownloading a model for the first time, importing a remote source, or
\nuse an external service.<\/p>\n

Is the data used to train a model?<\/strong>
\nNot on the WivenLLM side: nothing is sent anywhere by default. If you plug in a
\nexternal supplier, these are his<\/strong> conditions that apply \u2014 to be verified,
\nProfessional offers often differ from consumer offers.<\/p>\n

Do my colleagues see my conversations?<\/strong>
\nNo. An administrator, however, can view the instance's history.
\nThis is necessary for operation, and users must be
\ninformed<\/strong>.<\/p>\n

Where is the data physically stored?<\/strong>
\nWhere the instance is installed: your workstation, your server, or the datacenter of
\nThe hosted instance. There is no central storage.<\/p>\n

\n

Costs<\/h2>\n

How much does it cost to use?<\/strong>
\nIn an all-local configuration, there are no usage costs: the cost is that of the hardware and
\nof its operation. With an external supplier, the billing is
\nproportional to the volume exchanged.<\/p>\n

What causes the bill to increase with an external supplier?<\/strong>
\nPinned documents (injected with each message), a long history, a large
\nnumber of fragments found, and especially the agents<\/strong> \u2014 each step is a
\nadditional call.<\/p>\n

How to reduce costs without compromising quality?<\/strong>
\nA router models<\/a> : A
\ndefault economic model, a powerful model for complex cases.
\nAlso reduce the history and only pin what is strictly necessary.<\/p>\n

\n

Quality of responses<\/h2>\n

The assistant is making up answers. What to do?<\/strong>
\nThree actions, in this order: pass the space into Query mode<\/strong> ; lower the
\ntemperature<\/strong> ; specify in the prompt system<\/strong> that all information
\nIf an item is absent from the documents, it must be reported as such.<\/p>\n

He cannot find information that is nevertheless present.<\/strong>
\nCheck that the document is added to the space<\/strong> and not just imported.
\nThen lower the similarity threshold<\/strong> and increase the number of
\nfragments<\/strong>. Finally, check that the text has been correctly extracted \u2014 a PDF
\nScanned images may be illegible.<\/p>\n

He quotes unrelated passages.<\/strong>
\nThe similarity threshold is too low, or you are requesting too many fragments.<\/p>\n

The responses deteriorated suddenly.<\/strong>
\nA change in model, embedding engine, or search parameters.
\nChanging the embeddings engine without reindexing degrades search without
\nError message.<\/p>\n

He's responding in the wrong language.<\/strong>
\nAdd the instruction to the prompt system of the space: "Always respond in
\nFrench. \u00bb The interface language does not determine the language of the responses.<\/p>\n

\n

Models<\/h2>\n

Which model should I choose?<\/strong>
\nConsult the hardware diagnostics<\/strong> : it indicates for each model in the catalog
\nIf it's suitable for your machine. If in doubt, start small and increase the size.
\n\u2192 Choosing your models<\/a><\/p>\n

Can I change the model later?<\/strong>
\nYes, without consequence: the documents remain indexed and the conversations
\nlegible. However, Changing the embeddings engine requires reindexing.<\/strong>.<\/p>\n

Local responses are slow.<\/strong>
\nStandard without a graphics card. Choose a smaller model, or switch this space
\non an external supplier.<\/p>\n

Can I mix local and cloud storage?<\/strong>
\nYes. The model is chosen per workspace, and a router can arbitrate.
\nmessage by message.<\/p>\n

\n

Documents<\/h2>\n

How many documents can I index?<\/strong>
\nThe integrated local database easily holds corpora of several tens of
\nthousands of fragments. Beyond that, a business base is justified.<\/p>\n

Can a document be used in multiple spaces?<\/strong>
\nYes: it is extracted once and indexed in each space where you add it.<\/p>\n

What happens to a deleted document?<\/strong>
\nWhen removed from a space, it loses its vectors in that space but remains in the
\nlibrary. Removed from the library, it disappears from all spaces and
\nHis extracted text has been erased.<\/p>\n

My scanned PDFs are not being read properly.<\/strong>
\nCharacter recognition depends on the quality of the scan. A tilted scan or
\nLow resolution results in faulty text. Rescan properly, or use [a specific method\/process].
\nA dedicated upstream recognition tool is the only real solution.<\/p>\n

\n

Agents Verification Status<\/h2>\n

Can an agent send an email on their own?<\/strong>
\nMessaging skills are designed for prepare<\/strong> rather
\nto ship, so that a valid human can. And an unactivated skill does not exist.
\nfor the agent.<\/p>\n

Can an agent modify my files?<\/strong>
\nOnly if the corresponding skill is activated, and in the directory
\nAllowed. Each elementary operation can be disabled separately.<\/p>\n

My agent isn't using the right tools.<\/strong>
\nThe pattern is almost always the same: operating tools is more demanding than conversing.
\nAssign the most capable model available to the agent. Check
\nthen that the expected skill is indeed activated in this space.<\/p>\n

Why is an agent so slow?<\/strong>
\nIt makes several calls to the model, one per step. A ten-step task.
\ntakes about ten times longer than a simple question.<\/p>\n

\n

Exploitation<\/h2>\n

What needs to be saved?<\/strong>
\nThe storage directory and the configuration file. Nothing else.
\nThe application is reinstalling. Try a restore.<\/strong><\/p>\n

How do I update?<\/strong>
\nSave the file, restore the new version, and restart while keeping the volume.
\nstorage. The necessary migrations apply at startup.<\/p>\n

How many simultaneous users?<\/strong>
\nIt depends on the model and hardware, not the application. A local model
\nprocesses requests one after another; an external supplier absorbs
\ncompetition much more easily.<\/p>\n

Can I migrate from a workstation to a server?<\/strong>
\nYes. The content fits in the storage directory; moving it moves it
\nthe instance.<\/p>\n

A user cannot see a workspace.<\/strong>
\nHe is not a member. Add him from the space settings, tab
\nMembers.<\/p>\n

\n

One more question?<\/h2>\n
    \n
  • The screen Help<\/strong> The application links to this documentation.<\/li>\n
  • The support address defined by your administrator is located in the footer of
    \n application page.<\/li>\n<\/ul>","protected":false},"parent":7453,"menu_order":705,"template":"","meta":[],"class_list":["post-7490","wvn_doc","type-wvn_doc","status-publish","hentry"],"yoast_head":"\nQuestions fr\u00e9quentes \u2014 Documentation WivenLLM<\/title>\n<meta name=\"description\" content=\"Les questions les plus courantes sur la confidentialit\u00e9, les co\u00fbts, la qualit\u00e9 des r\u00e9ponses, les mod\u00e8les et l'exploitation.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/reference\/faq\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Questions fr\u00e9quentes \u2014 Documentation WivenLLM\" \/>\n<meta property=\"og:description\" content=\"Les questions les plus courantes sur la confidentialit\u00e9, les co\u00fbts, la qualit\u00e9 des r\u00e9ponses, les mod\u00e8les et l'exploitation.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/reference\/faq\/\" \/>\n<meta property=\"og:site_name\" content=\"Wiven AI\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/faq\\\/\",\"url\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/faq\\\/\",\"name\":\"Questions fr\u00e9quentes \u2014 Documentation WivenLLM\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.wiven.ai\\\/#website\"},\"datePublished\":\"2026-08-19T19:10:11+00:00\",\"description\":\"Les questions les plus courantes sur la confidentialit\u00e9, les co\u00fbts, la qualit\u00e9 des r\u00e9ponses, les mod\u00e8les et l'exploitation.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/faq\\\/#breadcrumb\"},\"inLanguage\":\"en\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/faq\\\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/faq\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.wiven.ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Documentation\",\"item\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"R\u00e9f\u00e9rence\",\"item\":\"https:\\\/\\\/www.wiven.ai\\\/wiven-llm\\\/docs\\\/reference\\\/\"},{\"@type\":\"ListItem\",\"position\":4,\"name\":\"Questions fr\u00e9quentes\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.wiven.ai\\\/#website\",\"url\":\"https:\\\/\\\/www.wiven.ai\\\/\",\"name\":\"Wiven AI\",\"description\":\"\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.wiven.ai\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Frequently Asked Questions \u2014 WivenLLM Documentation","description":"The most common questions about confidentiality, costs, quality of responses, models and operation.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/reference\/faq\/","og_locale":"en_US","og_type":"article","og_title":"Questions fr\u00e9quentes \u2014 Documentation WivenLLM","og_description":"Les questions les plus courantes sur la confidentialit\u00e9, les co\u00fbts, la qualit\u00e9 des r\u00e9ponses, les mod\u00e8les et l'exploitation.","og_url":"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/reference\/faq\/","og_site_name":"Wiven AI","twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/","url":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/","name":"Frequently Asked Questions \u2014 WivenLLM Documentation","isPartOf":{"@id":"https:\/\/www.wiven.ai\/#website"},"datePublished":"2026-08-19T19:10:11+00:00","description":"The most common questions about confidentiality, costs, quality of responses, models and operation.","breadcrumb":{"@id":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/#breadcrumb"},"inLanguage":"en","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/faq\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.wiven.ai\/"},{"@type":"ListItem","position":2,"name":"Documentation","item":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/"},{"@type":"ListItem","position":3,"name":"R\u00e9f\u00e9rence","item":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/reference\/"},{"@type":"ListItem","position":4,"name":"Questions fr\u00e9quentes"}]},{"@type":"WebSite","@id":"https:\/\/www.wiven.ai\/#website","url":"https:\/\/www.wiven.ai\/","name":"Wiven AI","description":"","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wiven.ai\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en"}]}},"_links":{"self":[{"href":"https:\/\/www.wiven.ai\/en\/wp-json\/wp\/v2\/wvn_doc\/7490","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wiven.ai\/en\/wp-json\/wp\/v2\/wvn_doc"}],"about":[{"href":"https:\/\/www.wiven.ai\/en\/wp-json\/wp\/v2\/types\/wvn_doc"}],"up":[{"embeddable":true,"href":"https:\/\/www.wiven.ai\/en\/wp-json\/wp\/v2\/wvn_doc\/7453"}],"wp:attachment":[{"href":"https:\/\/www.wiven.ai\/en\/wp-json\/wp\/v2\/media?parent=7490"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}