> For the complete documentation index, see [llms.txt](https://40factory.gitbook.io/wilson/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://40factory.gitbook.io/wilson/3.11/best-practices/strategies-and-tips-for-effective-implementation.md). # Strategies and Tips for Effective Implementation ## Knowledge Base *** ### Structuring the Knowledge Base #### Structure the Knowledge Base Create folders and assign clear names based on teams, machine types, topics, etc. The goals are: 1. Facilitate knowledge base management for document uploaders. 2. Make the knowledge base easily filterable for individual bots. 3. Ensure that folder, file names, and descriptions are "self-explanatory" to simplify automatic document filtering during search (auto document selector). #### Tag Management Tags allow finding documents located in different paths within the knowledge base. They are useful for filtering documents across folders. Use short, meaningful tags. Tags lose effectiveness if their names are identical to folder names. It's advisable to choose keywords that clearly describe the document's characteristics, avoiding abbreviations and acronyms. Alternatively, tags can selectively enable documents for different bots. For example: Folder ModelX: (first module) * doc ModelX-1 (tag "Bot1") * doc ModelX-2 (tag "Bot2") * doc ModelX-3 (tag "Bot3") Folder ModelX-sub1: * subdoc ModelX-1 (tag "Bot1") * subdoc ModelX-2 (tag "Bot1") * subdoc ModelX-3 (tag "Bot1") Folder ModelX-sub2: * subdoc ModelX-2 (tag "Bot2") Folder ModelX-sub3: * subdoc ModelX-3 (tag "Bot3") *** ### Document Upload #### **Selecting Documents** Avoid uploading the same file multiple times, ensure that files do not contain conflicting information, and do not upload incorrectly oriented documents (rotated). #### **Choosing the Parser** * If the document is a PDF with tables and images, use a dedicated parser. * If the document is a TXT, Excel, Word, or CSV file, use the basic parser. #### **Prevent logos or icons in manuals from being detected** Reduce the extracted image size ratio (Minimum image to page size ratio). If this excludes relevant images, logos should be manually removed from the image extraction section. Set a percentage size lower than useful images in the documents. #### **Document Description** * Identical descriptions for different documents make the auto document selector ineffective, especially when there are many documents. It’s better to use automatic description generation. * Including a series of examples in document descriptions or prompts that are not fully exhaustive can limit the scope of use for the document or its responses.\ For example:\ Instead of: “… useful for alarms like drive error, sensor fault, safety error.”\ Better: “… useful for alarms like drive error, sensor fault, safety error, and many more.” #### **Managing Excel Files with Multiple Sheets** Not recommended; instead, upload single files containing only one sheet at a time. #### **Verify Files** Ensure that files are not password-protected. *** ### **Document Processing** #### **Managing Electrical Schematics** It is advisable to wait for the release of specific functionalities. #### **Creating Descriptions of Extracted Images** Descriptions are generated from caption information. In the absence of captions, they are contextualized based on the page's text. It is important to review the generated texts. #### **Uploading a PDF but unable to view text within images** It is recommended to save the file in PDF/X format (). We recommend scanned or PDF/X documents to ensure optimal processing in Wilson. *** ## Virtual Assistants *** ### **Functionality** #### **How to create one or more bots?** Create a bot whenever a specific scope is defined for response management. For example, it might be useful to have a bot for each serial number or machine model, or a bot for the sales department with knowledge of all machine specifications. It is essential to define which documents the bot should consult for answers and which it should ignore. #### **How to enable the "Data Scientist" feature?** Enable the feature on the bot, then select the option from the Homepage and specify the document to analyze. Ensure that questions contain terms also present within the table. #### **Which formats are applicable for the "Data Scientist" feature?** Only CSV files. Currently, it cannot be applied to Excel or PDF files. #### **What to include in the prompt?** The prompt should be in English and clearly define the purpose and scope of the bot. #### **How to associate a bot with a folder or tags?** Select Settings > Bot Management > Edit Filters > select folders or tags to associate with the bot. *** ### Chat #### **How are responses generated?** The response generation process includes these steps: * Question analysis and rephrasing, influenced by the bot's prompt and the last five interactions of the conversation. * Document search based on the rephrased question (if the auto document selector option is enabled in the bot configuration); specific reference documents can be used for the search. * Search within documents by similarity or keywords and extraction. * Adding reference documents. * Generating the response. * Evaluating sources for relevance (with a score). #### **What information does the Auto Document Selector algorithm use to select documents?** File name, folder path, description, and tags. #### **How to interact with the bot?** It is recommended to ask complete and detailed questions; the bot does not function like a search engine. #### **Can I change topics during the conversation?** Yes, but it is preferable to make it explicit. To repeatedly ask the same question for response validation, start a new conversation. #### **How to check the documents used by Wilson?** Documents used in the response are listed in the sources. The score reflects document relevance to the response, not the question. #### **Can I teach Wilson new information directly from the chat?** You can add content to Wilson through two tools: 1. **Q\&A** - to create questions and answers for inclusion in the knowledge base. 2. **Feedback** - now related to bots; for information tied to a specific document or topic, it is preferable to use Q\&A. #### **What happens if I correct document information and then delete the document?** Applied corrections are not lost. Feedback is treated as separate “documents” and must be manually removed. #### **How can users provide feedback to administrators on the knowledge base?** Users can indicate the response score. Open feedback will be implemented in the future. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://40factory.gitbook.io/wilson/3.11/best-practices/strategies-and-tips-for-effective-implementation.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.