AI Search Triage - Prompt Mode: Research

Content-Type: application/json
Accept: text/event-stream
X-Auth-Token: <your-token>

json
{
  "question": "What are the key findings from our Q4 sales analysis?",
  "time_from": "2024-10-01",
  "time_to": "2024-12-31",
  "selected_file_types": "pdf,docx",
  "selected_collections": "123,456"
}

event: <type>
data: <payload>

event: query
data: sales figures Q4 2024

event: chunk
data: Based on the analysis of Q4 sales data

event: final
data: {"summary":"Based on the analysis...","answers":[{"answerId":"4799971_5"}],"queries":["sales figures Q4 2024"]}

{
"summary": "Complete summary...",
"answers": [
{
"answerId": "4799971_5",
"title": "Document Title",
"content": "Relevant content..."
}
],
"queries": ["sales figures Q4 2024"]
}

query    How does Lucy index content from sources such as SharePoint and Google Drive?    
17:51:52.911
query    What is the process Lucy uses to generate answers from indexed content?    
17:51:52.911
query    Does Lucy use natural language processing for answer generation?    
17:51:55.211
query    How often does Lucy update or re-index content from source systems?    
17:51:55.463
query    What technologies or algorithms power Lucy's content indexing and answer generation?    
17:51:55.734
chunk    Lucy    
17:51:59.776
chunk    indexes    
17:52:00.004
chunk    content    
17:52:00.004
chunk    and    
17:52:00.004
chunk    generates    
17:52:00.004
chunk    answers    
17:52:00.004
chunk    through    
17:52:00.004
chunk    a    
17:52:00.004
chunk    sophisticated    
17:52:00.226
chunk    AI    
17:52:00.226
chunk    -powered    
17:52:00.226
chunk    knowledge    
17:52:00.226
chunk    management    
17:52:00.226
    chunk    multiple    
17:52:01.033
chunk    common    
17:52:01.033
chunk    file    
17:52:01.033
chunk    systems    
17:52:01.033
chunk    such    
17:52:01.033
chunk    as    
17:52:01.266
chunk    Box    
17:52:01.266
chunk    ,    
17:52:01.266
chunk    Dropbox    
17:52:01.266
chunk    ,    
17:52:01.266
chunk    and    
17:52:01.266
chunk    Share    
17:52:01.266
chunk    Point    
17:52:01.266
...
....
.....
.......
{"summary":"Lucy indexes content and generates answers through a sophisticated AI-powered knowledge management system designed to handle vast amounts of unstructured enterprise data, including documents, audio, and video files from multiple common file systems such as Box, Dropbox, and SharePoint. Here's how Lucy works:\n\n1. **Content Ingestion and Indexing**  \n   Lucy ingests data from various internal sources in multiple formats (PDFs, PPTs, Word documents, videos, audio) without requiring users to tag or preprocess the data extensively. It creates an index of metadata and content that enables rapid retrieval without duplicating all original data, thus optimizing storage and access efficiency. This indexing includes enriching content with metadata, tags, and taxonomies to facilitate better search and retrieval [133173331_1], [118034592_458].\n\n2. **Natural Language Processing (NLP) for Understanding and Answering**  \n   Lucy functions as an answer engine that allows users to ask natural language questions. It uses advanced NLP techniques, including natural language understanding (NLU) and natural language generation (NLG), to interpret the intent of questions, identify relevant entities (people, places, dates), and generate precise, context-aware answers. It can extract answers from within large documents or multimedia files and present them alongside source references or export them for further use [133173331_1], [133173569_5].\n\n3. **Delivering Specific Answers, Not Just Documents**  \n   Unlike traditional search systems that return lists of documents based on keyword matching, Lucy returns specific answers or content snippets. It organizes information into smaller units (e.g., question-answer pairs, lists) rather than entire documents, allowing for efficient retrieval and helping users quickly find the exact information they need without sifting through long texts [117193008_11], [133173331_1].\n\n4. **Continuous Learning and Smart Insights**  \n   Lucy improves over time by learning from interactions and user behavior, thereby enhancing relevance and accuracy of responses. It also provides additional insights into the content such as highlighting topics, keywords, and key people appearing within video assets or documents, and facilitates exploration of timelines and transcripts in multimedia data [133173331_1].\n\n5. **Enterprise Integration and Scalability**  \n   The platform supports integration across various systems used by large organizations, enabling knowledge sharing across departments and geographies. This reduces information silos and prevents knowledge loss due to employee turnover or scattered data repositories. Lucy can scale to ingest tens of thousands of documents rapidly and provide access to the knowledge in a timely manner [133173331_1], [118034592_458].\n\nIn summary, Lucy’s ability to index content and generate answers relies on deep integration with enterprise data sources, enriched metadata indexing, natural language processing, and an AI-driven answer engine optimized for immediate relevance and ease of use within large organizations’ knowledge ecosystems [133173331_1], [133173569_5].","answers":[{"AnswerID":"133173331_1","MD5":"","Title":"Lucy AI: Transforming Enterprise Knowledge Management","Text":"<img src=\"/Documents/l2-0043a/133173331_133173331_1.jpg\" pdfurl=\"/Documents/l2-0043a/133173331_133173331_1.jpg\" pdfSize=\"0.5321512222290039\"/>","Confidence":0.0,"ExpertRating":0.0,"TrainingCount":0,"Company":"0043a","Source":"[ObjectStoreURL]/l2-0043a/DeepAnalysisReportonLucy.pdf?bsaccount=6ee4f7ce-ca46-43a5-86ae-2096efb5b749&bsid=eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhTldNSnE3cEQwRUthNC1NelRSTzI0ZHZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTVVWFBJNVFRQkZYUVJSSVJMWkRZNEpOM0JFRlRYNkdOIn0=","Cite":"Company Research","Answer_Concepts":"AI software,knowledge management,marketing technology; NLP,NLU,machine learning; venture capital,enterprise solutions","Answer_Taxonomy":"none","Answer_Keywords":"none","Filter3":"1of4","author_name":"","FileName":"Deep Analysis Report on Lucy.pdf","currentPageNumber":1,"totalpageCount":4,"Description":"","Language":"","Topic":"","assetDetailsUrl":"","section":"","V2Passage":"","isGPS":false,"isThirdPartySource":false,"shouldShowSourceNameInChat":false,"answer_locations":"Minnesota","answer_brands":"Lucy AI; Equals 3; IBM; PepsiCo","answer_persons":"Lucinda Watson; Thomas Watson","combinedData":"Lucy AI: Transforming Enterprise Knowledge Management","userSelectedDate":null,"TagData":null,"Taxonomies":[],"CustomTaxonomies":[],"Concepts":[],"Entities":[],"DiscoveryConcepts":[],"DiscoveryTaxonomies":[],"DiscoveryKeywords":[],"DiscoveryEntities":[],"CompanyandSource":[],"documentDate":"2022-04-07T11:17:50Z","createdDate":"2024-10-15T10:24:30Z","updatedDate":"2022-04-07T11:17:53Z","categories":"","Passage":"Lucy AI: Transforming Enterprise Knowledge Management","meta":{"site_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEifQ==","site_name":"Secured Data Folders","site_url":"https://equals3ai.sharepoint.com/sites/SecuredDataFolders","parent_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhTldNSnE3cEQwRUthNC1NelRSTzI0ZHZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTVVWFBJNVFQVldWU0taTkhUSkZJM1lWVDRLSkE0VURZIn0=","parent_name":"Research","parent_url":"https://equals3ai.sharepoint.com/sites/SecuredDataFolders/Shared%20Documents/Lucy%20Sales/Research","SourceFileName":"Deep Analysis Report on Lucy.pdf","modifier":"SharePoint App","file_created_date":"2022-04-07T11:17:50Z","last_saved_date":"2022-04-07T11:17:53Z","Language":"en","project_id":133145233,"file_id":133173331,"titleEnrich":"success","summary":"success","enrichv2":"success","Prompt":1,"vector_status":"success"},"relevancyScore":null,"answerDate":"Created Date","upVote":0,"downVote":0,"embeddings":null,"weightageByDate":0.0,"collectionDetails":null,"isVerified":false,"sourceMeta":"{\"url_regex\":\"\",\"source_replacement\":\"\"}"},{"AnswerID":"117193008_11","MD5":"","Title":"White Paper ... ","Text":"<img src=\"/Documents/l2-003tu/20220317_km_wp_renaissance_en1.pdf_117193008_11.jpg\" pdfSize=\"1.3876218795776367\"/>","Confidence":0.0,"ExpertRating":0.0,"TrainingCount":0,"Company":"003tu","Source":"[ObjectStoreURL]/l2-003tu/20220317_km_wp_renaissance_en1.pdf?bsaccount=1f441773-53bb-478e-9969-f71d3eaf6791&bsid=eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhY2pma215dFhTRVdvdmZZS0ttODVFOXZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTY1Q1oyQzdYUVNSR0VKMzVFNUYzWUhZQU1CN1ZKS1JLIn0=","Cite":"SharePoint - Misc","Answer_Concepts":"none","Answer_Taxonomy":"none","Answer_Keywords":"none","Filter3":"11of12","author_name":"","FileName":"2022-03-17_km_wp_renaissance_en (1).pdf","currentPageNumber":11,"totalpageCount":12,"Description":"","Language":"","Topic":"","assetDetailsUrl":"","section":"","V2Passage":"","isGPS":false,"isThirdPartySource":false,"shouldShowSourceNameInChat":false,"answer_locations":"","answer_brands":"","answer_persons":"","combinedData":"White Paper ...  White Paper \nWhy Knowledge Management is Undergoing a Renaissance\n11\nThis\n \nleaves vast troves of potentially valuable and also \nworthless information untapped. Knowledge mining \nthrough unstructured enterprise content, especially \nAI-driven will be a key factor in understanding, analyz\n-\ning, and making new knowledge available to compa\n-\nnies. While indexing has been sufficient until recently, \nthe pure volume of data created daily ensures it will \nnever be enough, and too much falls through the \ncracks. Mining knowledge includes three key steps:\n \n \nIngesting  data\n in structured and \n \nunstructured formats from a variety \nof\n  \ndifferent systems\n \nEnriching  Enhancing the data\n by   \n \n      adding metadata, tags, and taxonomies\n \nDelivering data\n via more natural search \n \n      with\n \nnatural language processing and in \n      a variety of different internal channels\n \n6. From documents to snippets\nThe document has been the basic organizing principle \nof information since the dawn of computing. And yet, \nhow many people enjoy wading through long walls of \ntext, particularly if only looking for a specific answer or \ntopic Next-generation KM systems will move away \nfrom the classic MS Word style text editors to block-\nbased systems that organize and structure information \ninto smaller units such as question and answer pairs or \nlists that can be indexed, accessed, and delivered \nindividually, without the baggage of an entire docu\n-\nment.\n \n7. Knowledge Bots\nBots, whether customer or employee-facing, are one of \nthe easiest and fastest channels to distribute knowl\n-\nedge and solve issues. Standalone bots, which require \ntheir own maintenance and management will wither \nwhile those powered directly by KM will grow. \n8. Relevance will be more \n \nimportant than ever\nRelevance has become critical as companies suffer \nfrom information overload. In a world of fewer docu\n-\nments, the time, energy, and efficiency cost of search\n-\ning and longer times to find the correct information \nhave been acceptable. Today, information is multiplying \nfaster than rabbits. Two key practices to increase \nrelevance, shorten the time to find results, and ensur\n-\ning they are useful are implicit and collaborative \nprofiling, fancy words for concepts many of us already \nunderstand from consumer products like Google. \n01\n \n\n \nImplicit profiling  KM tools must be capable of \nlearning what we search for, how we search, and what \nwe find relevant. This is what Google Search does for \nthose using it while logged into their account. Over \ntime, it delivers increasingly tailored results based on \nuser behavior, making searching, and finding faster and \nmore accurate. \n \n02\n \n\n \nCollaborative profiling  Collaborative profiling is \nsimilar, except rather than only on an individual basis, it \ntakes into account groups of users with similar profiles, \npositions, and needs. We already know this from \nAmazon and its recommendation engine Users who \nbought this also bought for example. \n -- This leaves vast troves of potentially valuable and also\nworthless information untapped. Knowledge mining\nthrough unstructured enterprise content, especially\nAl-driven will be a key factor in understanding, analyz- Relevance has become critical as companies suffer\ning, and making new knowledge available to compa- from information overload. In a world of fewer docu-\nnies. While indexing has been sufficient until recently, ments, the time, energy, and efficiency cost of search-\nthe pure volume of data created daily ensures it will ing and longer times to find the correct information\nnever be enough, and too much falls through the have been acceptable. Today, information is multiplying\ncracks. Mining knowledge includes three key steps: faster than rabbits. Two key practices to increase\n: relevance, shorten the time to find results, and ensur-\n Ingesting data in structured and ing they are useful are implicit and collaborative\nunstructured formats from a variety profiling, fancy words for concepts many of us already\nof different systems understand from consumer products like Google.\n Enriching  Enhancing the data by\n. . 01  Implicit profiling - KM tools must be capable of\nadding metadata, tags, and taxonomies \nlearning what we search for, how we search, and what\n Delivering data via more natural search we find relevant. This is what Google Search does for\nwith natural language processing and in those using it while logged into their account. Over\na variety of different internal channels time, it delivers increasingly tailored results based on\nuser behavior, making searching, and finding faster and\nmore accurate.\n02  Collaborative profiling - Collaborative profiling is\nThe document has been the basic organizing principle similar, except rather than only on an individual basis, it\nof information since the dawn of computing. And yet, takes into account groups of users with similar profiles,\nhow many people enjoy wading through long walls of positions, and needs. We already know this from\ntext, particularly if only looking for a specific answer or Amazon and its recommendation engine Users who\ntopic Next-generation KM systems will move away bought this also bought... for example.\nfrom the classic MS Word style text editors to block-\nbased systems that organize and structure information\ninto smaller units such as question and answer pairs or\nlists that can be indexed, accessed, and delivered\nindividually, without the baggage of an entire docu-\nment.\nBots, whether customer or employee-facing, are one of\nthe easiest and fastest channels to distribute knowl-\nedge and solve issues. Standalone bots, which require\ntheir own maintenance and management will wither\nwhile those powered directly by KM will grow.\nWhite Paper  Why Knowledge Management is Undergoing a Renaissance 11  ","userSelectedDate":null,"TagData":null,"Taxonomies":[],"CustomTaxonomies":[],"Concepts":[],"Entities":[],"DiscoveryConcepts":[],"DiscoveryTaxonomies":[],"DiscoveryKeywords":[],"DiscoveryEntities":[],"CompanyandSource":[],"documentDate":"2022-03-17T04:16:00Z","createdDate":"2023-10-13T17:39:27Z","updatedDate":"2022-03-17T04:16:05Z","categories":"","Passage":"White Paper ... ","meta":{"site_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEifQ==","site_name":"Lucy Demo content","site_url":"https://equals3ai.sharepoint.com/sites/LucyDemo","parent_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhY2pma215dFhTRVdvdmZZS0ttODVFOXZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTY1Q1oyQzZRNUVXN0JCVzRIWkUzWk9HTklZSFlIRjZDIn0=","parent_name":"Misc_Other","parent_url":"https://equals3ai.sharepoint.com/sites/LucyDemo/Shared%20Documents/Demo%20Files/Misc_Other","SourceFileName":"2022-03-17_km_wp_renaissance_en (1).pdf","modifier":"SharePoint App","file_created_date":"2022-03-17T04:16:00Z","last_saved_date":"2022-03-17T04:16:05Z","Language":"en","project_id":117190841,"file_id":117193008,"enrich":"error"},"relevancyScore":null,"answerDate":"Created Date","upVote":0,"downVote":0,"embeddings":null,"weightageByDate":0.0,"collectionDetails":null,"isVerified":false,"sourceMeta":"{}"},{"AnswerID":"133173569_5","MD5":"","Title":"JulyAugust 2022... ","Text":"<img src=\"/Documents/l2-0043a/133173569_133173569_5.jpg\" pdfurl=\"/Documents/l2-0043a/133173569_133173569_5.jpg\" pdfSize=\"0.929962158203125\"/>","Confidence":0.0,"ExpertRating":0.0,"TrainingCount":0,"Company":"0043a","Source":"[ObjectStoreURL]/l2-0043a/TheNewWorldofIntelligentSearchandContentAnalytics.pdf?bsaccount=6ee4f7ce-ca46-43a5-86ae-2096efb5b749&bsid=eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhTldNSnE3cEQwRUthNC1NelRSTzI0ZHZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTVVWFBJNVhLT1M1MkNOM1JCNUhMVTZTSEpNM1NBRFBEIn0=","Cite":"Company Research","Answer_Concepts":"none","Answer_Taxonomy":"none","Answer_Keywords":"none","Filter3":"5of6","author_name":"","FileName":"The-New-World-of-Intelligent-Search-and-Content-Analytics.pdf","currentPageNumber":5,"totalpageCount":6,"Description":"","Language":"","Topic":"","assetDetailsUrl":"","section":"","V2Passage":"","isGPS":false,"isThirdPartySource":false,"shouldShowSourceNameInChat":false,"answer_locations":"","answer_brands":"","answer_persons":"","combinedData":"JulyAugust 2022...  JulyAugust 2022 S30 K M W o r l d Sponsored Content Bid Writing: Generating Automated Responses for RFPs and RFIs By connecting your data with the right AI-driven technology, you can copy and paste a specific question from a tender and by using informa - tion from your data sources including pre - vious RFPs and RFIs, automatic and proper responses will be generated in seconds. Gener - ated answers will use all documents and infor - mation the intelligent search solution has access to. This includes blogs from your website, FAQs, whitepapers, corporate documentation, and more. We have natural language processing NLP, the branch of computer science fo - cused on giving machines the ability to under - stand natural human language to thank for this. Natural language understanding NLU can conceptualize the intent of the question and recognize the entities such as people, locations, times, and dates within the ver - biage and in multiple languages. This is known as entity recognition. Entity recognition enables the automat - ic extraction of the mentioned entities. In other words, it helps companies cut through the fluff. Business documents and RFPs, in general, are often filled with a lot of details that are not always relevant to the task at hand. Entity recognition and NLU can cut out what is not essential to a particular use, identify synonyms and words with similar meaning, and provide insights on given topics within your knowledge base. With Natural Language Generation NLG, the solution can form responses in complete sentences using proper sentiment and grammar. The task gets automated rath - er than thinking long and hard or asking a different department for resources on how to phrase an answer. Now, the sales team simply has to review the generated re - sponse and copy and paste it into the bid submission form. If that is not enough, it can even be tak - en a step further to cut out the repetitive copying and pasting. A bid manager can also upload full excel sheets and get every response filled out in an instance. Using the approach of uploading the entire excel file, all the sales team needs to do is run through the document and ensure accuracy, rather than writing new answers from scratch or copying the generated answers one by one. The machine learning techniques are con - sistent with both of these approaches and the use is fully dependent on how automat - ed you want or need the process to be. Tracking the Status of your Ongoing and Upcoming Bids Sales teams can be swamped following up with different leads, investigating potential client fits, and in many cases, pitching or demoing the actual product. In some organizations, bid man - agement is a full-time job. In others, it is a part of the previously listed tasks. On top of that, there are likely multiple opportunities and tenders running parallel to each other with overlapping deadlines and time-con - suming tasks, tracking and organizing where you stand in the process for all your bids is critical. Otherwise, the risk of los - ing a big piece of business or organization- defining contract is a real possibility. When competing for multiple tenders, having a clear 360-degree view of due dates and bid status helps keep sales teams organized and deadline-oriented. Artificial intelligence can extract the dates from dif - ferent documents and show your progress on a clear graphical timeline for easy use and reference. Whats on the Horizon A Detailed Overview Search has reached a fascinating time with the help of AI and ML technology. In - telligent search constantly makes individu - al departments more innovative and more productive based on their use case in a wide range of functional areas. The use case for sales and bid management is a very prom - ising example on what is on the horizon for search. Any area where you can auto - mate processes to make life easier for your workforce is important and we have seen so much success with this in recent years. Sales, however, is a department that keeps a business running with continued cash flow and customers for your other departments to work for. With the help of natural language processing and natural language understanding, natural language generation, entity recognition, and many other forms of business intelligence, it is giving workers the ability to receive in - sights in real human language while under - standing the context of specific questions related to bids and tenders. Freeing up time and automating certain components of the sales process can lead to more wins and more deals closing which ultimately helps every aspect of your business and operations. Companies are always looking for ways to stay ahead of their competition and us - ing intelligent search for bid management and sales could be your answer to doing just that. Consistent innovation in the right areas and defining the use case that makes the most sense to your organization will play a pivotal role to the future of your company. Contact: Mindbreeze Corp. 311 West Monroe Street Chicago, Illinois, 60606 Phone: 1.312.300.6745 Web: www.mindbreeze.com As much as companies like to think that their sales team knows every corner and crevice of your organization, the existence of data silos makes that unrealistic unless the proper AI solution is being used. Permitting sales teams to search across de partments prior to going down the question-answering process will help them identify showstoppers without the potential long wait times of asking a colleague in a different area of the business. Search has reached a fascinating time with the help of AI and ML technology. Intelligent search constantly makes individual departments more innovative and more productive based on their use case in a wide range of functional areas. The use case for sales and bid management is a very promising example on what is on the horizon for search. -- Bid Writing: Generating Automated : Responses for RFPs and RFIsBy : connecting your data with the right AF-driven : A much as companies like to think that their sales technology, you can copy and paste a specific : . question from a tender and by using informa- team knows every corner and crevice of your tion from your data sources including pre- : . . . . vious RFPs and RFls, automatic and proper Fganization, the existence of data silos makes that responses will be generated in seconds. Gener- : a . . ated answers will use all documents and infor- : Utealistic unless the proper Al solution is being used. mation the intelligent search solution has access: Here to. This includes blogs from your website, : Permitting sales teams to search across departments FAQs, whitepapers, corporate documentation, : - - . and more. : prior to going down the question-answering process We have natural language processing : . 2 . . NLP, the branch of computer science fo. will help them identify showstoppers without cused on giving machines the ability to under- . ne ge . stand natural human language to thank for this, potential long wait times of asking a colleague Natural language understanding NLU : H H H can conceptualize the intent of the question : ma differ ent ar ea of the b USINESS. and recognize the entities such as people, : locations, times, and dates within the ver- : biage and in multiple languages. This is : than writing new answers from scratch or : Whats on the Horizon known as entity recognition. : copying the generated answers one by one. : A Detailed Overview Entity recognition enables the automat- : The machine learning techniques are con- : Search has reached a fascinating time ic extraction of the mentioned entities. In : sistent with both of these approaches and : with the help of Al and ML technology In- other words, it helps companies cut through : the use is fully dependent on how automat. : telligent search constantly makes individu- the fluff. Business documents and RFPs, in : d you want or need the process to be. al departments more innovative and more general, are often filled with a lot of details : . . 3 productive based on their use case in a wide that are not always relevant to the task at: Tracking the Status of your Ongoing : range of functional areas. The use case for hand. Entity recognition and NLU can cut : and Upcoming BidsSales teams can be : sales and bid HATEEEMERE isa very rit out what is not essential to a particular use, : swamped following up with different leads, : ising example on what is on the horizon identify synonyms and words with similar : investigating potential client fits, and in: pooh Any area where you can auto- meaning, and provide insights on given Many cases, pitching or demoing the actual : 4, processes to make life easier for your topics within your knowledge base. PROMUGE, Ti Sine: Organizations, bid. Hiei - workforce important and we have seen so With Natural Language Generation : agement is a full-time job. In others, it isa : much success with dis in vevent years NLG, the solution can form responses in : Part of the previously listed tasks. On top of : Sales. however: Gk department that complete sentences using proper sentiment : that, there are likely multiple opportunities : keeps a business running ith eontinned and grammar. The task gets automated rath- : and tenders running parallel to each other : wasty flow and customers fer your viter er than thinking long and hard-or asking a With overlapping deadlines and time-con- 4, acments to work for. With the help of different department for resources on how : Suming tasks, tracking and organizing natural language processing and Aataral : language understanding, natural language a . . . . : generation, entity recognition, and many Search has reached a fascinating time with the : other forms of business intelligence, it is . : giving workers the ability to receive in- help of Al and ML technology. Intelligent search : sights in real human language while under- . on : standing the context of specific questions constantly makes individual departments more : related to bids and tenders. Freeing up . x 2 : time and automating certain components innovative and more productive based on their use: of the sales process can lead to more wins . . . : and more deals closing which ultimately case in a wide range of functional areas. The use CaS : helps every aspect of your business and . os : operations. for sales and bid management is a very promising 2 companies are always looking for ways example on what is on the horizon for search. 3. Siicent sourch for bid management : and sales could be your answer to doing : . : just that. Consistent innovation in the right to phrase an answer. Now, the sales team : Where you stand in the process for all your : a ea. and defining the use case that makes simply has to review the generated re- bids is critical. Otherwise, the risk of los- the most sense to your organization will sponse and copy and paste it into the bid i ing a big piece of business or organization- : play a pivotal role to the future of your submission form. : defining contract is a real possibility. : compeny, ll If that is not enough, it can even be tak- : When competing for multiple tenders, : : en a step further to cut out the repetitive having a clear 360-degree view of due : copying and pasting. A bid manager can z dates and bid status helps keep sales teams : oS also upload full excel sheets and get every : Organized and deadline-oriented. Artificial : ie se response filled out in an instance. Using the : intelligence can extract the dates from dif- : 311 West meres Street approach of uploading the entire excel file, : ferent documents and show your progress : Chicago, Illinois, 60606 all the sales team needs to do is run through : on a clear graphical timeline for easy use : phone: 1.312.300.6745 the document and ensure accuracy, rather : and reference. : Web: www.mindbreeze.com 30 KMWord JulyAugust 2022 Sponsored Content ","userSelectedDate":null,"TagData":null,"Taxonomies":[],"CustomTaxonomies":[],"Concepts":[],"Entities":[],"DiscoveryConcepts":[],"DiscoveryTaxonomies":[],"DiscoveryKeywords":[],"DiscoveryEntities":[],"CompanyandSource":[],"documentDate":"2022-07-22T03:41:23Z","createdDate":"2024-10-15T10:52:09Z","updatedDate":"2022-07-22T03:41:26Z","categories":"","Passage":"JulyAugust 2022... ","meta":{"site_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEifQ==","site_name":"Secured Data Folders","site_url":"https://equals3ai.sharepoint.com/sites/SecuredDataFolders","parent_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLGFiMDk2MzM1LTQzYmEtNDJkMC05YWUzLWUzMzM0ZDEzYjZlMSxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhTldNSnE3cEQwRUthNC1NelRSTzI0ZHZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTVVWFBJNVFQVldWU0taTkhUSkZJM1lWVDRLSkE0VURZIn0=","parent_name":"Research","parent_url":"https://equals3ai.sharepoint.com/sites/SecuredDataFolders/Shared%20Documents/Lucy%20Sales/Research","SourceFileName":"The-New-World-of-Intelligent-Search-and-Content-Analytics.pdf","modifier":"SharePoint App","file_created_date":"2022-07-22T03:41:23Z","last_saved_date":"2022-07-22T03:41:26Z","Language":"en","project_id":133145233,"file_id":133173569,"enrichv2":"failed","Prompt":1,"vector_status":"success"},"relevancyScore":null,"answerDate":"Created Date","upVote":0,"downVote":0,"embeddings":null,"weightageByDate":0.0,"collectionDetails":null,"isVerified":false,"sourceMeta":"{\"url_regex\":\"\",\"source_replacement\":\"\"}"},{"AnswerID":"118034592_458","MD5":"","Title":"Lucys just creating her adic, has this index of metadata that allows her to go get the document,... ","Text":"<lucy-player video=\"1ba0609387\" time=\"3050\" auGeneratedFrom=\"transcript\"></lucy-player>","Confidence":0.0,"ExpertRating":0.0,"TrainingCount":0,"Company":"003ux","Source":"[ObjectStoreURL]/l2-003ux/PepsiCoHRWebinar.mp4?bsaccount=1f441773-53bb-478e-9969-f71d3eaf6791&bsid=eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhY2pma215dFhTRVdvdmZZS0ttODVFOXZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTY1Q1oyQzc0NERBUldLTzNYWkJaTVBGVldVVVg2QzZPIn0=","Cite":"HR Docs and Policy","Answer_Concepts":"none","Answer_Taxonomy":"none","Answer_Keywords":"none","Filter3":"458of523","author_name":"","FileName":"PepsiCoHRWebinar.mp4","currentPageNumber":458,"totalpageCount":523,"Description":"","Language":"","Topic":"","assetDetailsUrl":"","section":"","V2Passage":"","isGPS":false,"isThirdPartySource":false,"shouldShowSourceNameInChat":false,"answer_locations":"","answer_brands":"","answer_persons":"Lucy","combinedData":"Lucys just creating her adic, has this index of metadata that allows her to go get the document,...  Lucys just creating her adic, has this index of metadata that allows her to go get the document, go get that answer and and provide it as opposed to duplicating all of the data, all the storage and uploading into that system.  ","userSelectedDate":null,"TagData":null,"Taxonomies":[],"CustomTaxonomies":[],"Concepts":[],"Entities":[],"DiscoveryConcepts":[],"DiscoveryTaxonomies":[],"DiscoveryKeywords":[],"DiscoveryEntities":[],"CompanyandSource":[],"documentDate":"2022-07-21T18:45:35Z","createdDate":"2023-10-24T05:49:20Z","updatedDate":"2022-08-04T20:17:54Z","categories":"","Passage":"Lucys just creating her adic, has this index of metadata that allows her to go get the document,... ","meta":{"site_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEifQ==","site_name":"Lucy Demo content","site_url":"https://equals3ai.sharepoint.com/sites/LucyDemo","parent_id":"eyJzaXRlIjoiZXF1YWxzM2FpLnNoYXJlcG9pbnQuY29tLDliZTQzNzcyLTU3MmItNDU0OC1hOGJkLWY2MGEyYTZmMzkxMyxkMTYyZWFkYi1jNjg0LTQ1NzQtYmU2Mi1hNWFlMzExNmJkZjEiLCJkcml2ZSI6ImIhY2pma215dFhTRVdvdmZZS0ttODVFOXZxWXRHRXhuUkZ2bUtscmpFV3ZmR1JlNkxBM1l5aFJwNndaaUNaZFhhcSIsImZpbGUiOiIwMTY1Q1oyQzdIWTJCVklHQ0dGUkNZWURMR1RMWU9LRFlZIn0=","parent_name":"HR Docs","parent_url":"https://equals3ai.sharepoint.com/sites/LucyDemo/Shared%20Documents/Demo%20Files/HR%20Docs","SourceFileName":"PepsiCoHRWebinar.mp4","modifier":"SharePoint App","Language":"en","project_id":118034285,"file_id":118034592,"enrich":"categories_unavailable","video_date_update":"success"},"relevancyScore":null,"answerDate":"Created Date","upVote":0,"downVote":0,"embeddings":null,"weightageByDate":0.0,"collectionDetails":null,"isVerified":false,"sourceMeta":"{}"}],"queries":["How does Lucy index content from sources such as SharePoint and Google Drive?","What is the process Lucy uses to generate answers from indexed content?","Does Lucy use natural language processing for answer generation?","How often does Lucy update or re-index content from source systems?","What technologies or algorithms power Lucy's content indexing and answer generation?"]}

[
{
"question": "What are the key findings?",
"summary": "The key findings include...",
"timestamp": "2024-01-15T10:30:00Z",
"answerIds": "4799971_5,4799972_3"
}
]

curl -X POST "https://ask.lucy.ui/api/qna/research" 
-H "Content-Type: application/json" 
-H "Accept: text/event-stream" 
-H "X-Auth-Token: your-token" 
-d '{"question": "What are the main features of our new product?"}'

curl -X POST "https://ask.lucy.ui/api/qna/research" 
-H "Content-Type: application/json" 
-H "Accept: text/event-stream" 
-H "X-Auth-Token: your-token" 
-d '{"question": "What were the sales figures for Q4 2024?","selected_file_types":"pdf"}'

curl -X POST "https://ask.lucy.ui/api/qna/research" 
-H "Content-Type: application/json" 
-H "Accept: text/event-stream" 
-H "X-Auth-Token: your-token" 
-d '{
  "question": "how answer engine generates an answer?",
  "time_from": "2025-1",
  "time_to": "2025-11",
  "selected_file_types": "PDF,Keynote,PPT,PPTX",
  "selected_collections": "af895cb6-a536-4d75-909d-4605635b65b1,4c80a5a2-b97e-4e4f-8204-f39b2cfa1a73,56132ffa-e55d-45ac-b4fc-722504e8c8fc",
  "selected_solr_companies": "4975,1987,5000,4976,4513,775,4627,GPS-4415,4999,4697,4980,GPS-2347"
}'

{
"error": "Invalid request parameters",
"status": 400,
"timestamp": "2024-01-15T10:30:00Z"
}