Cracked Conversations: What to Do When Chatbots Aren’t Enough

Enterprise Search to Compliment Your Chatbot ExperienceBy: Robert Smith, Sales Engineer and John Finneran, Product Marketing

Conversational AI, or chatbot, vendors, are everywhere, deafening customers with the promise of AI-Powered solutions for their customer service needs.  According to Capterra, 158 companies currently offer chatbot software.  In Forrester’s evaluation of the emerging market for conversational AI for customer service for Q2 2019, the analyst firm identified the 14 most significant providers in the category – [24], Avaamo, Cognigy, eGain, Indenta Technologies, Interactions, IPsoft,, LogMeIn, Nuance Communications, Omilia, Saleforce and Verint.

This makes understanding what works best to improve customer experience hard.

Chatbots work best guiding users along straightforward, well-defined conversational paths.  If a customer asks new, unpredicted questions the typical chatbot gets confused. More complex questions require complementary solutions.  

Sinequa offers one such complementary solution – Enterprise Search that can work with chatbots to help customers and employees find what they need.

We have spoken with a number of companies ranging from those considering the technology, to building prototypes, to deploying chatbots in customer-facing applications.

Several of the concerns about the value produced by chatbot deployments

  • Slow Conversation speeds
  • Conversation path-sets grow larger and longer
  • Low accuracy because the chatbot was unable to answer and was unable to maintain the chat
  • High development effort with too many expert hours spent conceiving, designing, deploying, and maintaining those conversational paths.

Some Reasons Why?

Chatbots work best when guiding a well-defined type of user through a set of preconceived conversational paths.

The typical chatbot’s tooling provides a graphical interface, and some testing capabilities; conceiving, designing, deploying, and maintaining those conversational paths will be up to you.

  • When you consider how many paths a user might take, multiplied by the number of user types, it can grow to an astonishing amount of work.
  • When chatbots have a lot of this work to do, they tend to slow down compromising, the chat experience
  • Most requests for information are ‘ad-hoc’ and therefore not well-suited for a pre-planned and pre-built conversation flow.

When Do Chatbots Make Sense?

An example is a chatbot at your local bank

  • They have a limited set of offerings for users to choose from
    • E.g. checking, savings, mortgages, lines of credit
  • Those offerings have a limited number of actions
    • Checking deposit, transfer, bill pay, balance inquiry
  • The site is often for reference, not as much for execution
    • To actually open up an account type, you typically have to apply in-person

If you can’t narrow the scope to specific user-types and paths like these, then the outcome of multi-step “chats” is by definition, less predictable, leading to a higher failure rate.

This also makes it difficult for some chatbots to get a PTO (Permit to Operate), because companies have not let applications go into production that couldn’t guarantee outcomes.  This is to avoid “Rogue AI” situations, among other things.

Addressing the Challenge

Enterprise Search, like Sinequa’s, leverages natural language processing (NLP) to get users the most relevant content, without the chatbot’s requirement that the conversational path be designed, built and maintained.

Where chatbot interactions are sometimes helpful, that chatbot can connect to enterprise search; when the chatbot gets a user’s request for information, the chatbot can refine and forward the request to the underlying Sinequa search, then channel the results back to the user’s conversation.

In Short

By using chatbots and a powerful enterprise search platform together for the jobs they were designed for, you can deliver profitable and productive solutions that enhance both customer and employee experiences.

+1Share on LinkedInShare on Twitter

Sinequa Snags Three Key Industry Award Wins in September

Sinequa Industry Recognition - September 2018

We’re off to a busy September here at Sinequa! We’re excited and humbled to have received a few different awards for our Cognitive Search & Analytics Platform and company as a whole this month. Sinequa recognition has included the following awards from leading industry publications:

KMWorld Trend-Setting Products 2018

KMWorld’s 2018 list of Trend-Setting Products features not only emerging software directed toward human-like functionality but also more traditional offerings impressively refined. It encompasses AI, machine learning, cognitive computing and the Internet of Things, as well as enterprise content management, collaboration, text analytics, compliance and customer service. Read more.

DBTA’s Cool Companies in Cognitive Computing for 2018

DBTA and Big Data Quarterly presented the 2018 list of Cool Companies in Cognitive Computing to help increase understanding about the important area of information technology and how it is being leveraged in solutions and platforms to provide business advantages. Read more.

Datanami Readers’ Choice Award Winner

Sinequa won the Readers’ Choice – Best Big Data Product or Technology: Machine Learning category.

The Datanami Readers’ and Editors’ Choice Awards are determined through a nomination and voting process with input from the global big data community, as well as selections from the Datanami editors, to highlight key trends, shine a spotlight on technological breakthroughs and capture a critical cross-section of the state of the industry. Read more.

Looking forward to continuing the momentum for the rest of the year!

For more information on Sinequa’s cognitive search and analytics platform visit:

+1Share on LinkedInShare on Twitter

Inspiration and Innovation: Learning From Artists

This article appeared in Wired Innovation Insights.

The crowd at an IT event in Paris was intrigued to see an art historian as the keynote speaker opening the conference. They had come to hear their peers talk about trendy topics in Big Data, Search, Content Analytics, Natural Language processing, etc. But the first thing they were asked to do was contemplate a painting by Veronese . Then, art historian Stéphane Coviaux showed them a drawing that looked like a sketch the artist made before embarking on the big painting. Yet the “sketch” had been created by a different artist some 50 years earlier! Veronese a plagiary?!

When asked to compare the two works, the IT audience was not shy and came up with many major and minor differences. It became clear to everyone that Veronese had been inspired by the work of his predecessor. Through his changes in the conception of the image, in his use of space and color and through his own symbolic, Veronese had produced a major work of art from a comparatively minor source of inspiration.

The message to the audience: do not expect cooking recipes or “best practices”! Transpose what you see from others – the innovations they have implemented – into your own environment. And, don’t be overawed by impressive projects that you may see, view them as sketches for your own projects and “go create!”

In an emerging market or one that is radically changing, there simply are no “best practices” and no “recipes.” Take inspiration from others but use your imagination to create innovations that advance your business. A specific message for the modern times IT-audience was added: Aim at “co-creation”; find partners you trust to help you along in the creation process and accompany you in uncharted (or not completely charted) territory. In uncharted territory, your procurement services cannot take a standard contract out of a drawer and hope it fits.

The “distance” between the artistic and the business environments worked well to get the message across. The presentation of even an exemplary business project coupled with the injunction “be inspired! Do not copy, but transpose,” would certainly have provoked reactions like “fine but not applicable in my environment.” In the distant world of art, everyone could easily agree on the necessity of inspiration to create innovation, took this lesson home to apply to their enterprises.

Xavier Pornain – VP of Sales & Alliances for Sinequa.


+1Share on LinkedInShare on Twitter

Inclusion of Unstructured Data into Business Intelligence Analytics

data5Today’s large enterprises and administrations accumulate ever bigger volumes of ever more heterogeneous and volatile data. More than 90 percent of this data is unstructured, consisting of texts in many different languages. Information workers spend too much time searching for the information they need for their work and companies lose business opportunities when employees are drowning in these mounds of data.

Enter Sinequa.

Sinequa’s real-time Big Data Search & Analytics platform helps its customers meet this challenge. But what makes Sinequa different than other big data analytic platforms? See what Sinequa’s Vice President of Marketing Hans-Josef Jeanrond says about the value of Sinequa’s platform and how companies can best leverage the software.

Q: What differentiates Sinequa from other big data analysis platforms?

A: There are two basic reasons why Sinequa is different from other platforms. One being the data sources we are dealing with: We are not tackling the World Wide Web nor the complete internet of things. We focus on the enterprise world with different analytical tools.

The second reason that we stand apart from other big data platforms is that we do not choose between structured and unstructured data. We offer combined statistical and semantic analysis of big data and use the structured information to refine linguistic analysis. With Sinequa the semantic analysis of unstructured data is used to create structured data; one analytic method is used to refine the other. What sets Sinequa apart here is our capability to query an index of up-to 200 million documents in real-time with sub-second answers and our ability to deal with semantically related subjects in 19 different languages.

Q: Can you please further explain the importance of semantic analysis?

A: If you limit yourself to statistical data analysis, enterprise data often poses problems in sample size, sample error and sample bias. This is why tools from the Web don’t work well with enterprise data.

Keyword search is not a sufficient option either: The keywords you use in an information request may not occur in many of the documents that are actually relevant for your work. These keywords may not have been added as “meta data” by the people who classified and stored the documents in your document management system – they may have been looked at from a different perspective.  You need to find documents dealing with concepts that are semantically related to your request, thus needing semantic analysis to find them.


As you can see, the Sinequa platform offers functionalities that truly differentiates it from competitors. Sinequa combines deep content analytics, including Natural Language Processing with an extremely scalable IT architecture, offering users simple and secure access to the most relevant information. Stay tuned for our next post that will explain how a leading bio/pharma company implemented Sinequa to index millions of R&D documents and break down barriers between information silos, worldwide.

+1Share on LinkedInShare on Twitter

La coopération homme-machine distance les « supercalculateurs »

Vous connaissez peut-être la vielle blague sur le supercalculateur à qui on a posé la question « quel est le sens de la vie », et qui sortait (après un long temps de calcul) la réponse « 27 ». Cette blague a été racontée pour illustrer bien des aspects différents  sur des ordinateurs, sur l’interaction homme machine, la philosophie et la vie en général. Ici, je voudrais tirer votre attention sur les questions floues et réponses précises, ou vice-versa, et à l’amélioration de l’interaction entre les hommes et les machines.

Shyam Sankar a donné une conférence intéressante à ce sujet à  TED, sous le titre de « l’ascension de la coopération homme-ordinateur ». Dans son discours il explique pourquoi « résoudre de grands défis (comme arrêter des terroristes ou identifier des tendances émergentes) n’est pas tant une question de trouver le bon algorithme mais plutôt de la bonne relation symbiotique entre calcul d’ordinateurs et créativité humaine ».

Son premier exemple est bien connu mais il mérite d’être raconté encore une fois. C’est l’histoire de deux championnats d’échec de niveau mondial : En 1997, le champion du monde Gary Kasparov perd contre l’ordinateur « Deep Blue » d’IBM. En 2005, dans un championnat d’échec ouvert à tous, dans lequel des hommes peuvent jouer avec des machines comme partenaires, un supercalculateur a été battu par un grand-maître avec un ordinateur portable assez médiocre. Mais à la surprise de tous, le tournoi a été remporté non pas par un grand maître associé à un supercalculateur, mais par deux amateurs avec trois ordinateurs portables assez faibles. Sankar  pense que c’est la façon d’interagir avec leurs machines qui a fait gagner des hommes moyens avec des ordinateurs moyens contre les meilleurs hommes avec les meilleures machines.

Très bien, mais quelle relation avec le Search ou l’Accès unifié à l’information (Unified Information Access UIA) ?

Peut-être la relation est-elle tenue, et peut-être je ne l’exprime pas bien, mais je vous sollicite de m’aider avec votre « puissance symbiotique cerveau-calculateur » pour affiner  mon argument.

Dans l’informatique “classique”, où l’on travaille avec des bases de données, des entrepôts de données (data warehouses), des systèmes de BI, etc. , des questions précises sont posées au (super) calculateur par des gens qui connaissent la structure de leurs données et maitrisent la façon de poser ces questions, et l’ordinateur sort des réponses précises du genre « 27 » ou des tableaux de bord sympathiques qui illustrent des chiffres et même des trends. Mais si vous voulez poser des questions qui vous amènent en dehors des structures de vos données ou de la logique prédéfinie de vos « systèmes décisionnels », vous n’aurez pas de chance.

Le Search, par contre, vous permet de poser des questions floues en langage naturel et il ne vous retournera pas une réponse du type « 27 », mais un ensemble de réponses – des documents ou des entrées d’une base de données – classées dans des catégories (des « facettes ») dans lesquelles vous pouvez naviguer. (Des informations de sources multiples, y inclus des applications métier,  peuvent être agrégés dans une catégorie) Vous pouvez zoomer sur des sous-catégories que votre intelligence humaine reconnait instantanément comme les plus prometteuses. Vous pouvez aussi raffiner votre question suite aux idées que le premier lot de réponses vous aura données. En effet, vous pouvez poser n’importe quelle question que vous voulez sans aucune nécessité de (re) programmer quoi que ce soit. Et dans un ping-pong de trois échanges avec votre solution de Search vous avez de fortes chances de découvrir une réponse que votre supercalculateur avec ses logiciels élaborés n’aurait pas trouvée. Ou peut-être il l’aurait trouvée, mais après quelques milliers de jours-hommes de développement et de mise au point, et des millions d’Euros dépensés pour un matériel de pointe – tout comme Watson a gagné le jeu Jeopardy.

Chez Sinequa, nous aimons penser que nos logiciels sont meilleurs que la moyenne, mais même si vous présumez qu’ils soient tout justes dans la moyenne, l’interaction des utilisateurs avec notre plateforme de Search et d’accès unifié à l’information (Unified Information Access, UIA) se rapproche assez de celle des deux amateurs avec leurs ordinateurs portables qui ont battu le champion de l’échec avec son supercalculateur.

+1Share on LinkedInShare on Twitter