AiThority Interview with Dorian Selz, Co-Founder & CEO at Squirro

Hi Dorian, welcome to our AIThority interview Series. Please tell us about your tech journey so far.

My tech journey started with a PhD in information systems at the University of St.Gallen. Out of the institute we founded Namics, which became the largest e-business consultancy in Switzerland and Germany.

One of my clients turned into my next company local.ch. We digitized the Swiss yellow page space. We grew local.ch to be the market leader in four years. In 2012, I co-founded Squirro, an AI company specializing in data-driven insights.

Please tell us more about your AI offering and how you market these to your ideal customer base?

Squirro provides AI-driven insights and recommendations extracted from our client’s internal and external, primarily unstructured data. We call it “generative AI-enabled Semantic Search and Insight Cloud”. Our insights applications offer effectiveness and efficiency benefits to a broad range of corporate functions from Knowledge Management, Sales, and Service Management to Risk, Compliance and Audit Management, or Supply Chain Management primarily in Financial Services, Manufacturing and Pharmaceutical industries.

To put it into perspective: We observe three global trends driving demand for these types of applications: 1) The siloed IT landscapes; 2) The data flood and 3) the current economic, geopolitical and demographic developments.

To start with the first trend: In 2023, an average company is working with around 1’000 applications. This number is constantly increasing with around 100 new applications being added to the corporate application landscape every single year. Less than 30% of these applications are integrated. The overall data volume has been growing exponentially for years, and it is expected to double in size between 2022 and 2026 specifically in the B2B sector. 90% of this total data volume is unstructured data. This means that the employees are forced to process an exponentially growing volume of text-based information to be able to perform their duties. Given the current cost constraints and the shortage of skilled employees, the majority of all companies are facing nowadays, this results in a so-called “sampling” – instead of sifting through all the relevant information (which becomes impossible). Subject matter experts just focus on some specific parts of it. However, this bears a very significant risk of overlooking or missing relevant insights: Not recognizing a new sales opportunity will negatively impact your sales revenue, and not recognizing a customer problem fast enough will damage the customer experience, which can result in the loss of the entire customer relationship. And overlooking some specific risks or regulatory changes can result in quite significant fines.

Given the current challenging economic, geopolitical and demographic environment, companies are facing significant supply chain risks, an ever-growing regulatory pressure accompanied by an ever-growing  lack of experienced resources, and significant cost pressure. The severity of risks is increasing. With baby boomers retiring, new younger employees, which don’t have the advantage of decades of work experience and are literally “lost” in the corporate data floods and siloed application landscapes, are now entering the stage.

How does Squirro help here? :

• Gathering:  Connecting to and continuously consuming all of the relevant data sources
• Understanding: Using cutting-edge machine-learning techniques to find the patterns, anomalies, and trends within that data that produce the insights and recommendations
• Acting: Delivering those insights and recommendations to the right people, at the right time through easy, intuitive, and customizable dashboards (or through a ChatGPT-like interface, see next question) that can be accessed directly or embedded within the user’s work environment

Squirro’s Insight Cloud is built on its core product, called the Insight Engine (Squirro is awarded Visionary in Gartner Magic Quadrant for Insight Engines 2021 and 2022).

Recommended:  AiThority Interview with Itamar Kandel, Chief Executive Officer at Vista.ai

AI firm Squirro recently launched a new enterprise-ready Generative AI solution, that has impacted enterprise use of Gen AI services such as Chat GPT. Could you tell our readers some more about this technology?

Large Language Models like ChatGPT hold great potential for enterprise applications. However, certain limitations must be addressed to make them genuinely useful in a business context. By focusing on reliability, contextual understanding, and security, LLMs can be optimized for enterprise usage.

Reliability is crucial for enterprise adoption as generative AI models sometimes produce hallucinated outputs. To tackle this issue Squirro’s approach employs rigorous evaluation, monitoring, and evidence basing of results. This proactive approach can significantly enhance the reliability of AI-generated content, making LLMs more dependable for enterprise applications.

Incorporating context and local training is essential for LLMs to comprehend specific business needs and industry jargon. Customizing these models with proprietary data and domain-specific knowledge can vastly improve their performance in understanding and generating responses that align with the organization’s requirements.

Lastly, addressing security concerns is paramount in the enterprise environment. LLMs’ potential disregard for data security can be mitigated by implementing local access controls on the organization’s data. By enforcing strict data management protocols and ensuring that AI models only access authorized information, businesses can maintain data integrity and compliance while utilizing the capabilities of LLMs.

Squirro was able to solve these genAI issues in such a timely manner thanks to its core product, the Insight Engine, which provides the relevant foundation to perform all of the mentioned activities.

How do customers leverage Squirro to get more value from their data?

Customers leverage Squirro to extract more value from their data by utilizing its advanced AI-driven insights and analytics capabilities. Squirro’s platform enables users to automatically discover, understand, and visualize data patterns, trends, and anomalies, transforming raw data into actionable intelligence. By providing contextually relevant and personalized recommendations, Squirro empowers businesses to make data-driven decisions, enhance customer experiences, and uncover untapped opportunities, thus maximizing the value of their data and driving growth in a competitive landscape.

To give you some specific examples: By using Squirro in Knowledge Management, and specifically now with the availability of a GenAI user interface, our clients save about 95% of their time on searching for relevant information. In a case of a large global investment company, they are now able to produce investment-related dossiers in about 30 minutes instead of 2 weeks. In a case of a large manufacturing company, a sales report is produced within less than 5 minutes instead of 45 minutes. Efficiency and speed are just some of the benefits. Others address the quality and effectiveness of their internal decision-making in risk and compliance, in patent screening, in drug monitoring, and alike.

Recommended: AiThority Interview with Jim Kaskade, Chief Executive Officer at Conversica

The appetite for Gen AI is vast combining techs like LLMs and Composite AI. Please explain how AI is being used today?

AI is being used today in numerous applications, leveraging LLMs and Composite AI to create sophisticated, interconnected solutions. Industries such as healthcare, finance, marketing, and manufacturing benefit from AI’s capabilities to analyze vast datasets, generate insights, and automate processes. AI assists in personalizing customer experiences, detecting fraud, optimizing supply chains, and improving diagnostics and treatment plans. Furthermore, AI-powered chatbots and virtual assistants help streamline customer support, while natural language processing (NLP) and computer vision technologies enhance human-computer interaction and enable new forms of content analysis.

In one of your statements you have commented on ChatGPT- its results lack up-to-date context and it cannot access information sources outside its own model. Can your new model beat ChatGPT?

It is well known that ChatGPT is currently limited to public data until about 2021. In an enterprise context, you want to tap into enterprise data, which ChatGPT does not offer. By incorporating up-to-date context, accessing external information sources, and using continuous model updates and training we have seen significant improvements in the resulting quality.

What’s the future of ChatGPT-like tools for customer experience and sales generation?

The future of ChatGPT-like tools for customer experience and sales generation is promising, as AI-driven language models continue to evolve and improve. These tools are expected to become more personalized, context-aware, and capable of real-time engagement. By seamlessly integrating with various customer touchpoints, they will enhance customer support, generate targeted product recommendations, and nurture leads more effectively. As AI models become better at understanding human emotions and intent, the resulting customer interactions will be increasingly authentic, leading to enriched experiences and increased sales generation opportunities for businesses.

What are the major challenges in making AI more accessible to local communities? How do you overcome such hurdles?

Major challenges in making AI more accessible to local communities include limited resources, lack of infrastructure, and insufficient technical expertise. To overcome these hurdles, partnerships between governments, NGOs, and tech companies are crucial in providing affordable resources, infrastructure development, and capacity building. Initiatives such as localized AI training programs, community-driven AI projects, and accessible open-source tools can empower communities to harness AI’s potential. Additionally, promoting digital literacy and offering multilingual AI solutions can break down language barriers and foster inclusivity, further democratizing access to AI technologies.

Could you please explain to our readers about  “NLP text classification” along with an example?

NLP text classification is an AI-driven technique used to categorize text into predefined groups based on its content or context. It employs natural language processing (NLP) to analyze and understand human language, making it possible to automatically sort and organize text data. For example, an email filtering system may use NLP text classification to categorize incoming messages as “spam” or “non-spam.” By analyzing keywords, phrases, and linguistic patterns within the emails, the system can accurately classify them, ensuring a clutter-free inbox and helping users focus on relevant content.

As an AI leader, which industries do you think would be the fastest to adopt AI/ML?

I believe industries with a significant reliance on data analysis and automation are the fastest to adopt AI/ML. These include finance, healthcare, manufacturing, and retail. The finance sector leverages AI for fraud detection, risk assessment, and algorithmic trading. Healthcare uses AI for diagnostics, drug discovery, and personalized medicine. Manufacturing applies AI for process optimization, quality control, and predictive maintenance. Finally, the retail industry adopts AI for inventory management, personalized recommendations, and customer support. These sectors continue to push boundaries as AI and ML technologies evolve and mature.

Recommended:  AiThority Interview with Yan Chow, Global Healthcare Leader at Automation Anywhere

What are your predictions for the AI domain for 2023-2024?

My predictions for the AI domain for 2023-2024 include continued advancements in AI capabilities, increased integration of AI in daily life, and a stronger focus on ethics and responsible AI. We can expect breakthroughs in natural language understanding, computer vision, and reinforcement learning, leading to more sophisticated and human-like AI applications. The proliferation of AI will impact industries such as healthcare, education, and transportation. Additionally, the global conversation around data privacy, fairness, and transparency will gain momentum, resulting in stricter regulations and guidelines to ensure the responsible development and deployment of AI technologies.

Thank you, Dorian ! That was fun and we hope to see you back on AiThority.com soon.

[To share your insights with us, please write to sghosh@martechseries.com]