The concept of incorporating AI into information searches isn't novel. This is something Alex had been involved with during his tenure in his previous role, automating machine learning categorization and the search for customer requests. What has changed, however, is that recent strides forward in AI conversational tech have fundamentally altered the way individuals wish to interact with information.
We were fed up with navigating through unwieldy FAQs. Possessing a solid computer science background and extensive machine learning knowledge, we felt compelled to explore whether it would be possible to make this process more intuitive and intelligent. After conducting numerous interviews and creating an application for the LiveChat marketplace, we observed the first signs of positive progression through the test conducted in collaboration with the LiveChat team. We recognized the potential for broader applications of this technology in knowledge management. Consequently, we decided to intensify our efforts on conducting interviews and creating prototypes, all while exploring various utilization possibilities.
What
applications
are we talking
Domains most affected by technological advancements are those where people naturally interact with information a lot, seeking either knowledge or elaboration. AskGuru focuses primarily on the following areas:
Customer support. Here, our tech comes as a support agent copilot and customer self-service chat. Tests we conducted with LiveChat Inc. showed 60% better answers than existing chatbots, and 55% of solutions were the same as the real human agent.
Knowledge management portals. Powerful semantic search is robust to typos and synonyms and provides accurate answers and source links for company intranets, knowledge bases, and such.
Learning management systems. Our most recent direction. We are running pilots of AI tutors helping students catch up with course materials by answering their questions and pointing to relevant resources.
So, is this
yet another
AI chatbot?
Not quite.
Our interviews revealed a significant trend in our explored domains: most of their software is Software-as-a-Service (SaaS). In-house solutions for customer support and knowledge management are exceedingly rare. Usually, businesses utilize various services provided by various B2B companies, both large and small. It seems logical that these service providers will eventually adopt the AI capabilities that their customers desire. This would render any externally built third-party tool redundant, however appealing and AI-rich it might be.
While giants like Zendesk and Intercom have already promised or rolled out AI functionality to their business customers, the situation with small to medium-sized business (SMB) providers is different. These providers maintain a ready client base and have budgets designated for AI enhancements; however, they often lack a dedicated machine learning team to implement these enhancements in line with rapidly evolving AI technologies. Some may prefer to acquire these capabilities in the form of plug-and-play tools. This is where we see our opportunity. Offering a web API and on-premise deployment options, AskGuru empowers companies to test their product theories and deliver high-quality AI solutions much quicker than ever before.
But isn’t that just a couple of
OpenAI API
calls?
The answer is both yes and no. Projects like Langchain have helped significantly streamline the prototyping of language learning models (LLMs), making them sufficient for basic Q & A applications. However, real-world applications necessitate providers to offer a host of additional features like:
Robust user-data permission models,
Multilingual conversation support,
Accurate citation of sources,
Integration of multiple sources such as documents, chats, and many more,
A continuous feedback loop for iterative improvement,
Detailed logs, metrics, and dashboards,
Importantly, what's often overlooked is that for tasks like these, using ChatGPT— a common approach— is akin to using a sledgehammer to crack a nut. More nimble and precise domain-specific models can be trained, which would help optimize latency and cost-effectiveness. This constitutes an area far beyond mere prompt engineering. So, as you can see, the process involves complexities that go well beyond just a couple of API calls.
Can
we
achieve all of this?
The short answer is we already have. While we recognize that there's always room for improvement, AskGuru addresses many business use cases. Moreover, we are constantly developing new features and tools to serve our clients better.
The composition of our core team speaks volumes about our capabilities.
Alex is a seasoned Natural Language Processing (NLP) machine learning engineer with a successful career stint at industry-leading companies like Yandex and Booking.
Sergei also brings a wealth of experience. With work experience as an ML Engineer in both startups and large corporations like Snapchat, he carries a unique blend of expertise. Complementing his Computer Science and Data Science credentials, Sergei is also a graduate of Product Management school. He has effectively led projects and teams as a technical lead, demonstrating an ability to steer our endeavour in the right direction.
How far-reaching is this
opportunity
?
The Knowledge Management software market is gigantic. In 2022 alone, it's estimated to be worth a staggering $547.5 billion and is expanding at an impressive compound annual growth rate (CAGR) of 17%. At the outset, we aim to target the subset of this market, focusing on small to mid-size knowledge base and customer support software, a field valued at a substantial $500 million.
What's the current status
status
of the project?
From a traction standpoint, we have made notable progress. An active contract is underway with OneClick, a customer support provider based in India. Furthermore, we’re actively engaged in discussions and currently operating four pilot projects with companies across Knowledge Management, Customer Support, and LMS domains. Our application in the LiveChat marketplace is installed on more than 2,000 accounts.
On the financial side, we have secured $20k in funding from the Pioneer accelerator. This investment is a solid foundation for our project's ongoing growth and development.
What's necessary to guarantee our
ultimate success
?
To ensure the continued growth and success of AskGuru, we need to augment our engineering capability and business development efforts. To this end, we're looking to raise $500k. Our plans for this funding include:
We are devoting resources to our engineering efforts to support current service providers and develop new features. This includes the enhancements of AI agents to respond, clarify and take specific actions, such as cancelling subscriptions.
We are investing in sales and business development to reveal new potential avenues for our technology.
However, our needs continue beyond financial backing. We're also on the lookout for strategic advice and networking opportunities. If you see potential in our solution and can assist us through advisory services or connecting us to interested parties, we're more than willing to explore such possibilities.
