Arkive.AI

At present, 20% of a typical employee's workweek is spent gathering information to complete tasks (McKinsey Global Institute). This time is not spent generating new information, but instead dedicated to searching for old information. By addressing this wasted fifth of time, we will help employees focus on creating value in their work.

Our product uses proprietary NLP technology to answer questions live in chat. The software curates a knowledge base to 'learn' about the business.

After scaling globally, our product will drastically improve the quality of electronic communication. This will improve the ability of people to collaborate remotely and rapidly equip new workers with the skills they need to thrive at a company.

Technological change has transformed the worldwide economy. In some ways, it has improved the quality of life for billions of people around the world. However, technological advancement is not without its difficulties. New technology is making some jobs redundant and furthermore, increasingly high skill levels are required of workers across industries. In fact, 53% of workers believe automation will significantly change or make their job obsolete within the next ten years and 77% of adults would learn new skills now or completely retrain to improve their future employability (Upskilling Hopes and Fears - PwC).

According to research by BCG, these problems will affect the developing world disproportionately. In the past, countries such as South Korea and Singapore were able to achieve economic development by leveraging low-cost factories to serve export markets before pivoting towards more highly skilled industries. This route, with the advent of Industry 4.0, will cease to be an option.

By facilitating better electronic communication, we will be able to improve access to technologically skilled work for communities in untapped countries. Our product will make it easier to re-skill workers through online training and allow collaborative enterprises to tap into the new economy.

Our solution is a 'chat bot' which operates within existing, free-to-use, chat platforms such as Slack and Discord. Our goal is to use NLP technology to learn the answers to company-specific questions.

At present, our prototype can group Question/Answer (QA) pairs together, giving managers, trainers or educators a single point to answer questions - removing the risk that they'll miss a crucial query. Moving forward, we are developing a technology to detect similar QA pairs and answer them automatically, as well as forwarding questions to the members of the organisation most qualified to answer them.

The front end of this technology will be indistinguishable from a colleague on the internal company chat platform. In fact, we like to make the analogy that Arkive.AI will fill the role of an incredibly knowledgeable colleague who always answers right away and finds out the answers to questions they don't know the answers to. This has many use cases across an organisation, from on-boarding new employees by providing a virtual helper to helping them work faster and with less tedium after they've spent months at the company.

Our solution targets entrepreneurs and workstation-based workers across the world. The solution was borne out of frustration at searching through email chains and Slack conversations for information we knew we already had. Further research has confirmed our suspicions - 20% of employee time is spent searching for information already available (McKinsey Global Institute).

The most impactful relationships we will develop will be with companies and individuals in untapped communities. As we deploy our product, it will be essential to continue a conversation with these people, to improve our product for them. Our software will improve with use, the QA pair detection will become better and better. We will also request feedback and reiterate our product time and time again.

The solution allows company-specific training and communication to be vastly sped up. For instance: Across the world, companies spend millions of dollars on Training and Development, in untapped communities however, access to capital seriously decreases the pace of this training. By providing our tools, we will reduce the barriers to company specific training and speed up the re-skilling of workers. Our product will continue to improve their working lives, and help companies in untapped communities grow.

Our project is aligned to both dimensions 1 & 3, but we've chosen to focus on the impact we can have on small and new businesses.  By augmenting communications within these organisations, we will directly impact their prosperity. Our technology, through a reasonable pricing structure, will dramatically lower the barriers to growth that these companies face. By reducing the number of hours spent on each problem, we will allow these companies to grow faster and employ exponentially more people than they otherwise would be able to.

Our two major competitors are Bloomfire and Guru. Both include some chat integration, however we feel that they both approach the problem in a way that we don’t feel addresses the underlying challenges: namely that people don’t like to write documentation or search for knowledge, they just want to ask a friend a question, and get results.  Both Bloomfire and Guru have approached this problem in the sense of “how do we integrate knowledge bases like wiki to make them easier to use?” our approach is “everything is chat now, nobody wants to use a knowledge base, they just want to ask an all-knowing colleague or friend”.

Arkive.AI approaches the issues of information retrieval differently to our competitors. For one, we focus on gathering information from chat. This gives us access to a far richer set of contextual information than is available from keyword search. We have also applied the latest NLP algorithms and award-winning proprietary techniques to build a methodology that outperforms conventional approaches.

Our focus is on building out technology to automate information processing. To this end, we have developed a methodology that outperforms conventional approaches by using proprietary NLP techniques and applying the latest research of the field.

Natural Language Processing (NLP) is our core technology. Ontology-Based Interpretation of Natural Language makes the analogy that while human beings have a very easy time understanding non-formatted text, interpreting nuance and implication without difficulty, machines have a far harder time. Interpreting language beyond its literal context is the key difficulty. NLP, therefore, is the technique of teaching machines to understand speech or text that hasn't been specifically formatted for their consumption.

Within NLP, we are focussed on Question-Answer (QA) pair detection. In the first instance, this allows our software to learn the answer to a specific question. Applying NLP means that we will be able to detect similar questions and respond with information already shared on the chat platform (i.e. "What's the capital of Peru?" and "What is the name of the capital city of Peru?" should both be answered with "Lima").

NLP also facilitates the front-end Software/Mobile Application we are building. This application provides an easy-to-use user interface for tapping into NLP, both for queriers and answerers.

NLP is already very widely used. Autocompleting, Virtual Assistants and even Search Engines utilise the technology. Autocompleting uses NLP to apply context to incomplete sentences in order to suggest possible words. Search Engines apply NLP to make related keyword suggestions.

Virtual Assistants such as Siri and Alexa make use of NLP in a very similar way to our product. They use NLP to process requests, whether they're web searches ("Hey Siri, what's the weather like today?") or commands ("Alexa, play Old Town Road"). The difference with our application of NLP is that we're focussing much more on specific knowledge, rather than generic web search or IOT connectivity ("Arkive, how did my boss ask me to format the sales data?").

Our initial assumption is that the trend towards digitisation in the workplace will continue at an increasing pace. This will not only change the nature of industries, but also the nature of work and collaboration. As technology becomes an even greater part of all industries, we anticipate that workers will need to be re-skilled to take advantage of the newly available jobs of the digital economy and that digital communication will become an increasingly valuable component of work.

First, we'll address the training issue. Already, there are communities with untapped potential for economic growth, around the world. They risk being left even further behind by technological innovation. Part of the reason for this is a high cost to train workers. Our technology will decrease this cost, by providing tools for automated training. Short term, this decreases the barriers to recruiting workers into the digital economy.

The digital communications point has longer term implications. By improving information retrieval and knowledge management, we'll reduce the support infrastructure needed to start and grow companies. In an increasingly remote and gig-economy, being able to work remotely will facilitate workers from all over the world to collaborate on projects.

Our solution is currently a prototype, so doesn't yet serve any customers. 

Within 12 months, we will have at least 10 accounts from the education sector and 3 enterprise accounts using the product as part of every-day work. These accounts will have at least 5000 total daily active users.

Over the next 5 years, we aim to target the Enterprise Knowledge Management Industry. It's hard to put a number on the number of people who'll work with our product, but our software has an addressable market of every worker using a computer workstation.

Within the next year, we will launch our education sector product to the public and launch a private preview for our enterprise product. In total, we are aiming to have 5000 daily active users in the next year. On the technical side of the product, we will be able to achieve high levels of accuracy and recall on question detecting and topic matching.

Over the next 5 years, we aim to further develop the product. Our aim is to realise the dream of building a widely used Knowledge Management tool.

Our product is not yet at the point where we can deploy it to customers. We need to improve our NLP technology to be able to provide the service we have described. Specifically, we need to improve our accuracy and question recall in order to be able to deliver a useful product to our customers.

Beyond technology, we will also need to focus on marketing. At present, none of our potential users are aware of the advantages that Arkive will be able to bring them.

On the technological side, we are very lucky to have a strong team of NLP researchers with a track record of publishing impactful research. We also intend to initially work with small groups of key users to understand the challenges they face and how our product can improve to meet the challenges.

The marketing barrier is partially a matter of building the technology, without a product there's no way we can gain traction, but also down to listening to our beta-testers. These initial customers will help us understand their pain-points and better articulate to new customers how we can help them.

In order to overcome this barrier, we have employed NLP researchers with a track record in publishing. 

Currently our team has 3 full-time staff, 3 part-time staff and 3 contractors. Our full-time staff are working on Software Engineering and NLP research, part-time staff in project-advisory roles and contractors on UI/UX design.

Our founder, Zhen Wei, has recently graduated from Imperial College London with a PhD in Machine Learning she is joined by Zhenhao Li, a PhD candidate in Machine Learning at Imperial College London and Chao Wu, a Lecturer in Machine Learning at ZheJiang University. Together, our team have already won the KDD 2019 award for Best Startup Research and the CogX award for Outstanding Research Contribution in AI from Zhen Wei's PhD thesis.

Yuan Gao, our Technology Advisor, graduated with an MEng from the University of Oxford and has held CTO and senior leadership roles at several technology startups. He was named a 2018 Forbes 30 Under 30 Enterprise Technology Listee and is a Forbes Technology Council member. He will advise us on taking the NLP technology to market and created our product prototype.

The rest of our team were selected from top Universities across the United Kingdom (Oxford and Imperial) and represent our commitment to top flight talent.

Our product is a Software as a Service offering providing a digital assistant. This product is integrated into existing chat technology, such as Slack and Discord, in order to improve knowledge retrieval.

The product is useful to people across the world, given that it removes time-consuming tasks (such as searching for information a boss has already sent in order to complete a task). However, we think it's especially useful to people in as-yet untapped communities. Our software will lower the barriers to engagement with the high-tech ecosystem. This will help members of these communities to benefit from economic growth.

Clearly, our product is not a catch-all solution to creating good jobs and sustainable entrepreneurship. However, our customers will recognise the power of software and technology to help augment job creation in their communities. 

We are looking to embed our social goals into our enterprise. We'll be charging enterprise customers for different features and functionality, which will be the long term driver for our financial sustainability. Alongside that, we are currently in the process of raising funds to help continue the development of our platform.

Solve can help us overcome 3 challenges: Marketing, Honing our Business Model and Funding.

We hope to be able to leverage the Solver network to better spread the word about our solution. By partnering with similar organisations and receiving marketing and exposure from Solve, we will be able to augment our efforts to take our product to market and maximise its impact.

Already, we have received helpful feedback on our business model from outside advisors. The decision to focus on a core group of committed customers to help build our user base and platform came from such advice. By working with Solve mentors, we hope to be able to hone our business model even further and leave no stone unturned in search of the best approaches.

Finally, funding from Solve would contribute to our ongoing fundraising efforts. We have relatively high staffing costs, due to the high standards we look for in employees and will need to carry on funding our product development before we take it to market.

Solve can help us overcome our marketing barrier. We hope to be able to leverage the Solver network to spread the word about our solution. Specifically, we are interested in partnering with similar organisations to deliver a more useful 

We are looking to partner with the Solve ecosystem to gain exposure for our brand. We acknowledge that the impact of our innovation is limited by our ability to spread the word about what we're doing and one of our primary goals from a partnership is to address this.

We also acknowledge that outside advice is the best way to iterate and hone our business model, so that we can maximise our impact and the number of people we are able to reach.

Finally, a core part of our current funding model depends on raising money from a variety of sources - including grants offered by Solve.

One of the key applications of our software is in rapidly training new workers and lowering the barriers to entry at companies across the world. Another key effect is reducing the amount of middle-office support that companies need in order to begin to scale.

If we were successful in our application for the GM Prize, we would be able to afford to employ a NLP researcher for over a year and reduce our time to market, and impact, greatly.

Our product is a chat-based virtual assistant, which functions as an ever-present and always helpful colleague for the employees of an organisation. We use Natural Language Processing (NLP) to analyse chat data and train our software to augment digital communications between colleagues. We have already won the KDD 2019 award for Best Startup Research and the CogX award for Outstanding Research Contribution in AI from a PhD thesis.

Using our technology, the product will be able to lower the barriers to entrepreneurship and good jobs in untapped communities. Our product will greatly reduce the cost of on-the-job training and reduce the amount of time employees spend on tasks that don't add value, such as searching for information to complete a task.

The AI for Humanity Prize Money would enable us to step up both our pace of research into NLP and development of the user interface that enables us to create a real impact with these technologies.

Our product is a chat-based virtual assistant, which functions as an ever-present and always helpful colleague for the employees of an organisation. We use Natural Language Processing (NLP) to analyse chat data and train our software to augment digital communications between colleagues.

Our current business plan involves rolling out a free-to-use version of the software to educational institutions in the first instance, which will allow us to improve the accuracy of our NLP technologies. From there, we will build up our enterprise technology to create sustainable revenues to drive the company forwards.

The Future Planet Capital Prize would help us to bridge the pre-revenue period and subsidise our free-to-use offering.