Follow us on social media

The Solver Case: How AI Turns Business Data into Gold

The Solver Case: How AI Turns Business Data into Gold

If there is one thing that any company anywhere in the world wants, it is to make the right decisions at the right time to make its business profitable and expand its horizons. Old-school methods relied on a blend of a nose for business, experience, leadership, and organization. Without relinquishing any of these factors, the age of artificial intelligence brings with it an additional linchpin: data.

In its most powerful form, data can be used to build predictive models. When will demand for a clothing brand’s e-commerce be highest? How likely is a factory machine going to break down in the next ten years? How can you better logistics by planning the best possible delivery routes? What price should a hotel charge for its rooms to attract more tourists?

Solver has the magic wand to turn thousands of statistics into business knowledge. Founded in 2016 as a spin-off from the Universidad Politécnica de Valencia (UPV), the startup began its journey through the hard work and vision of Jon Ander Gómez (CDO), Roberto Paredes (CTO), and Francisco Casacuberta.

“The idea was to market a technology that we knew was mature enough to help the business community,” said Jordi Mansanet (CEO). “AI is a very wide-ranging tool that can improve many processes, yet one common feature is extracting value from company data, automating operations that are not streamlined, and making the right decisions to attain levels of efficiency which would be impossible for humans to achieve.”

Solver’s software fits each customer like a glove. Today, with the current boom in AI, many companies are taking an interest in the Valencian startup’s solutions.

First of all, Solver explores the business model of the company in question to find out where it makes sense to apply artificial intelligence. “We are especially keen to learn what their challenges are and where they need to improve,” commented Mansanet. “Based on that we can then assess whether AI makes sense or not. Sufficient historical data is always a must for the answer to be ‘yes’.”

If industrial machinery is not fitted with sensors, there will also be no history to feed the algorithm’s knowledge. In contrast, if a retail business has previously digitally transformed its routine operations, AI can hone demand forecasts by 30% to 50%. The more data the platform receives, the better its predictions will be.

In tourism, which is highly digitalized in all its aspects (accommodation, flights, car rental, entertainment in the city visited), Solver moves like a fish in water. With its support, industry businesses can enhance the user experience and personalize purchasing processes by analyzing behavioral patterns and putting in place individualized strategies. Why didn’t the traveler book the hotel room they were looking at? At what point did they stop? What if they were emailed a deal? There are lots of options.

“If the project with a customer goes well, our AI usually spills over into other areas of the company,” pointed out Mansanet. “We normally set KPIs at the beginning of the job. This is how we quantify the project and how much the outcomes need to improve. We have extensive experience in a range of sectors: retail (fashion and food), industrial logistics, pharmaceuticals, and tourism.” Solver is also active in areas such as finance, insurance, and energy.


The AI chess board

Aside from the current crop of headliners (ChatGPT, DALL-E-2, Stable Diffusion, Midjourney), Mansanet said there are three types of companies competing on the artificial intelligence board. Some of them offer closed AI-based products. “This is not our business as it is extremely difficult to tailor these kinds of products to the customer’s specific needs.”

Another group are the more generalist, “almost consultancy” firms which are not specialists in this type of technology even though they are engaged in software development.

And then there are startups like Solver. “The fact that our niche is AI and we have a proven track record means we have a better chance of success” argued Mansanet. “There is not as much competition here.”


Entry barriers

Mansanet noted above that the secret of AI lies in the contracting company meticulously measuring its processes. These must be highly digitalized organizations. Once this requirement is met, the power of the algorithm is unleashed.

“A lot of companies put lots of resources into document analysis. If it takes a person 50 seconds to a minute to review a dossier and extract the fields by hand, AI will complete this task in two to three seconds.”

The algorithm cannot thrive without sufficient historical data. The other common mistake Mansanet identified has to do with the approach to AI projects. “It makes little sense for just the innovation department of a firm to be involved. Several departments in the company have to pitch in and everything must be geared towards addressing a business need.”

Next August Solver will be eight years old with sixteen professionals on its payroll (90% technology backgrounds). After consolidating its position in Spain with premium customers, the startup is now expanding internationally (finalizing agreement in Mexico) and taking part in two European Union flagship projects which might be joined by a third in 2024.

To round off two solutions it has successfully developed and tested in practice with numerous customers, the startup is soon to release two other tools: Forecaster (demand forecasting) and Docreader (document information search).

“AI has been around for a while even if it often works in a less showy way than ChatGPT,” said Mansanet. “This is no passing fad. We keep a close eye on what goes in universities, and we can see how fast this technology is evolving and the awesome impact it will have when it goes mainstream among users and companies.”

Solver is here to stay.

Apr 11 · 2024 GoHub Ventures

Aunoa and the AI Customer Service Revolution

Aunoa and the AI Customer Service Revolution

Ask any technophile about the latest buzzword and two words will immediately come to mind: artificial intelligence. No metaverse, no blockchain, no web3. No other phenomenon matches today’s most creative algorithm’s excitement, analysis, insights and potential, capable of mimicking human voices, solving complex equations and writing academic essays in the blink of an eye.

ChatGPT, Claude or Gemini have taken centre stage in front of a public eager for novelties. They are the current stars of today’s show, the benchmark tools on everyone’s lips, yet in the wings there are hundreds of other solutions created by highly disruptive startups. The Valencian startup Aunoa belongs to this avant-garde club.

Launched in 2019 by four founders who have been working together for more than two decades, Aunoa focuses on automating processes linked to customer service, “although it is not the only solution we market,” notes Fernando Pérez Borrajo, the startup’s Corporate Director.

“Our goal is to offer companies a comprehensive solution for all their customer management processes, from the resolution of queries to the management of transactional and documentary processes, also using other conversation channels such as WhatsApp and Instagram,” explained Pérez Borrajo.

The approach could not be simpler: free customer service departments from repetitive and tedious tasks and instead let the combination of machines and humans handle all consumer and citizen queries, as Aunoa’s software can be used in both the private sector and public agencies.

In addition, agents in these areas have a kind of internal chatbot that provides them with response suggestions based on what other agents have answered in similar situations; it helps them classify incidents and suggests semi-automatic responses. The goal is to save time and avoid bottlenecks.

All too often, customer service is a headache for companies. Turnover is high because people get in touch when they have a problem, but if the response is slow in coming, tempers start to flare.”

Losing an employee leaves a big void. Their skills disappear with them, and someone else has to be trained. This costs a lot of money. With our tools, we try to keep the workers from leaving. However, if they do, there are no big gaps because there is already a well-organized knowledge base,” says the Corporate Director.

Despite their shortcomings, large language models like ChatGPT are outstanding: they have been trained with all the data available on the Internet and provide an ideal environment for completing a battery of tasks. A human agent, for example, will give a relatively simple response to the consumer. Still, this action can be supplemented by the algorithm composing the response to make it friendly and assertive.

While this additional help is invaluable, Aunoa never leaves it up to these linguistic models to respond directly, adding instead an intermediate layer of self-experience in the form of supervision. In fact, the transformer models used by OpenAI are also the basis of Aunoa’s AI, designed ad hoc for each client.

The two systems are thus seamlessly connected: Aunoa has a variety of anonymized and pre-trained models for different industries. There are e-commerce and logistics models for utilities (electricity, water, gas, telecommunications, banking and insurance) and for the public administration. These models safeguard the rules imposed by Europe, where the algorithm cannot be trained with data provided by customers. Aunoa then connects to these other major dominant models for very specific tasks (summaries, document compilation, response improvement, etc.).


Business overview

Aunoa has more than 100 relevant clients in 10 countries and exceeds the 1 million € of ARR currently. Its artificial intelligence works in Spanish, English, Portuguese, Italian and the co-official languages in Spain (including Basque). To implement each language, it is necessary to have at least one linguist in what seems to be an ode to the humanities in extinction.

“Our average ticket is around 1,200 euros per month. This may seem like a high figure, but it is not. The language models downloaded from the Internet are not sophisticated. If a company really wants to turn customer service around, it needs complex structures that require training and follow-up. That’s what Aunoa offers.” summarizes Pérez Borrajo. The churn rate is practically non-existent with this service model.

In addition to not losing customers, Aunoa benefits from another competitive advantage. More than 40% of its revenue comes from product upgrades. A company usually starts with an essential service but typically chooses to up the ante and deploy more ambitious ideas and services in more channels.


New developments on the horizon

The startup’s next giant leap will come with the widespread adoption of generative AI, not just in text, but in voice, images, or video to deliver a fully integrated environment” in less than a year. These new services will not only deploy ‘multi-sense’ environments to help the frustrated consumer at any time of the day. Still, they will also provide powerful tools to human agents to help them analyze and understand customers’ problems and sentiments.

“The human service is often outsourced. Large companies are looking to pay the lowest possible price, so they turn to call centers located in other countries. However, if the service is poor, they lose a customer that has cost the company 700 or 800 euros to win over. Nobody can afford that, which is why customer service has become such a strategic issue. The potential of the combination of people and robots is extraordinary.”

Aunoa is not dazzled by the hype surrounding AI, and his pitch is as slick and thorough as his solutions.

Mar 25 · 2024 GoHub Ventures

Felix Laumann (NeuralSpace): “Our goal is to become a world leader in language AI in Asia and Africa”

Felix Laumann (NeuralSpace): “Our goal is to become a world leader in language AI in Asia and Africa”

UK-based NeuralSpace debuted in 2021 with a powerful language AI model that can translate text and speech into over 100 languages and regional dialects with an emphasis on Asia, Africa, and the Middle East. Translation accuracy, linguistic variety, and data privacy are the hallmarks of its platform.

Felix Laumann is the CEO of this startup based in King’s Cross, London. To date, the company has raised more than $3 million with the backing of GoHub Ventures.


Question: What would NeuralSpace’s pitch be like today if you were on stage at the most important event on the planet?

Answer: I honestly believe that nothing would change in my speech compared to the ones I gave at the outset. In the last couple of months, AI applied to language has attracted a lot of public and media attention, especially owing to LLMs (large language models such as OpenAI’s with ChatGPT). However, the number of sensitive tasks that call for in-depth understanding of language whether spoken or written is still huge, ranging from scrutinizing competitors based on their online presence to putting together a compelling marketing kit or disentangling a call between a customer and an agent in an attempt to optimize that communication.

Many of these tasks can be automated in part or in whole with the help of AI. Plus, much of the technological race in this field is anchored in using only English and other European languages, whereas NeuralSpace seeks to deliver highly accurate solutions for languages and regional dialects spoken in other parts of the world.


Q: What makes NeuralSpace different in a market with so many rivals?

A: Most AI tools work with European languages even though 90% of the world’s population speaks another native language. Our company was set up to redress this imbalance. We have developed the most robust and customizable technology regardless of the industry or use case to automatically translate from speech to text and to enhance the effectiveness of virtual assistants (chatbots, voicebots).

In addition to helping developers to craft their own software using NeuralSpace’s APIs, we also offer enterprise solutions in which we personalize our products. When it comes to complex enterprise solutions, we strive to fully understand the customer’s business problem before suggesting any solutions. For example, as explained in GoHub Ventures’ Tech Unpacked inaugural edition, oil giant Aramco uses our DocAI solution to analyze complex documents and extract qualitative content, a process previously done 100% manually.


Q: Let’s talk about NeuralSpace’s next milestones and where you see the company five years from now when the AI revolution will have fully spread its wings.

A: Our goal is to become the world’s leading provider of language AI in African and Asian languages and dialects. The potential fields of use of our technology are vast. In the short term, we are expanding the capabilities of our AI-driven enterprise solutions and a good example of this is our speech recognition tool, VoiceAI. We are also delivering a world-class development experience, enabling more professionals and organizations to harness the transformative power of language AI in their software applications.

Our long-term commitment is straightforward: to help all kinds of companies in Asia, the Middle East, and Africa to ramp up their productivity, efficiency, and cost savings, thereby putting them in a prime position to significantly enhance their performance.


Q: Let’s talk about the AI boom. Will there be room for everyone or will big tech firms verge on oligopoly with hegemonic solutions and/or buying out startups with disruptive proposals?

A: The market is so large that big tech firms and startups will rub shoulders. It is likely that all tools which can tap into the capabilities of artificial intelligence will leverage this technology in the future as companies need to stay competitive. The surge of innovation we have seen over the last 12 months is incredible.

In this scenario, startups have the agility and speed to make the most of technological breakthroughs and adapt to customer needs which often call for bespoke models or product adaptations. The big players can’t offer this level of customization.


Q: Turning to regulation, there are two opposing schools of thought: the more liberal led by people such as Marc Andreessen who advocate letting artificial intelligence develop with hardly any restrictions, and the more moderate players (Geoffrey Hilton, Sam Altman) who are pushing for clear limits.

A: The AI industry needs to evolve responsibly and transparently. Safeguarding security, privacy, and human rights must be a priority. Our team fully grasps this integrity requirement and applies it to every aspect of the process, from data collection and algorithm design to product rollout and customer interaction. In fact, my feeling is that the AI and machine learning community is more than willing to engage in a healthy regulatory discussion and work with society to strike a balance.


Q: Some compare the emergence of generative AI to the invention of the printing press or the automobile. Where is humanity heading as a result of this revolution?

A: There is no question that artificial intelligence will completely change the way businesses operate. More productivity means more growth. Automating tedious tasks also enables the employee to focus on more complex issues and feel more satisfied with their mission. At the same time, the company gets more out of its workforce. New jobs will also come along, including the position of head of AI strategy (a kind of CAI working hand in hand with the CEO). The conversation has shifted significantly over the last year: the question is no longer whether a company should use artificial intelligence. The question is how to leverage it to stand out from the competition.

Feb 05 · 2024 GoHub Ventures