Conversational AI For The Enterprise.
AI-based Virtual Assistants and Live Chat.
Unparalleled flexibility in end-to-end chat automation.
End-to-end chat automation
Unlimited flexibility in chat automation.
Use virtual assistants and intelligent routing to optimize your chat processes.
We Differ By Offering Unrivalled Depth In Building Virtual Assistants
These companies use our product
1,000,000 customers
Leading telecom in Lithuania
Telecommunications
Our work: virtual customer assistant on the website.
2,000,000 customers
UK Digital-Only Bank
Banking
Our work: virtual customer assistant in iOS and Android.
200,000+ customers
Nasdaq Tallinn: LHV1T
Banking
Our work: Virtual customer assistant on the website and agents’ live chat product.
9,000,000 passengers / year
Nasdaq OMX Nordic: TAL1T
Marine transport
Our work: Virtual customer assistant on the website.
500,000+ customers
Leading telecom in Estonia
Telecommunications
Our work: Virtual customer assistant on the website and agents’ live chat product.
500,000+ customers
Leading telecom in Latvia
Telecommunications
Our work: Virtual customer service & sales assistant. Agents’ live chat product.
Employing 300+ people
Estonian Statistics Agency
Government
Our work: Virtual customer assistant on the website and agents’ live chat product.
100,000+ customers
Dynamic German energy company
Energy
Our work: Virtual customer service assistant in Facebook.
Case Studies
“AlphaBlues has made our customer service even more efficient with their artificial intelligence based solution“
– Lauri Haav, Head of Customer Loyalty, Monese Digital-Only Bank
Business Issue
Monese is a revolutionary neo-bank founded with the ambitious goal of building a global instant-open bank account for the on-demand age. Monese was born from its founder’s painful experience when he was denied a bank account when he moved to the UK. With the Monese app, customers from all over Europe can open a bank account in as little as 120 seconds, free from the hidden fees and restrictions that legacy banks impose. Monese was the first 100% mobile bank account to ever launch in the UK. The company’s customer base is fast approaching six digits and customers are moving over 0.5€ billion annually. To handle the increasing amount of customer service requests from a growing customer base, Monese was looking to automate its core customer service chat.
Solution
To manage customer support more efficiently, AlphaBlues deployed its Virtual Customer Assistant (VCA) technology into Monese in-app customer support chat. The artificial intelligence based VCA was made the front-line customer service representative for the thousands of customer support requests each month. It was trained to automatically handle natural language and specific questions that customers have about Monese and its services.
Results
- The Virtual Customer Assistant has had thousands of conversations with customers.
- The VCA has autonomously solved (i.e. without any human help) more than 15% of the customer service requests it has encountered.
- It works 24/7 and is available during off-hours with an instantaneous response time. The solution provides instant alerts to customer requests in case of urgent updates and service announcements.
“With the AlphaBlues product our customer service in Facebook Messenger gives a better customer experience.”
– Liisa Ruusamägi, Client Communication Product Manager, LHV Bank
Business Issue
LHV Group is the largest domestic financial group and capital provider in Estonia that is currently holding 2 billion euros of Estonian assets. LHV Group’s key subsidiaries are AS LHV Pank and AS LHV Varahaldus. LHV has representative offices in Tallinn and Tartu. The company employs over 300 people and has more than 5000 investors. LHV Bank services are used by more than 120,000 customers. Pension funds managed by LHV have over 175,000 customers and are the second largest in Estonia by volume. The company is one of the most innovative banks in the region and was looking for a new and efficient way to engage with their customers 24/7.
Solution
To bring their customer service into the messaging era, LHV turned to AlphaBlues and its Virtual Customer Assistant (VCA) technology. AlphaBlues worked with the bank to integrate their customer service knowledge base with hundreds of topics and intents into the VCA which was launched through the bank’s Facebook page onto Facebook Messenger. Artificial intelligence based training was used to increase the relevancy and accuracy of the VCA in order to allow for more self-service by the bank’s customers. The VCA was given a friendly persona and it was taught to understand free text and natural language so as to be able to help people with the wide variety of questions they might have.
Results
- During the first weeks of the launch the Virtual Customer Assistant already handled more than a thousand customer conversations.
- More than 50 percent of the users have rated their experience with the VCA as useful and helpful in providing the answers they were looking for.
- The solution is available 24/7 and also during regular customer service off-hours providing automatic support to customers.
“AlphaBlues increased conversions with their intelligent virtual customer assistant in our online store.”
– Inese Zarina, Head of Chat Automation, Lattelecom
Business Issue
Lattelecom is a dynamic and innovative company on a mission to bridge the gap between technology and the individual. They are one of the biggest companies in Latvia employing more than 1500 people. Among other business lines the company operates one of the largest electronics e-commerce stores in Latvia that sells smartphones, TVs, computers and other electronic devices. The company was looking for a solution how to increase the engagement and conversion rates of their customers during the Christmas holiday shopping period.
Solution
To provide their e-shop users with a new experience, Lattelecom turned to AlphaBlues and its Virtual Customer Assistant (VCA) technology. AlphaBlues worked with Lattelecom to create a chat-based virtual customer assistant to help users in purchasing products. The assistant, named Anete, was put on the main page of the e-shop as a chat icon. Users could click on the chat icon and immediately see the numerous ways how Anete could help them. They could browse for products for the family as Anete suggested top gift recommendations for husbands, wives, kids and friends. They could also pick the best selling gifts based on product categories (e.g. phones, TVs, computers) or browse products with the lowest price. In addition Anete helped answers questions regarding tracking the ordered package and ways how to lease products.
Results
- During the Christmas shopping period in the second half of December, more than 4,500 users engaged with the virtual customer assistant.
- Users who engaged with the virtual assistant were on average 260% more likely to buy products on the website than those who did not engage with the virtual assistant.
- Users who clicked on products recommended by the virtual assistant were on average 380% more likely to buy products than those who did not engage with the virtual assistant.
Features
VIRTUAL ASSISTANT (VA)
Building environment for virtual assistants
Graphical interface for creating intents (i.e. topics) for the virtual assistant.
Group intents under specific folders, expand folders for easy visibility and organization.
Add buttons, quick replies, conditions, images, URLs to VA answers.
Customizable answer delay setting in milliseconds.
JS API for front end integration.
Create different menus which are triggered in different URLs where chat window is deployed.
Deep nested multistep workflows.
Unlimited number of actions.
Context specific responses.
Training
Insert training phrases into the VA.
Change the intent matched to phrase.
Easy visibility into training data across intents.
Free text search across all intents and trained phrases.
Export and import phrases into and from CSV.
User chat messages sorted by a customizable date range.
AI predicted intents for users’ chat messages.
Ignore predicted intent when training.
Measure AI intent prediction accuracy.
Machine learning
Ensemble of machine learning and deep learning algorithms to understand the customer intent from free text.
Functional in any European language + Russian + Arabic.
Co-intent understanding.
Intent ranking and prioritization based on confidence scores.
Offer choice of multiple intents for low confidence matches.
Statistics
Time interval selection for seeing the number of VA conversations.
Top intents sorted by the number of occurrences.
Least triggered intents by the number of occurrences.
General helpfulness score of the VA for the customers.
Detailed intent based helpfulness score from customers.
Count of interactions with text; with buttons; with text and buttons.
Intent report with count of sessions, feedback requested / provided / positive / negative per specific action.
Button click report.
CHAT
General
Add chat to multiple websites.
Customers can rate agents.
Customers can provide comments as feedback to agents.
Receive messages as tickets or have them sent to email during off-hours.
Canned answers (saved replies) in the chat answering window and emojis.
Customers can send and receive files through chat.
Agents can assign chats to other agents.
Info on customers waiting in each queue and agent availability.
Adding comments to the chat conversation for leaving notes to other agents.
Agents can see history of the conversation between the customer and the VA.
Clear distinction of the handover from the VA to the agent.
URL shown from where customer came in.
Accepting/not accepting chats button for each agent.
Ability to gather various info about customer including IP, OS, chat ID.
Notes field for agents to type notes.
Configurable custom attributes in chat dashboard.
Queues
Create custom queues for receiving tickets.
Assign agents to queues.
Visual notification of new messages in browser tab.
Sound notification of a new message.
Unread messages highlighted in the queue.
Timestamp showing time since incoming.
Tag tickets.
Autoclose inactive tickets.
Statistics
First response time, average response time.
Agent first and response time.
Number of chats with agents.
Rejected and abondoned chats per queue.
Proactive chats offered, accepted and abandoned.
Maximum time of waiting in queue before reaching agents.
Queue abandonment rate.
Agent activity with 15 minute intervals of accepting, not accepting and writing chats.
Agents online, customers in queues, customers waiting – live view, last 15 minutes and last 24 hours.
Tag counts.
Security
Encrypted data transfer over HTTPS.
IP based restrictions (e.g. banned visitors).
Single sign-on (ADFS).
Operations log.
Password expiration and complexity checks.
Management
Supervision of agents by administrators while answering chats.
Role based access to chat solution features.
Change agents’ avatar pictures, maximum chats, name.
Add and delete agents.
Archive search.
Archive summary data.
DEEP COLLABORATION BETWEEN VA AND AGENTS
Intelligent routing from VA to the relevant agent with prioritization.
Detailed configuration of chat and handover in synchronous (live) and asynchronous (messaging) mode.
Customized flows in VA for handing chats over to agents (e.g. forms, additional questions).
Conversations in each queue are marked by anonymous name if customer is unauthenticated.
VA responses customized based on agents status and queue availability.
Chat window icon and name change in handover from VA to agents.
Proactive chat automatically disabled if queues are full.
When queues full customers can stay on-hold or leave custom message.
VA can trigger actions on the website on customer’s behalf (e.g. open windows, views).
Co-pilot mode where AI monitors what customer says and suggests answers for agents to use.
Agents can use and modify co-pilot mode suggested answers with one click.
Automatic training of the VA when agents use suggested answers.
CHAT WINDOW
Change chat window color, design, fonts, text and size with customizable CSS.
Change icon in the chat window and the chat widget bubble with text.
“Someone is typing” indicator in the chat window.
Show VA or agent name in the chat window.
Show videos, pictures, buttons, GIFs, text, emojis and links.
Carousel view in chat window to showcase products.
Mobile optimized.
Customizable forms for collecting information.
URL specific configuration.
3 VA-agent configurations: 1) VA only 2) Agent only 3) VA with handover to agent.
Proactive chat window popup.
Automatic choice of language for customer.
Ability for customers to accept terms of use and privacy.
VA response templates filled with customer specific information.
JavaScript SDK for programmatic control of the chat window.
Observable JavaScript events about actions performed in the chat window.
Compatibility with iOS and Android apps.
INTEGRATIONS
VA depolyment into social channels (Facebook, WhatsApp, Twitter, Slack).
Customer authentication via front end or back end.
Deployment with Docker containers into company’s private AWS, Azure, Google Cloud.
Deployment on-premises.
APIs for downloading chat transcripts, events and operational logs.
Integration with any 3rd party API (e.g. CRM, HR tools, e-mail/chat, RPA).
GDPR compliance mechanisms.
Speech-to-text and text-to-speech integrations with 3rd party APIs.
How We Work With Customers
01
Plan Your Virtual Assistant
Before starting to build the Virtual Assistant you should decide what is the business case. Is it for customer service? Is it for sales? Form filling? Or it can also be all of the above.
02
Existing Chat Data Analysis
If you have existing chat or email data (as .txt, .csv) we can use that to train the Natural Language Processing algorithms specifically for your Virtual Assistant. This speeds up the learning process.
03
We build the Virtual Assistant
We build the Virtual Assistant on our AlphaAI™ Platform – answering logic, language understanding, integration to your systems through our API. From you we only need intents and text answers for each intent.
04
Deployment
We deploy the Virtual Assistant where ever you want it to be – your website, mobile app, Facebook page or Slack. Your IT can relax, deployment is usually just a line of JavaScript.
05
Virtual Assistant answers customer queries
The Virtual Assistant is the first to respond to your customer queries 24/7. It understands natural language in any European language and solves customer problems without human help.
06
Modify the Virtual Assistant yourself
You can use our AlphaAI™ Platform to easily tweak the Virtual Assistant. You can change existing intents and answers, structure and train the Natural Language Understanding yourself.
07
Handover to human agents in AlphaChat™
If the Virtual Assistant gets stuck, the conversation is automatically forwarded to AlphaChat™, our live chat system. The Virtual Assistant intelligently routes customer queries to the correct human agent to handle the case.
08
Virtual Assistant learns automatically
As human agents answer customer queries, the Virtual Assistant observes the answers and automatically learns from them. This decreases the need for additional separate manual training.
09
End-to-end conversation statistics
See detailed statistics about the whole automation pipeline – from Virtual Assistant performance and conversation close rates to human agent responsiveness and customer satisfaction.
Product Video
Custom NLP already deployed in 8+ languages across Europe.
Powering more than 250,000 customer queries each month.
Serving more than 4 million users for our customers.
AlphaAI™ + AlphaChat™
AlphaAI™ Virtual Assistants
Custom AI and Natural Language Processing.
In-house developed AI. Tailor-made flexible algorithms. Inventory of industry specific intents.
Pre-trained domain specific models. Conversation Builder.
Create intents and multi-dialogue. Easy to use interface with no programming skills required. Compact design. API available.
Authentication built into Virtual Assistants. Sales automation.
Personal info for users with authentication. Personalized sales offers to users. Integration with ecommerce product feeds.
Detailed Analytics. GDPR compliant.
Learn what customers say, detect patterns, measure & improve performance. Comply with regulation.
AlphaChat™ Live Chat
Human & AI blending.
Topic prioritization in handover from bots to agents. Prioritize by URL or agent skills. Custom queues and ranking.
Automated Chatbot Training.
AI learns automatically by observing human agents answer customer queries. Decrease the need for manual AI training.
Agent Assist Co-Pilot Mode.
AI gives real time suggested answers to your agents in live chat. Agents can answer quicker and use best performing answers.
All Standard Chat Features.
Agent accounts, ticketing system, team grouping, abandonment rates, ratings, response time measurement and more.
Technology
Custom AI and NLU with an ensemble algorithms.
Confidence scores, thresholds and intent detection in all European languages.
Semantics based models built to work with very limited and small datasets.
Outperforms traditional token based models.
Works in parallel with other methods to boost text understanding accuracy.
Our Partners
AlphaBlues is AWS Select Technology Partner.
Amazon Web Services is the leading cloud service provider globally. AlphaBlues is certified AWS Technical Partner which gives us deep competence and wide insight in building and running custom solutions on the AWS infrastructure. Our team is verified by AWS technical standards and we can assist you with your digital transformation and virtual assistant needs in the cloud. Read more here.
AlphaBlues is part of the PwC Startup Collider program.
The aim is to work with the PwC team in Poland and CEE on joint customer projects. The program chose 12 technological startups from the European region and AlphaBlues was one of them. The companies were selected out of 300 applicants to partner with PwC in Poland to expand their product offering in Central Europe. Read more here.
AlphaBlues is part of the NVIDIA AI startup program.
The program nurtures dedicated and exceptional startups who are revolutionizing industries with advances in Artificial Intelligence and data science. This virtual accelerator program helps startups during critical stages of product development, prototyping, and deployment. We were also one of the few startups in the program that took part of the Deep Learning Summit in San Francisco in 2018. Read more here.
We Integrate With All Major Platforms
Meet Our Team
Indrek Vainu
CEOSales, partnerships, product management. Ex-CEO at 70-person data science company. Speaker at tech and business conferences. Harvard educated. Kitesurfing enthusiast.
Hendrik Luuk, PhD
CTOArtificial Intelligence expert with deep knowledge in NLP, machine learning and deep learning. PhD in neuroscience. Published author. Member of award winning synth-pop band.
Vladimir Rjabkov
Full Stack EngineerDeveloping front end modules of the AlphaBlues product and connecting with our API layer. Fluent in React, AngularJS and several web development frameworks.
Sten Koobas
User Analytics and AI TrainerAnalysis of user behaviour and training the virtual assistants for increased accuracy and autonomous work. Competitive football player in the National League.
Karel Koppel
Full Stack EngineerSkilled React, Node.js, Python and front end developer. Extensive experience with Docker, databases and cloud services.
Andres Tiko, PhD
Backend DeveloperPreviously worked at CERN as particle physicist in the Higgs boson project. Experience in data analysis (Pandas, Scikit-learn) and backend development.
Karl-Kristjan Kokk
DeveloperMSc in Computer Science. Skilled Python and Java developer. Travelling and sports enthusiast.
Joann Klaas
Project ManagerExperience in project delivery and customer communication. Snowboarding fan and sound designer.
Hannes Kinks
System AdministratorMSc in Computer Science with a focus on Artificial Intelligence. Experience in back end and front end development.
Dainis Rudzitis
AI TrainerSeveral years of experience in leading customer support teams. Fluent in Spanish, Italian, English, Latvian and Russian. Former radio show host & opera aficionado.
Mari Joller
AdvisorProduct and business builder with over a decade of experience in digital and mobile services. Background from Nokia and Virgin Mobile. Harvard Business School graduate.
Join Our Team
Who we are & what we do
AlphaBlues (www.alphablues.com) makes customer service better.
We’re the leading Artificial Intelligence company in the region with deep skills in Natural Language Processing and chat automation. Our product enables automating customer service chat. It is already used by major banks and telecom companies in the Baltics and Europe. We are expanding and are seeking for talented individuals like you to join.
Why you should care
It’s up to you. If you have interest in AI and front end development, this is your chance. We’ve got practical experience in deploying machine and deep learning systems to enterprise customers and know the ins and outs of our world. You get to learn from the best in a minimally bureaucratic environment with emphasis on delivering results.
What we expect from you
- Experience developing in JavaScript is required. Experience with other JS frameworks (React) is a bonus.
- Experience with Python and MongoDB is a plus.
- That you are an independent thinker that can look at the big picture and find best ways of achieving goals.
- Ability to work with minimal supervision.
- Enthusiasm and drive.
What you get in return
Satisfaction. No, for real. You get to work apply your skills in a no-nonsense environment and have major impact on the product and the millions of users that use it. In addition Tallinn city center office with great views and flexible work environment.
What next
Send CV and your github link to indrek (dot) vainu (at) alphablues (dot) com.
Follow Our Blog
How To Train Your AI: After Going Live
European Commission Language Technology Seminar
How To Train Your AI: Getting Started
Featured in Press
Subscribe to our newsletter for the latest in AI
Contact Us
