Tuesday, 5:00 pm, and you’re late for a business conference. When you’re stuck in traffic, it happens to you that you forgot to book a babysitter – and find a restaurant for the scheduled team at the end of the trip.
Karl Moritz Herman expects that instead of solving problems on your own, you will turn to Saiga, the new virtual concierge service, or many for help. Harman launched Saiga in an attempt to create a device that could “do everything”. [long as] It’s (a) legal and (b) doesn’t require a physical presence,” he explained to gaming-updates via email last week.
“Our users interact with Saiga through the app as their main interface, where they have a chat-like interface for every task. The chat has been enhanced with interactive suggestion cards to improve the experience and speed up decision making for some of our clients,” said Harman. “We do what’s important, not what’s necessary.”
Countless startups have tried to make Harman’s vision a reality. Perhaps the most infamous, Magic Assistant has promised to order, reserve or facilitate just about anything on behalf of customers texting their assistants. It took Magic years to figure out how to grow without spending a lot of capital, which eventually led to premium plans and an easy-to-use marketplace for specialized features like calendar management.
But Harman thinks where experiments like Magic and Facebook’s Gobutler and M Assistant have failed, Saiga can succeed.
“I suppose that [digital assistants] There is still a problem that deserves to be solved, and when it is solved, and what is not, will be solved satisfactorily. “Saiga has eliminated big stumbling blocks where others have failed before.”
AI. background in
Prior to founding Saiga, Herman was a consultant at McKinsey and did a brief internship with Morgan Stanley on the company’s mergers and acquisitions. He then founded Dark Blue Labs, an AI company that develops algorithms for learning from structured and unstructured data. After DeepMind acquired Deep Blue Labs in 2014, Harman joined DeepMind as a Research Scholar where he helped form the Language Research Group.
Harman was inspired by the launch of Saiga, which he described as “a combination of bad life choices and a difficult situation to start with”. In 2018, he moved from London to Germany to open DeepMind’s Berlin office, and because his wife is French, Hermann has had to file tax returns in four countries this year. There was a lot of juggling.
“Add three little kids and bureaucracy to that and you have a super new life — bad enough that I could seriously think about how to get rid of this part of our lives,” he said. Hermann. “It’s good that this is a fairly common problem. We all suffer from life administration, and automating life administration is a clear plus for humanity.”
Of course, Saiga is not a charitable enterprise. The service has multiple pricing tiers, the cheapest being 299 Euros (~$330 million) per month with additional charges for “very complex tasks”. The high barrier to entry serves both to cover costs and to purposefully shrink the target market, says Harman, making Saiga’s company truly scalable – in theory.
Saiga also focuses on certain types of tasks, usually tasks that can be processed asynchronously. For example, a service can schedule an appointment with a psychiatrist but cannot order pizza to be delivered within the hour. Herman says the goal is to strike a balance between “important tasks” and complex bureaucracy with concierge-style operations.
As for behind-the-scenes tech, Harman says Saiga is using a combination of tech to expand its customer service team. Saiga maintains a knowledge graph for each client that stores everything it knows about that person in a structured format. This, according to Herman, allows Saiga to gather all the data it needs for a particular task and only query the bits that aren’t already in the database.
Natural Language Processing allows Saiga to determine which tasks clients may need help with and whether those tasks are repeatable or predictable. Meanwhile, robotic process automation (RPA) helps automate repetitive workflows that typically require a single employee.
“Our goal is to automate only 80%. Given 20% manual labor, our product is much more viable,” explains Harman. “We use an intent parser to find what our customers need and, if possible, map that intent to an existing process, allowing our operators to work efficiently, but more importantly, let the automation be…think RPA – a version of UiPath for the consumer side. Instead of sending data from Salesforce to Oracle, our automated processes are focused on renewing your vehicle registration or ensuring your family passports are renewed before they expire.
road to success?
Saiga is currently available primarily in the UK and Germany, with only “a few dozen” early access users. But in late 2020, the company closed a €3 million (~$3.31 million) seed round led by Mosaic Ventures and Seedcamp and set its sights on “late spring” for its Series A.
“Currently at Scale-Us, we are targeting senior managers and C-level founders. We sell to both individual consumers and companies that want to present Saiga as an asset to their management teams,” said Hermann. “We plan to gradually expand this target audience in the next phase to include all households with household incomes over $100,000.”
Much of Saiga is in the pipeline, but the exact salary and benefits structure for its maintenance staff is not. Harman says they are full-time employees with additional benefits, salaries “well above the minimum wage” and proven track record.
A recent study by the Customer Service Institute found that 60% of help desk employees have experienced hostility in 2021, but Harman believes that Saiga employees are not receptive. In their opinion, since Saiga deals mainly with administrative matters, customers are more likely to talk about the slow turnaround time for orders in a bureaucratic institution than about Saiga or its support staff.
“Given our very stressful and manual onboarding process, we also have a personal relationship with each client, further reducing the risk of clients harming our operators in any way. However, we take the training and support of our customer service professionals very seriously and have built a team around them so they can do their best job in a safe environment,” added Harman.
Regarding privacy and data retention, Saiga only states in its published policy that it will “comply with applicable data protection laws”, will not use personal data for any purpose other than to provide the Service, and will not be used by customers . .allow him to store his personal data. Delete at any time. Documents not marked for deletion by the Customer will be retained “until the expiration of statutory or potential contractual warranty rights” and possibly longer, depending on applicable trade and tax laws.
“As a starting point, it is extremely important to understand our business model – the fees we charge for our services – to ensure that we have fully structured incentives for our clients. We do not send products or services to them, as they may contain a commission for Saiga, and we use their data in any form other than to train our machine learning systems and perform administrative tasks for our clients. Don’t do it,” Hermann said. “We do not sell data or sell data in aggregated form; We also do not advertise customer data. Our customer service professionals are trained in data privacy measures and all customer data is encrypted during transit and storage and stored in accordance with European servers. [relevant] Rule. ,
Harman plans to rely heavily on user feedback to tailor the service as it expands.
“By charging a (significant) service fee, we can deliver a viable product right from the start. Automation is the path to economies of scale, higher margins and lower prices, but it’s not a prerequisite for success,” he continues. I[O]Our main goal is to increase the level of automation of the experience of our customers and operators … In all scenarios, this will be a multi-billion dollar business. ,