\n<\/p>\n
What is AI good for? Automating repetitive tasks for the very busy people running small businesses, reckons Berlin-based startup Synthflow<\/a>, which is announcing a $7.4 million seed round for its SME-focused no code platform for AI voice assistance. <\/p>\n Since being founded around spring last year, the startup has now banked a total of $9.1 million, underscoring ongoing investor enthusiasm for accelerating applications of generative AI.<\/p>\n The startup also claims to be approaching 1,000 customers \u2014 touting \u201cdouble-digit\u201d monthly growth rates since it stepped out of stealthy development to launch its browser-based \u201cno code\u201d tool in December 2023. That suggests there\u2019s a healthy appetite among SMEs to adopt \u2014 or at least experiment with \u2014 generative AI tools that promise easy-to-reach productivity gains.<\/p>\n The new funding will be ploughed into R&D, according to Synthflow CEO and co-founder Hakob Astabatsyan, who says the team is keen to keep stoking its early momentum by increasing product utility and broadening the scope of SMEs to which it\u2019s appealing.<\/p>\n \u201cWe have very many ideas. We know exactly what the customers need,\u201d he tells TechCrunch.<\/p>\n Astabatsyan, a serial entrepreneur with a business background, is ex-Rocket Internet. Joining him in his latest venture are his brother, Albert, who also worked with him on a prior no code startup; and Sassun Mirzakhan-Saky, who brings a software engineering background and CTO expertise to the team.<\/p>\n While Synthflow\u2019s product kicked off with English-language call handling because its largest markets are English-speaking, it has since added German and French language versions (note: these are still in beta). So dialing up its focus on the latter markets in Europe is also on the cards.<\/p>\n Call centers<\/a> were early adopters of AI voice agents, pulling on large language models\u2019 (LLMs) APIs to power systems that could answer phone calls in a human-like way \u2014 just with indefatigable energy and enthusiasm 24\/7, if not always flawless comprehension. <\/p>\n Synthflow is taking the concept in a slightly different direction, targeting service industry-focused SMEs directly, including those towards the smaller end of the category with a DIY \u201cno code\u201d offering. The goal is to offer SMEs an \u201cend-to-end\u201d experience, per Astabatsyan, who argues that the return on investment from being able to automate core tasks like appointment scheduling will be immediately obvious to its target resource-strapped businesses.<\/p>\n \u201cThe AI can do it in a more affordable manner, more reliably, and humans can do other stuff,\u201d is his concise pitch for voice assistance. <\/p>\n He gives the example of a handyman or mechanic who would typically answer the phone themselves when not working on a job \u2014 meaning they inevitably end up missing a lot of calls and losing out on some business as a result; or a dentist who employs a receptionist who works limited hours so, again, isn\u2019t always around to pick up the phone.<\/p>\n Having a tool that can handle basic customer enquiries could be a gamechanger for small businesses, Astabatsyan argues. <\/p>\n Synthflow\u2019s target being SMEs necessarily means a core focus for the startup is making AI technology accessible to non-technical users \u2014 which is why it\u2019s built a no code interface for its customers to design voice agents that fit the needs of their businesses. <\/p>\n \u201cWe wanted to try to build something simple,\u201d he explains. \u201cA no code layer on top [of AI agents] so that\u2026 business owners, business oriented people, can go and play around with this and get familiar and explore the what LLMs can do for their businesses.\u201d<\/p>\n Synthflow\u2019s interface lets customers drag and drop elements to configure voice AIs that can perform specific tasks for them \u2014 such as scheduling appointments; running through FAQs; or performing \u201cinformation extraction,\u201d such as obtaining personal information from a potential customer so a human can call them back.<\/p>\n \u201cLet\u2019s say if someone has to call, and there is a particular set of questions to be asked and particular pieces of information to be collected \u2014 especially static ones, such as, address, home, etc \u2014 AI is very good,\u201d he argues. <\/p>\n The customer can choose to configure the AI assistant so it discloses it\u2019s a robot. \u201cI think it is very good practice to disclose that it is a virtual assistant,\u201d says Astabatsyan. \u201cMy personal favourite opening is: \u2018Hello. My name is [so-and-so], right now, all our lines are busy. I apologise for that. I am the virtual assistant here at [the name of the business]. How can I help you?\u2019.\u201d<\/p>\n Another big utility for voice AIs is recognizing when a call needs to be transferred to a human agent, according to Astabatsyan. So, essentially, using AI to filter inbound calls based on complexity \u2014 with automation taking care of the simple requests which then compounds the benefit by freeing up human agents to have more time to spend on more complex customer enquiries. <\/p>\n He stresses the goal isn\u2019t to replace human jobs but rather suggests AI can help SMEs be more productive and efficient than they could otherwise be with their limited resources.<\/p>\n This is why, as well as letting customers deploy voice agents, Synthflow\u2019s system is designed to take care of post-call data entry tasks too \u2014 adding appointments to a calendar tool for instance. Building out integrations with third-party software is thus another big focus for the team.<\/p>\n \u201cThis is what the AI is so good at,\u201d he argues. \u201cBecause it can take this information [extracted from a call] and, let\u2019s say, update particular fields in particular CRM \u2014 and if you do these things at scale, on hundreds or 1,000s of calls, suddenly we\u2019re seeing this technology advantage that we saw [when businesses first adopted] computers.\u201d<\/p>\n For the voice agents, the startup is building on OpenAI\u2019s GPT LLM but also incorporating its own AI models on top \u2014 which Astabatsyan says have been trained on its own data and fine-tuned to specific customer use-cases.<\/p>\n He says it has also built its own \u201cvoice orchestration layer\u201d which converts the customer\u2019s speech into text that can then be fed to the AI model as a prompt, returning an automated answer that the system converts from text into speech the customer hears as a synthesized voice on the other end of the phone line.<\/p>\n For now, Synthflow is focused on using AI with inbound calls \u2014 which Astabatsyan suggests are the low hanging fruit for automation for resource-strapped businesses. But he hints at more sophisticated capabilities in development, with R&D fuelled by the chunky seed round.<\/p>\n One thing he mentions they\u2019re working on is a feature that will enable Synthflow\u2019s voice AIs to perform what he refers to as \u201clive actions\u201d or \u201cconnections\u201d \u2014 meaning that during a call the AI would be able to run a check on live inventory in a warehouse. Or pull in another other piece of requested info and \u201cpush it elsewhere,\u201d as he puts it.<\/p>\n He also sketches a scenario where task-focused AI voice systems would be able to expand their utility collectively. They could hand off a call to other dedicated voice AIs trained for different tasks being requested by the customer.<\/p>\n \u201cThe key here is to have focus on who your customers are. Because, depending for whom you are building this, your product is going to be very, very, very different,\u201d he adds.<\/p>\n One impact to consider is, if voice AIs and voice assistance systems live up to the productivity hype \u2014 slickly delivering on the promise of efficiently handling a whole layer of customer enquiries, including by expertly redirecting more complex stuff to the right system or human to deal with \u2014 it could end up meaning the average SME discovers they have an awful lot more work than they\u2019re able to take on. <\/p>\n \u201cI think that\u2019s an interesting question for a lot of managers and leaders to think about, right?\u201d he responds, discussing this scenario. \u201cLike, if there\u2019s so much capacity \u2014 and productivity gets unleashed \u2014 how do we channel this human resources in other sectors of the economy? Because I think this question is not answered yet, but it\u2019s a very interesting question indeed.\u201d<\/p>\n Synthflow\u2019s seed funding is led by Singular, with participation from existing investor Atlantic Labs and a number of investors in the AI space, including the founders of Krisp AI<\/a>. <\/p>\nEnd-to-end experience<\/h2>\n