Hakob Astabatsyan, Co‑Founder and CEO of Synthflow, is main the event of cutting-edge voice AI expertise that helps companies automate buyer calls with human-like responsiveness. Based mostly in Berlin, Hakob co-founded Synthflow in 2023 with the aim of constructing superior conversational AI accessible to corporations of all sizes.
Underneath his management, Synthflow has quickly scaled throughout Germany and the U.S., enabling purchasers in actual property, healthcare, finance, and different industries to deploy customized no-code voice brokers in simply weeks. These AI brokers deal with inbound and outbound calls, combine with 200+ apps and CRMs, and function at low value with distinctive accuracy. With fast income development and a rising international group, Synthflow is turning into a number one voice automation platform—and Hakob is on the forefront of its growth.
What impressed you to begin Synthflow, and the way did your background as a strategist and serial entrepreneur affect the corporate’s mission?
Once we began Synthflow, voice AI was nonetheless in its infancy, restricted to robotic IVRs and hard-coded scripts. However having labored throughout a number of startups and tech ventures, I might see a deeper alternative. My background as a strategist taught me to identify moments when a shift in infrastructure reshapes an business, and that’s precisely what massive language fashions did for voice.
We knew that if we might mix cutting-edge AI with real-time voice interplay we might rework buyer communication. That mission to democratize voice AI wasn’t only a tagline; it was a response to a market hole. Our north star has at all times been to construct human-like automation that’s simple to deploy, scalable, and safe.
Your mission is to “democratize entry to AI voice automation.” What does that appear like in sensible phrases for small and mid-sized companies?
Democratizing voice AI means eradicating the technical and monetary obstacles which have traditionally restricted entry to this expertise. Whether or not you are a fast-scaling startup, a mid-sized operation, or an enterprise group trying to modernize buyer engagement, our aim is to make human-like voice automation deployable in days.
This is the reason we constructed Synthflow as a no-code platform. As a substitute of counting on engineering sources or prolonged integrations, groups throughout gross sales, help, and operations can design and launch voice brokers by way of a easy browser-based interface. It’s a shift from AI being a back-office R&D challenge to a frontline enterprise device.
Are you able to stroll us by way of what it takes to construct and deploy a voice agent from scratch utilizing your instruments?
It’s remarkably intuitive. Our no-code, browser-based interface lets customers drag and drop components to design an entire name move dialog from greetings to follow-ups. You’ll be able to outline tone, fallback behaviors, and integrations together with your current methods like CRMs or calendars.
We care for every little thing behind the scenes—transcription, speech synthesis, LLMs, and name setup—so groups can deal with the expertise, not the plumbing. That’s how we earned G2’s Quickest Implementation badge for AI Brokers globally, as a result of most of our clients go from zero to dwell in underneath two weeks.
What breakthroughs allowed you to realize human-like latency of <500ms and near-perfect uptime?
Attaining sub-500ms latency was a serious technical milestone. It required a mix of edge computing, real-time streaming structure, and model-level optimization. We decoupled our speech-to-text and text-to-speech layers from the LLM inferences to make sure parallel processing wherever doable.
On the infrastructure facet, we designed for reliability from day one—leveraging redundancy, failover routing, and distributed cloud methods to keep up uptime even at excessive volumes. Immediately, we course of over 40 million calls per thirty days, and our clients profit from that battle-tested scale.
How does Synthflow’s AI keep away from hallucinations and keep inside brand-safe “guardrails” throughout conversations?
Not like general-purpose chatbots that attempt to deal with something and every little thing, our voice brokers are constructed to be goal-driven. That focus dramatically reduces the possibilities of hallucination. If an agent is designed to schedule calls, it merely gained’t reply to unrelated prompts like “inform me a joke” or “what’s the climate.” It sticks to the duty it was created for, protecting the dialog purposeful and on observe.
On high of that, our platform features a built-in data base that allows you to outline precisely what the agent ought to know—and extra importantly, what it should not. This makes it simple to maintain conversations brand-safe, correct, and related, whereas providing you with full management over how the AI handles info.
What makes customers belief an AI voice agent, and the place do issues sometimes go flawed in buyer expertise?
Belief comes down to hurry, relevance, and tone. If an AI responds shortly, understands intent, and sounds pure, persons are extra more likely to have interaction. We’ve inbuilt refined human cues like pauses and filler phrases to make conversations really feel actual.
Issues go flawed when AI tries to do an excessive amount of. That’s why our brokers deal with particular duties and escalate when wanted. It’s not about doing every little thing—it’s about doing the proper issues effectively.
That stated, we all know some persons are nonetheless skeptical. And that’s truthful—each new expertise goes by way of that section. As a pioneer in voice AI, we see it as our job not simply to construct nice tech, however to assist individuals really feel assured utilizing it.
How do you make sure that Synthflow voice brokers deal with delicate information securely whereas complying with HIPAA, SOC2, and GDPR?
Safety and compliance are core to our structure. We’re licensed underneath SOC 2 and GDPR and function with HIPAA-compliant safeguards for healthcare purchasers. This contains information encryption at relaxation and in transit, role-based entry controls, audit logs, and consent frameworks baked into name flows.
What position do people nonetheless play when Synthflow is deployed — and the place is human-AI collaboration most crucial?
People will at all times play a crucial position. Synthflow isn’t right here to switch individuals, we’re right here to make them higher at their jobs. Our AI voice brokers tackle the repetitive, routine calls so human groups can deal with what they do greatest: dealing with complicated, emotional, and high-impact conversations.
We expect that the way forward for buyer communication is sort of a relay. The AI runs the primary leg of the dialog—capturing intent, gathering particulars, dealing with routine actions. When issues get extra complicated, it passes the baton to a human with full context to allow them to decide up seamlessly. That form of human-AI collaboration creates higher experiences for either side of the dialog.
You’re based mostly in Berlin. How does the European tech ecosystem form Synthflow’s development technique, particularly round privateness and AI regulation?
Being in Europe, particularly in Berlin, has sharpened our deal with privateness, moral AI, and compliance-first improvement. European clients demand excessive requirements, and we’ve embraced that as a aggressive benefit. Our structure is designed to be data-sovereign and regulation-ready, which positions us effectively as international AI rules tighten.
Berlin additionally presents a novel expertise pool. It’s helped us develop thoughtfully whereas staying grounded in belief, which is crucial when your product talks on to clients on behalf of a model.
The place do you see the way forward for AI voice brokers heading — and what’s one main development in enterprise AI that’s presently underhyped?
The longer term is accelerating quicker than anybody predicted. Inside a number of years, I consider over half of all B2B voice interactions might be AI-driven, and we’re already seeing glimpses of AI-to-AI conversations managing whole workflows.
One underhyped development is the rise of no-code and low-code AI orchestration. Everybody talks in regards to the fashions, however the actual unlock comes when companies can deploy and iterate on AI options with out writing code. And we’re excited to be a part of this wave.
Thanks for the good interview, readers who want to be taught extra ought to go to Synthflow.