Joscha Koepke, Head of Product at Connectly – Interview Series

Joscha Koepke, is the Head of Product at Connectly, a code-free platform that lets you create campaigns and interactive bots to easily automate two-way conversations – to both leads and loyal customers – at scale.

Connectly has a vision is to create the AI-powered infrastructure of the future to make this as easy as possible; enabling businesses to operate the full flywheel – marketing, sales, transactions, customer experience -all within the customer’s thread of choice.

What initially attracted you to computer science and AI?

My path to the tech industry, and product management in particular, took a bit of an unconventional turn. Initially, I was immersed in the world of product development within the hair care sector. There was something incredibly rewarding about diving deep into human needs and mastering the art of user-centric design. However, I found myself yearning for a more dynamic environment, which led me to embark on a nearly decade-long journey at Google.

Starting in sales, I gained invaluable insights into customer pain points and the intricacies of building relationships. This experience laid the groundwork for my transition into a product role within the Ads organization.

I’ve always believed that life and career are punctuated by moments that challenge us to choose between safety and risk. Joining Connectly was my leap into the unknown—a decision fueled by the thrill of building something from the ground up. As the fourth employee, armed with little more than a vision and a few PowerPoint slides, I stepped into a journey of immense growth and learning.

You described the challenges of moving to the earliest stages of a startup as being akin to a constant roller coaster traveling at hyperdrive while being thrashed around by a big wave. What was it that attracted you to the hectic lifestyle of startup life?

What initially attracted me to the startup life was the accelerated pace, continuous demands, and the constrained resources and data for decision-making. Recognizing the startup journey as a marathon filled with sprints has been crucial in my experience. There were days I felt on top of the world and others where the path forward was murky at best. Embracing this dynamic, understanding that each day brings its own set of challenges and opportunities has been key. This perspective helped me and Connectly as a whole navigate through uncertain times, especially when we were refining our product-market fit, allowing us to adapt and evolve swiftly. Startups usually do not die because of competition but because of a lack of resilience and focus.

In this same article you also referenced the importance of first principles thinking, what are some ways that first principles are applied at Connectly?

Product Principle 1: Action Provides Information: It’s easy to sit in a room and intellectualize about possible learnings and assumptions (we have spent days doing this), and it’s a trap Product Managers easily fall into. Every experiment we have launched with real customers, regardless of how small, has provided us with more learnings than any theoretical approach. Before any test or launch, we define a set of hypotheses, and every action becomes an experiment. The results, whether they are a success or failure, provide clearer direction for our next steps.

Putting our “First Principles” lens on, we seek foundational data instead of making assumptions or relying on second-hand knowledge. By taking action, we gather this primary information, allowing us to base decisions on direct experience rather than hearsay or theory.

Product Principle 2: Think Big, Start Small: By starting small, with a singular feature or a specific market niche, we can validate our assumptions, refine our approach, and progressively inch closer to our overarching vision. At Connectly, our mission is to automate every sales conversation with AI. But, to avoid being overwhelmed and to ensure consistent forward momentum, we narrow down our focus.

Putting our “First Principles” lens on, we strip Connectly’s product down to its fundamental purpose, which is to solve the needs of a user who requires help in their purchase decision. While our end goal is vast (automating all sales conversations), we begin by addressing a singular, fundamental need (advising a customer interested in buying a white t-shirt). As we validate and learn from this, we can expand, but always stay grounded in the essential truths we’ve uncovered.

Product Principle 3: Simplify Our Customers’ Lives: Our commitment at Connectly is to have every feature we roll out to make interactions more effortless for businesses and end consumers. By constantly asking, “How does this simplify the user journey?” we ensure that our product stays aligned with the needs and desires of our users.

Putting our “First Principles” lens on, we’re prompted to ask: “What is the most direct way we can provide this value?” For example, we noticed an increased usage of our campaign tool via API. We previously directed customers to our API documentation. However, after employing first principles to identify the most straightforward value-add, we integrated a ‘copy’ button within our campaign flowbuilder. This allows users to instantly copy a pre-configured cURL, enabling immediate action without delving into our API documentation first.

Can you briefly discuss how Retrieval-Augmented Generation (RAG) is being used at Connectly to design a conversational sales bot?

Retrieval-Augmented Generation (RAG) is one part of our AI architecture. We have also trained our own models and embeddings that perform better than the benchmark, especially in the multilingual product embedding space. A great conversational sales bot needs access to almost real-time information like stock availability and price updates. This is where RAG comes into play to connect our bot with product catalogs.

How have AI chatbots evolved to better understand and adapt to human language nuances, transforming from mere tools to active partners in digital experiences?

As Connectly’s Head of Product, I’ve observed the transformation of chatbots into proactive, learning agents. These chatbots autonomously perform actions, replicate human decision-making, and learn from interactions. A key development is their enhanced ability to understand human language with greater nuance, thanks to advances in natural language processing and large language models.

Can you elaborate on the ways AI chatbots are becoming integral team members across various industries and the implications for human-AI interaction dynamics?

AI chatbots that are becoming integral team members across industries go beyond mere query responses; they adapt their interactions and have the ability to re-prompt during sales conversations, where the chatbot uses its extensive product catalog knowledge to guide customers through targeted follow-up questions. This method narrows down options effectively, finding the ideal product for the customer’s specific needs. For example, if a customer is looking for a t-shirt, the chatbot would prompt questions like “Is this t-shirt for you or someone else? Do you have a price range and color preference in mind?” mirroring a skilled salesperson’s approach by following a sequence of ideas or steps to solve complex recommendations and answer questions.

How do you see the role of AI chatbots evolving in terms of autonomous decision-making and learning from human interactions?

Mature AI chatbots will be able to seamlessly transition from selling and recommending products in Brazil or California (environmental adaptation) to making decisions on which product SKUs to recommend (decision-making) and improve the sales conversation after receiving pushback or not being able to sell a product (learning).

At Connectly, we’re exploring game-like scenarios for learning, where chatbots engage in simulated conversations to improve their strategies, akin to role-plays in sales training. This continuous learning and adaptation are crucial for successfully deploying and distributing chatbots.

What is your vision for the future of chatbots?

Looking ahead, we anticipate improvements in emotional intelligence, particularly in a multimodal world where AI can interpret emotions from tone and facial expressions. With ongoing hardware innovations and advancement topics like memory from an AI will become a lot more achievable in the near future. This will truly allow for personal shopping AI bots that know your style and preferences over time and can remember what you like and dislike.

For readers who are interested, could you give a brief summary of what Connectly offers with its AI-Powered Marketing Automation Flowbuilder?

Connectly uses proprietary AI models to help businesses automate their communications with their customers and see their products from any messaging platform. Its code-free platform lets you create campaigns and interactive mini-bots to automate two-way conversations – to both leads and loyal customers – easily and at scale.

Thank you for the great interview, readers who wish to learn more should visit Connectly.