Building an AI product for sales using your domain expertise

A dream come true - I ended up building software with an AI twist, which was designed for people like me 😎

I'm a big believer that in order to increase the odds of startup success, you should build for the target audience that you know a ton about (their daily work, pains, goals & desires). This will allow you to build software faster, make fewer mistakes along the way and if you're able to dogfood the software, then you'll be flying!

I've worked in B2B sales for over a decade, predominantly in SaaS selling marketing software to marketing people, so most of my experience lies in the field of martech, however apart from being an individual contributor in sales, I also led sales teams of up to 50 people so along the way I also became a somewhat of a domain expert in sales tech.

Looking at the landscape, there's just way too much software in martech including AI and it's relatively easy to build AI products in the space than some other categories I find, so I had to embark into the world of sales technology as it's a much harder category due to limited amount of structured & labeled datasets. I felt there's still so much disruption to be had, especially considering that most of the people in sales do not leverage ML & AI at scale to drive sales productivity. Sales productivity is a huge issue according to Salesforce 2023 research: https://www.salesforce.com/news/stories/sales-research-2023/

So this is where Salesforge.ai idea was born to disrupt the market. I've decided to go through Antler.co accelerator in Berlin, which is focused on zero-to-one founders, to focus fully on making your idea a reality with a cohort of 50 other founder wannabes. We even had the possibility to raise a bit of money at the end to get your idea off the ground, though the deal terms aren't that great. When it came to assembling the top 1% team to execute on the idea, I met my 2 co-founders via Y Combinator match-making platform rather than Antler. The likelihood of matching is really low, though I somehow managed to find not one, but TWO co-founders who are both technical and who knew each from the first grade at school. For the humongous scope, we have in mind, having the two of them is advantageous, so it made sense to start working on this together. In the past, they managed to build a few cool apps, which even trended in app stores, though nothing really took off. The missing piece of real success for them was a commercial co-founder a.k.a. the sales guy.

So many start-ups fail these days because they don't have the distribution figured out. My advice is to find someone who can figure out the distribution from the get-go because it doesn't matter how good your product is if no one buys it. And showing early traction to investors is paramount if you're looking to raise.

So the first thing you want to do once you have an idea, the team, and the vision of where you can take your product is to validate it with your future customers. In our case, it was doing 45 interviews with VP Sales, Head of Revenue/Sales Ops, heads of sales development etc. We used the https://www.momtestbook.com/ method when talking to our customers combined with watching a ton of Y Combinator YouTube videos. I highly recommend it as there are so many golden nuggets there in terms of what to avoid.

We figured out that our grand idea of building a multi-product AI-powered sales engine was too big to execute within a year or two, so we need a narrower scope and more focus to get the early traction we hoped for. Through the interviews we were able to validate that people would see a lot of value in leveraging an email sequencing product, which is a subset of our vision, to personalize every email at scale using generative AI. Email is one of the 3 main B2B sales channels, so it made sense to focus on this channel and then expand later down the road.

For us pulling out a credit card and preordering the product from our future customers without writing a single line of code was very important to see whether there was a demand for Salesforge. We've managed to get 4 paid pre-orders without spending a single cent on ads, so we knew we were on to something here. We ticked the box that there's a very good likelihood of product market fit.

Also as GPT4 API was released during that same timeframe, it was a no-brainer we should immediately go ahead and pursue this. Additionally, the cost of OpenAI tokens has dropped by 10x, which made it also financially feasible to scale our product. In terms of what features to focus on initially, we've used the RICE framework to help us with our decision-making process: https://www.productplan.com/glossary/rice-scoring-model/

Beyond our preorders, we wanted further validation and mingle in places/platforms where our ICPs (Ideal Customer Profile) were on a daily basis. Over a short period of time, I somehow managed to get over 200 companies to sign up for the waitlist via subreddits (use F5bot.com for social listening!), sales and start-up Slack communities (Find some on thehiveindex.com), Discord, and good old LinkedIn consistently posting value content (LinkedIn Social Selling). Over the years I've also been added to a few sales and startup founder Whatsapp groups, which I was also able to leverage by promoting that I'm building something new, which would benefit them immensely.

I must emphasize it's also important to connect to your ICPs offline through events whether, in our case, it's designed for sales professionals or startup founders. It's a totally different experience than chatting to someone on a Zoom call. Every event we went to, we were able to generate at least one additional sign up and these events really help a lot with practicing your pitch and boosting word of mouth. Sometimes we would invite our ICPs for a chat in a coffee shop to have a deeper conversation about their daily struggles. You should do this with people who are mid-level managers because they're able to share the pains that are being observed on the sales floor, the reporting struggles & tracking priorities at the mid-level, and the needs of the top-level management. Talking to ICPs offline is especially important for your technical co-founders who are not talking to users daily as they're busy shipping more code.

Using various ICP inputs, we were able to ship our lightweight MVP to validate the lift in performance we were expecting to see back in March. During that time we started to see on social media sales influencers talking significantly more about how they were leveraging specific prompts via ChatGPT to help with cold email outreach. We also saw people hacking together personalized emails via Google Sheets using https://workspace.google.com/marketplace/app/gpt_for_sheets_and_docs/677318054654 and then pushing that data for each lead into their CRM or sequencing tools in order to send a personalized message. This further helped us to validate that we're on the right track.

In order to build a wildly successful business though, it's not just about building a kickass product, but it's also about disrupting the market in other ways. Most of the sales technology market charges per seat meaning the more sales reps you have, the more seats or licenses you would need to get from places like Salesforce, and in a lot of cases this is not very CAC (Customer Acquisition Cost) efficient. We wanted to align the cost of the software with the success our customers are seeing and we're optimizing towards increasing the pipeline numbers, so we've implemented a consumption-based pricing where the more personalized emails you send, the more you would pay us. And this makes a ton of sense because usage of our product should correlate to an increase in pipeline numbers the customer would see on their end. Plus the more users are using your product, the higher the impact this has on positive retention & expansion numbers, hence charging for seats just didn't make much sense to us. I highly recommend this pricing course: https://cxl.com/institute/online-course/pricing-packaging/

Now the fact that we have a product, which needs a lot of resources to run, figuring out unit economics is very important to aim at max 30%-40% for COGS (Cost of Goods Sold) in the early days. We need to pay for multiple services to be able to run our software in the early days before bringing some parts of it in-house by developing it ourselves, so you need to find a path forward to reduce your COGS down the road.

Our next step is to do a launch on Appsumo followed by a launch on Product Hunt to achieve a milestone of 1k customers.

Join us for the ride! 🚀