In this post, we chat with Mert Deveci, Founder of Godmode. Godmode is a startup building a go-to-market operating system for modern sales and marketing organizations that need to build relationships with their customers and prospective customers across Linkedin, email, Whatsapp and Slack.
Mert and the team at Godmode evaluated several frameworks for developing LLM-Powered Applications in their quest to make their vision of building AI coworkers for their customers’ businesses a reality. They chose to put the open source Griptape Framework to work as a solution for the challenges that they encountered. Let’s jump in and learn more from Mert himself.
We began our conversation by asking Mert to give us some background on Godmode, how he came to found the company, and what problem he and the rest of the team are trying to help their customers solve.
Why was Godmode created?
Mert told us that the story of Godmode began when he was working in financial services sales where his role involved establishing relationships with C-level executives at large public and private companies. Building these relationships required significant knowledge of each prospective customer and Mert found himself researching each company at a detailed level, understanding their financial performance, and creating comprehensive research reports. The insights in these reports were vital in crafting compelling outreach to individuals in Mert’s target customers.
Mert found that this approach was simultaneously highly effective and extremely time consuming. The time taken to complete the research phase meant that he was only able to complete two to three research reports during a week, limiting the number of customers that he was able to engage with.
When Mert became aware of GPT AI models, he wondered whether he could give these models access to the knowledge and real time information sources that he used in his customer research and have the models automate this research task to make him more effective. Like most with the founder-mindset, he also wondered whether this solution might be generally applicable and therefore represent an opportunity to create a product.
Godmode was born to make B2B enterprise sales teams omnipotent through the provision of actionable intelligence about their customers. Godmode aims to solve two key problems for their customers:
- Remove pipeline constraints providing an autopilot for pipeline generation, avoiding AI slop with well thought out and relevant research that leads to effective personalization.
- Minimize unproductive outreach by providing sales teams with daily guidance on prospects to focus on in order to convert them to customers in the most optimal way.
Challenges Godmode overcame
Next, we focused on some of the challenges that Godmode faced during the early stage of their product development process. At this stage, Godmode were experimenting with a variety of different approaches. In doing so, they identified two primary issues that they needed solutions for:
- Large Language Models are non-deterministic, meaning that they do not always give the same output in response to the same input, or a consistent answer to the same question. While this can prove useful, or simply isn’t an issue in casual usage or for chat interfaces, business customers need reliability, predictability, and outputs in a specific range to meet their quality needs.
- In experimenting with several other frameworks focused on developing LLM-Powered applications, such as LangChain, they found it very difficult to generate reliable outputs. This led the team to resort to creating rigorous and overly long prompts, which still resulted in probabilistic output. As engineers, they hated prompting and wanted a way to program LLMs and Agents rather than to prompt them.
Godmode & Griptape Framework
Now we understand the issues that Mert and team faced, let’s talk briefly about the solutions that they built and their plans for the future.
Using Griptape Framework, the team at Godmode have built more than 10 pipelines and workflows to identify leads, research them, generate outreach, and find signals all over the web and social media on a highly reliable basis. With Griptape they are able to control the outputs and show the trails to their end users, allowing them to audit and improve responses for better human and AI collaboration.
Looking forward, Mert & Godmode plan to continue to develop additional workflows with more complex capabilities that will deliver more and more valuable insights to their customers. They also plan to connect Griptape Framework into more tools and applications so AI is embedded into their customers' workflow, and accessible by using a natural language prompt or at the click of a button, becoming fully usable and useful for their customers.