Suggestions
Jamie Cuffe
Experienced tech professional and co-founder
Jamie Cuffe is a prominent figure in the tech industry, currently serving as the Head of New Products and Self-Serve at Retool, a platform designed to simplify the creation of internal applications. He has been with Retool since March 2020 and has played a significant role in integrating artificial intelligence into the company's offerings, including the recent launch of Retool AI, which aims to enhance application development speed and efficiency for businesses.23
Before joining Retool, Jamie co-founded Cheer, a company that was later acquired by Retool. His professional journey also includes experience as a Data Scientist at Sequoia Capital, where he developed his analytical skills to identify promising startups. Additionally, he has worked as a Technical Project Manager at Salesforce and as an Engineering/Operations Engineer at TimescaleDB, gaining expertise in project management, engineering, and operations.12
Jamie holds a Bachelor of Science in Computer Science from Princeton University, graduating Summa Cum Laude. His academic background and diverse professional experiences have equipped him with a strong foundation in technology and innovation, making him a valuable asset to Retool and the broader tech community.1
Highlights
The blank canvas problem is the top issue in onboarding. For horizontal SaaS like Retool, Notion, or Figma, users repeatedly say: “I can build anything! Great… so what do I build?” AI can fix this 👇
The solution to the blank canvas problem today can be anything from a structured onboarding tutorial to templates that get users to an app in 1-click.
But most developers want to jump straight into building. In the early days of Retool, we’d hop into Intercom and build alongside every new user—valuable, but not scalable.
What if we could give each user an engineer to pair program with them?
Enter Workflows AI ✨
We recently launched Retool Workflows, a visual IDE for automation. Workflows supports a ton of use cases, from bespoke billing scripts to outage alerts, but they all start with a blank canvas.
So over a hackathon, Yuke built an agent that creates a full workflow from a simple prompt using OpenAI's GPT-4. Think of an idea, press generate, and let AI walk you through the core concepts you need to get your first workflow into production.
Our learnings: ⚙ Users like examples that are as relevant to their use case as possible ⚡ People hate tooltip overviews but streaming an AI walkthrough captures the imagination 🤖 AI won’t solve all your problems in 1 prompt, yet. Get users started then let them iterate fast
There’s been a lot of hype around LLMs but very few AI apps are delivering ROI today. It’s time to make good on the promise of AI. Here’s what we learned building Retool AI 👇
-
The only way to know whether LLMs can really solve your use case is to try it. Experiment with your own prompts and iterate fast.
-
Customers told us over and over that every vendor is selling chatbots but no one has what they need. To get usable results, you need to customize with your own logic and prompts.
-
Every point solution is adding AI, but for AI to really work you need data from across your apps. A lot of the data you need to make chatbots work is locked inside your internal systems.
-
For large customers that stand to gain the most from AI, data security is the biggest threat keeping AI apps from getting to production. Using open-source LLMs like Llama2 and enterprise agreements with Azure are the current solutions.
AI demos have been a lot of fun, but it’s time to deliver some real value.
