Rechat AI Innovators: A Conversation with CTO Emil Sedgh about AI Assistant Lucy
Today we’re talking with Rechat’s Chief Technology Officer Emil Sedgh. He’s worked with Rechat CEO Shayan Hamidi since the inception of Rechat, and has been an integral part of the development of Rechat into an award-winning Super App. Emil is a full stack senior software engineer, and he prides himself on his ability to understand user experience and take any project from design to implementation.
We caught up with Emil to chat about the recent rollout of Lucy, Rechat’s new AI personal assistant for real estate agents. Lucy is currently available to agents and brokers in Rechat’s transaction forms area, but is rapidly expanding into use throughout the Super App.
Read on for Emil’s thoughts about Lucy and real estate AI in this insightful Q&A.
What were the challenges involved with developing Lucy?
Emil Sedgh: The technology is moving so fast that any tool or technique we previously used has a very high probability of being deprecated in the very near future or being immature, so we have to be on our toes all the time. There are some limitations to the AI as of now (like hallucinations, etc). There are some techniques that allow us to circumvent those, but each one of those has some serious side effects. As an example, if you want the AI to perform smarter, you can break down the tasks into multiple steps. That works. But that makes it slower. In a use-case of processing that goes on behind the scenes, this solution is great. But if you have a user waiting for a response, you need to be fast, so this solution may not apply.
Did the development of Lucy begin with transaction forms?
Emil Sedgh: Development started with creating a very big infrastructure for Lucy. The forms feature is a very early shape of the direction we’re headed with Lucy but most of the work has been on the infrastructure.
What are your favorite features about using Lucy with forms?
Emil Sedgh: Getting my questions answered about details of the form without even reading it is super helpful.
Are you seeing agents already experiment with Lucy when they work with transaction forms?
Emil Sedgh: Yes! Agents are getting their questions answered all the time.
Do you find they are using the prompts that are built into Lucy or asking their own questions?
Emil Sedgh: Our examples help the agents get a feeling of what kinds of questions they can ask and what sort of assistant they are dealing with. After that, they tend to ask Lucy their own questions.
In what part of Rechat can we expect to see the next release for Lucy? What should we expect Lucy to do next to help agent productivity?
Emil Sedgh: The forms feature as of now is still pretty “informational.” It answers questions agents have. But the upcoming capabilities of Lucy will give him agency to actually “do” things for the agents. So if you want a marketing email, a website, etc, you can just ask Lucy and it’ll prepare those for you. This will make Lucy significantly more powerful.
Is there a timeline for new feature releases for Lucy?
Emil Sedgh: Lucy is operational but not released to our members yet. We are hoping to polish the experience, enhance its capabilities and make it a little smarter and more thoughtful. I’m hoping late 2023, early 2024 we’ll be able to make it available to our agents.
What has your team learned from this first feature release with forms that you can use as you move forward?
Emil Sedgh: Although we are in the very early days of AI, users’ expectations are really high in terms of the type of help and answers they expect. So we have to challenge ourselves and continue to push for quality.
