In the ever-evolving landscape of artificial intelligence and automation, staying connected with the community and iterating based on real user feedback are paramount. Straico, a recently hosted an insightful AMA (Ask Me Anything) session on our Discord server. This session brought together key members of the Strico team and esteemed guests to discuss the current state, challenges, and future directions of their innovative solutions, particularly focusing on the Retrieval-Augmented Generation (RAG) API. Joining Arturo and Kit from Straico were RoboRiley and Jrinaldi, two active members of our community who specialize in automation.
In this blog post, we delve into the highlights of this engaging conversation, unpacking the contributions, concerns, and key insights shared by the participants.
Riley is an automation enthusiast with a keen interest in leveraging AI tools for educational technology (ed-tech), and JRinaldi, a software engineer proficient in integrating AI into programming and home automation. Their diverse backgrounds provided a rich perspective on how Strico’s solutions are being utilized in various real-world scenarios.
Introducing the Guests: Riley and Jrinaldi
Riley began by expressing his passion for automation and AI and he highlighted his involvement in ed-tech, juggling a day job in the field while also managing a startup focused on similar endeavors. Riley’s dual roles underscored the practical applications of Strico’s tools and API in both every day tasks as well as more demanding technical tasks like automation.
JRinaldi shared his experience as a software engineer who extensively uses Straico’s API to enhance his work processes. From programming with AI as a co-programmer to implementing Striaco into his Smart home automation, his projects exemplify the versatile applications of Straico’s prompt completion AI as well as our upcoming RAG based Agents API. His hands-on experience provided valuable insights into the technical aspects and user experience of integrating advanced AI solutions into a Smart Home to provide useful enhancements to everyday life such as providing him with accurate transit directions.
RAGs Aren’t Just For Doing The Dishes
The heart of the AMA session revolved around Retrieval-Augmented Generation (RAG) – an advanced way for Large Language Models to use documents, books and other written sources as their knowledge base. Far more advanced than simply attaching a single document in a single chat, something Straico already supports. RAG can breakdown multiple documents into small chunks and understand the context of full sentences. This not only leads to more reliable factual information but can even provide citations about where the knowledge the LLM is drawing upon came from.
Riley praised the alpha RAG system for its robustness and usability. He emphasized the flexibility and enhanced embedding models that Strico has integrated, which significantly improve the API’s performance. Riley shared his practical experience, noting that the RAG API enables him to deploy automation tasks efficiently— His endorsement highlighted the API’s potential to transform how users manage and interact with data, particularly in automation-driven environments.
“I have not gotten to test the brand-new Just announced today innovations in the API, but I can say My experience is like first of all like with an API you need a ton of stability Especially if you’re using it for automations, right? It’s got to respond every single time…” – Quote from RoboRiley about the upcoming RAG API.
When Developers and Users Collaborate Magic Happens
Arturo, one of Straico’s two co-founders acknowledging the fundamental role Riley has played in testing and suggestion enhancements for our RAG API. He highlighted that many of Riley’s ideas were being implemented, steering the RAG API towards greater functionality and robust options for non-coders to use the RAG API in the future.This underscores how user contributions and feedback to our development result in collaborative efforts that enhance Straico’s features and usability.
One concern raised was the lower-than-expected traction in the alpha test group for the RAG API. It might sound puzzling at first why adoption for such an amazing addition that enhances LLM discussions hasn’t seen more use. It might be due to the fact that right now RAG can only be used by programmers, wider usage may come when a web UI that is more user friendly is available.
“We haven’t received the traction in the alpha test group that we were expecting…[maybe] it’s because it’s too technical” – Quote from Kit discussing the lower than expected traction of the private RAG alpha test.
Another technical limitation highlighted was the RAG API’s current support limited to PDF files. Users requiring functionality with other formats or resources, such as website submissions have found this restrictive. This limitation curtailed the API’s usability in cases where varied document types are needed. Arturo acknowledged these constraints, mentioning ongoing efforts to expand file support and enhance the API’s versatility.
While discussing automation Riley made a great point that reliability and response time are crucial when building them. When testing an automation it can sometimes take between 100 and 150 attempts to really gauge the level of reliability. When a component of an integration, such an API endpoint, fails having good error correction is vital. Riley pointed out that the end user wouldn’t want the entire process to hang just because an error was returned. One point that the group discussed was the wait times for Large Language Models to generate responses. Sometimes responses can take several seconds to receive, which for shorter or repeated automation could become an annoyance.
Learning Languages With Straico?
Learning a brand new language is no small feat it requires daily commitment and can take years to become proficient in another tongue. It is estimated that 43% of the population – or 3,3 billion people – speak at least two languages and with the popularity of language learning apps on mobile devices the appeal is obvious.
Jrinaldi shared with us that’s been using a combination of home automation and Straico to help accelerate his learning journey. This novel approach allows his Smart Home to translate things like weather alerts from English to Japanese at random intervals, providing him with an unexpected opportunity to flex his developing linguistic skills. When asked if there was a “dream automation” that he’d like to accomplish; Jrinaldi told us he’d love to be able to automate Smart Home notifications for the local transit system. This would allow him to know whether his train has been delayed or to take alternate routes if crowds are to be expected.
Wrapping Up
As we conclude our look at Straico’s second AMA, we extend our heartfelt gratitude to our esteemed guests, Riley and JRinaldi, for sharing their invaluable insights and experiences. Riley’s passion for automation in educational technology and his dedication to innovation, alongside JRinaldi’s expertise in integrating AI into software engineering and home automation, have provided our community with a deeper understanding of the versatile applications of Straico’s RAG API. We also want to thank our dedicated readers and community members for their continuous support and engagement. Throughout the session, we explored the advanced capabilities of Retrieval-Augmented Generation, discussed the importance of user-friendly configurations, and highlighted the collaborative efforts between developers and users that propel our advancements. Your participation and feedback are integral to our mission of building a connected and innovative community. We look forward to future AMA sessions and blog posts centered on what can be achieved with generative AI. If you’d like to join our community, our Discord is just a single click away.
