As we reach the end of 2024 it has become clear that we’ve seen large improvements in Large Language Models (LLMs) whether it is OpenAI’s logic based o1 series of models or the sometimes irreverent GROK-2 from X – with these new tools available to all new possibilities arise for artificial intelligence. We’ve also seen improvements with image generation – models like Flux 1.1 Pro showed us that National Language Process (NLP) can be used by models to better understand the context of prompts but also be used to parse prompt text into readable text on the final images. What was your favorite development in the A.I space in 2024?
“We started Straico 14-15 months ago, and the landscape has already changed drastically. We advance step-by-step with the industry,” – Straico Co-Founder Juan
In our last Ask Me Anything (AMA) of the year Straico co-founders Arturo and Juan joined Kit and Jorge to discuss what major developments could be coming for Straico int he first half of 2025. While sometime was spent looking over the past 12 – 16 months the main focus was what is coming in the more immediate future. Broadly speaking these can be defined by 3 categories – the development of A.I Agents, Experiments that focus on user retention / monthly recurring revenue (MRR) and the importance that building a thriving and vibrant community is to a product like Straico.
“The community helps shape our direction, whether through feature requests, feedback, or creative solutions like custom integrations.” – Straico Co-Founder Arturo
A.I Agents: They’re Like 007 But For Productivity
The concept of an “A.I agent” might seem confusing at first glance, what is an A.I agent? To simplify things, an A.I Agent is LLM that is able to interact with third party services via A.P.I and make decisions on when it is best to use those capabilities. In a way OpenAI’s o1-series of models might be considered an “agent” of sort because it has reasoning capabilities to evaluate its own conclusion. But for the purpose of Straico’s vision for A.I agents – it’s best to think of them as helpers who can call upon capabilities that normal LLMs do not have. When speaking with a support bot for an airline, for example, the bot might offer you alternative flight dates or recommend other services that would make travel much easier.
Our Co-Founders envision a world where A.I agents are customization – where the person designing the agents can provide a custom knowledge base for their A.I agent (see our previous blog post about RAG) which is used to retrieve answers to specific questions. Straico agents should be able to use Function Calling to enhance their utility and capabilities beyond what is currently possible. There should be a number of different ways to interact with agents that aren’t just limited to text, including real time voice conversations.
In summary, A.I agents are LLMs with a end-user provided knowledge-base that is based upon documents and references provided by the end user. This is handled through RAG technology combined with a custom prompt that outlines the agents’ personality, scope, and functionality. By interacting with users in real-time using voice or text combined with function calling to third party applications provide a solid foundation for helpful A.I chatbots.
The Art of Experimentation
Creating an innovative and reliable product is not achieved through a single breakthrough or even a small number of improvements. Enhancing Straico involves collecting community feedback, implementing cutting-edge advancements in Artificial Intelligence, and understanding how these elements collectively improve the user experience.
Improvements such as the introduction of Smart Mode, the release of our unified user interface, and the addition of engaging tools like Side by Side mode have significantly enhanced Straico’s usability. As we look towards 2025, initiatives such as integrating Retrieval-Augmented Generation (RAG) into our main web UI, and developing comprehensive video tutorials and written documentation, highlight our commitment to continuous enhancement. Additionally, our experimentation with the existing Coin System demonstrates our dedication to evolving our platform based on user needs and feedback.
The data we gather from these experiments and enhancements, alongside community feedback, industry trends, and internal testing, plays a crucial role in shaping Straico’s future. A recent AMA session brought attention to a privacy concern—previous API versions showed a small snippet of prompt history in transaction logs. However, due to our privacy-conscious community, this issue was identified and resolved, leading to the removal of these snippets and reinforcing our focus on user privacy and data protection.
“After our new UI release, we saw free user retention improve significantly.”
Community at the Core: How Your Insights Are Improving Straico
Our community sits at the very heart of Straico, without people our product we wouldn’t have been able to deliver all the improvements we did in 2024 nor would we have such a clear vision for the first half of 2025. We’ve seen thousands of new users join us in our journey throughout 2024 and several hundred new faces in our official discord in the last couple months alone.
2024 has seen some amazing community discussions and contributions. Several members of our discord have shown off impressive demos that integrate Straico in unique ways, such as the Straico Assistant Chrome extension by Hossam Gouda that allows any user to chat with our models without needing to load our site.
Many users have requested advanced features like text to video, the ability to generate infographics among other suggestions. As we look at the roadmap for 2025 many of these proposed ideas will soon become a reality – January 2025 will see the availability of Ideogram, shareable templates and an API update. March 2025 will see us release the ability to perform prompt chains, the use of RAG for agent context as well as the ability to deploy agents within the Web UI. By May 2025 multi-agent conversations, built-in-memory and custom agent creation / sharing will all be possible.
If you want to check out our roadmap, click the image below:
