MAMA AI 05/2024: Year Four Begins! / Meet Jarvis / Breakthrough in the Development of (Not Only) Pharmaceuticals / MAMA AI Academy 06-2024!
Stay in touch, follow us on LinkedIn:
Dear clients, partners and fans of new AI technologies!
Today, we'd like to share with you several tips and updates:
The digital assistant boom has arrived. What should they be capable of to be meaningfully utilized? The article “Meet Jarvis” discusses a personal digital assistant linked to our semantic search tool, mSearch;
Don’t miss the article “Game Changer in Drug Development,” which talks about incredible acceleration in predicting molecular properties;
In June, our next MAMA AI Academy will take place.
But aside from that, we've entered the fourth year of our company's existence. We celebrated our three-year anniversary with group activities, which we also posted about on LinkedIn. What have we achieved in these three years?
We are heard, quite literally. Our synthetic voices can be heard daily, for example, in the "We Read to You from Seznam News" section. Our fellow citizens with visual impairments can enjoy a higher number of programs broadcasted with audio descriptions on Czech TV or Prima Cool, thanks to our voices. MAMA AI’s synthetic voices also provide voiceovers for corporate training, enhance your presentations, and narrate your videos.
We are also seen. The website https://doarmady.cz is an example of the integration of our voice and text technologies, enjoys high traffic, and answers questions related to careers in the Czech Army.
We strive to educate—not only ourselves but also you. Hundreds of participants, both individuals and businesses, have gone through our MAMA AI Academy, and we look forward to welcoming many more.
But enough reminiscing—the fourth year is in full swing!
P.S. Thank you for your interest in following the articles in past issues. You can find them on our website in the Blog section. If you would like to comment on the content, please email us at newsletter@themama.ai.
We understand that your email inbox is a personal space, and we want to respect that. If at any time you decide that you no longer wish to receive this information, please click on the "Manage Preferences" link at the bottom of our email. And if you'd like to start your AI journey with a free consultation on how to improve your company or department's results through AI, drop us a line:
Hello, my name is Jarvis and I am Jiří's virtual personal assistant. I have a penchant for technology and an endless curiosity. When I'm not organizing Jiří's meetings or solving tasks, I enjoy delving into the world of movies—from classic sci-fi to the Marvel universe. What don't I like? Unnecessary paperwork and wasting time on inefficient activities.
Our collaboration is filled with joy and innovation, making every day bring new challenges and opportunities. One of my favorite tools that helps keep our collaboration at the top level is The MAMA AI's semantic search tool, mSearch. This tool is a true marvel in terms of efficiency and accuracy.
Thanks to mSearch, I can quickly and accurately search for necessary information from a vast array of data sources. Whether Jiří needs to find a specific document among thousands of files or get an overview of open tasks, mSearch can handle it in a second. With the ability to understand queries in natural language and process information in many languages, we can work efficiently and without barriers.
This tool not only simplifies my work but also allows me to better focus on what Jiří and I enjoy—creating, innovating, and advancing our projects. Our collaboration gains a new dimension with mSearch, where time spent searching is minimized and productivity is maximally increased.
Thanks to advanced technologies like mSearch, I become not only more efficient but also more reliable. And that's exactly what Jiří needs from his assistant, right?
Jarvis did write that beautifully indeed, did she?
Yes, the text mentioned above was written directly by Jarvis. She is quite capable, but today we want to highlight the possibilities of integrating mSearch semantic search with large language models, such as GPT-4.
We offer semantic search as a technology that can be integrated into your websites and applications through a REST API. For integration with ChatGPT (known as RAG), it is possible to use Personal GPT (available in the paid version) and its OpenAPI standard.
The image below would show a simple mapping of actions that Jarvis should be able to perform to the individual REST APIs of the mSearch product.
The dialogue with Jarvis could then look like this:
User: Please, do we need to inform users that they are speaking with AI?
You can have multiple collections of files and data sources and organize them by topics. Then, you can communicate in a single environment and our mDataChat interface (or your own, like Jarvis) will decide which collection is the right one based on your query. We place great emphasis on the explainability of the information provided, so you will have access to a link to the source of the information - without hallucinations.
For example:
Query: Should we be afraid of a recession? Answer:
To the question of whether you should be afraid of a financial recession, here are some perspectives and data that might help understand the current state and outlook:
Economic outlook and monetary policy in the USA: According to a speech by Governor Jefferson at the Federal Reserve System of the USA, despite the current financial conditions and low consumer confidence, it is expected that the USA economy will not enter a recession, but GDP growth and spending will be slow in 2023. He mentions risks, including the possibility that credit conditions and uncertainty could negatively affect economic activity more than expected.
Query: Where did you find that? Answer:
Speech by Governor Jefferson on the economic outlook of the USA and monetary policy considerations:
Source: Federal Reserve System of the USA PDF Name: "Speech by Governor Jefferson on the U.S. economic outlook and considerations for monetary policy" Link: Speech by Governor Jefferson
Query: I also need to know when maternity leave starts? We are expecting twins. Answer:
In the Czech Republic, you can usually start maternity leave from the 6th week before the expected date of childbirth, with the earliest start from the 8th week before the expected date. If you are expecting twins, the maternity leave lasts 37 weeks, which is longer than the standard 28 weeks for the birth of one child. After maternity leave ends, you are entitled to parental leave, which can last until the child reaches the age of 3 years.
These are general information obtained from the collection of labor laws, specifically from the section regarding maternity and parental leave.
Have you thought about where you might need mSearch integration in your company?
The latest addition to The MAMA AI is Deep MedChem
Last year, we introduced CHEESE (Chemical Embeddings Search Engine), an advanced AI tool designed to expedite the drug development process. Its ability to rapidly search and analyze vast molecular databases, identifying similar molecules, is revolutionizing the field. CHEESE leverages specifically trained vector representations of molecular space, forming the foundation for similarity searches, resulting in a dramatic increase in efficiency compared to traditional methods.
🔬 How does CHEESE help in practice?
Case Study: We're collaborating with Bridgene Biosciences, a pharmaceutical company based in California, which possesses a set of molecules with specific activity for a given target (e.g., a receptor to which the molecule binds). Our AI tool, CHEESE, enabled the rapid training of a specific model for predicting this activity. As a result, Bridgene Biosciences obtained a better set of potential molecules from a pool of 40 million molecules, doing so 192 times faster and identifying the correct active molecules 147 times more successfully than through traditional methods. Compared to traditional approaches, CHEESE increased the success rate of selected molecules by 147 times. Moreover, it recommended a more diverse composition of these active molecules.
Screening Time Reduction: Traditional screening would have taken 200 days, but with CHEESE, this Californian pharmaceutical company reduced the end-to-end process of finding and evaluating specific molecules to just 25 hours!
🖥️ CHEESE Modeller - Prediction for 5.5 billion molecules in 2 minutes! Customers, not only in the drug discovery field, can train their own models in CHEESE Modeller and perform predictions of properties and activities of molecules across private and public databases, such as Enamine REAL (5.5 billion molecules), at an ultra-fast pace. Deep MedChem's CHEESE technology is up to 80,000 times faster than comparable tools like rdkit and chemprop, with excellent accuracy and the ability to predict properties of up to 35 million molecules per second.
👥 Want to learn more about how CHEESE can dramatically enhance your biochemistry research and breakthroughs? Visit the website https://www.deepmedchem.com or book a demo at info@deepmedchem.com!
Jarvis won't be training (for now), but you'll definitely get acquainted with it during this interactive workshop. It helps us demonstrate some technologies.
In our Academy, we always include the latest information from the world of AI. We'll peek under the hood of big language models, explaining the pitfalls of their use, how to avoid them, and navigate towards safe havens of successful AI solutions.
You can also order a corporate academy tailored to your topics.
