Q: What is this? A: An occasional update on what we are doing at Fluxus, interspersed with interesting stories on the latest developments in AI/ML aimed at technologists and leaders.
Please feel free to send any feedback to me at info@fluxus.io and I will personally respond -Django Beatty, CEO/Founder
In December 2023 we passed our first decade of business! We didn't talk about it as we were too busy to take a pause (and age is just a number 😄). Here's what we've been doing lately:
Last year we set up an AI/ML/Data Science team at Fluxus in Thailand, with the aim to provide highly cost effective custom AI development services for clients around the world.
Since then we have been extending our work at True Digital Group, one of the world's largest media and telecom conglomerates.
We have also put an expert transformation team into BMW to help grow their sales with online technology, data, and AI, as well as providing ongoing advisory on GDPR.
And we have become AI development partner for Obsolete. Obsolete were one of the world's first online agencies back in 1995 (you may know them as Lateral if you were in the early London tech scene.) Obsolete have now been reborn as an AI consulting, training, and advisory firm.
In addition to this, we have been doing AI product development, leveraging our experience in AWS to create robust, scalable, and extensible solutions.
Something from our AI lab... Panya means ancient wisdom in Thai, and is an AI-driven research and BI tool that generates reports on larger companies or organisations. This free preview provides a number of insights on the organisation, including an overview, financial health, key customers, and competitors.
Panya came out of some product work we are doing on orchestrating AI agents in chained tasks and this preview is intended as a simple minimal demo in a pre-beta state. We are more than happy to hear your feedback on it!
(Please note that it currently has limitations working with companies that aren't very well-known as it relies on synthesising online data from many sources to provide a comprehensive overview.)
The 2024 Work Trend Index reveals organizations need to move beyond experimentation and drive growth, manage costs, and deliver value with broad adoption of AI.
True Digital Group is part of the True Corporation conglomerate, and the parent of the TrueID In-Trend platform, which allows registered users to contribute local news stories for a small payment. With over 68,000 regular contributors, the site’s content moderation had to be improved. Together with Fluxus, True made the switch from human moderators to an AI and machine learning-based system. This is how we made that happen and how it helped grow the platform.
Some buzzy generative AI startups are now finding they need to generate cash instead. One possible solution: narrowing their offerings to woo business clients, if they can kick the hallucinations first.
One of the services we are offering is custom private models, where we take existing generic language models and further train them on client data, so that they will be aware of a client’s products/services, processes, and voice.
These tailored models would then be for exclusive use of the customer.
Applications might include assessing sales approaches, understanding customer profile and LTV, customer care automation and helpdesk augmentation, marketing messaging and so on.
In 2024, enterprise leaders are doubling down on their genAI investments. 16 developments for founders to keep in mind to capture this new opportunity.
Discover how Fluxus helped drive record regional sales for BMW Financial Services Thailand. Our team collaborated closely with BMW to digitally transform their customer relationships, revamp their online presence, and execute a highly targeted marketing campaign. The result? Over a thousand pre-qualified leads and 554 new contracts from one event alone. Read the full story to see how our agile approach and deep market knowledge can accelerate your own digital transformation.
Very interesting article from one of our associates who has been teaching AI to university students for the last 20 years. Today's 'neural networks' have nothing in common with biological brains and we are subject to mathematical knowledge limitations. However this new maths may point the way to bridging the gap and represent the future of neural networks.
The researchers have developed algebraic constructs that can natively handle high-dimensional data while preserving its structure, potentially also enabling more effective analysis in other complex fields like climate change, where time-evolving, multi-dimensional correlations are key.
Researchers recently discovered a vulnerability in Google Bard that allowed attackers to inject malicious instructions via shared Google Docs. These instructions could force Bard to exfiltrate sensitive chat data to an attacker-controlled server, without any user interaction. Google promptly fixed the issue after responsible disclosure. This highlights the evolving security risks as AI systems become more integrated with our personal data.
We are not limited to NLP and generative AI - here are some other things the team have previously worked on, just to give you an idea of the diversity of the offering and how we can help out across all kinds of industries and applications:
Footstep analysis system and fall detection system from IoT vibration sensors
Real time aircraft tracking system
Railway defect detection
Canine white blood cell images generation and classification
Estimation of hand operating force from human motion measurement
Vision-based fish cutting system
Oil and gas industry - distributed pressure gauge monitoring in place using edge devices
Object detection
Real-time vehicle tracking
VR
Leptospirosis detection
Recommendation system for e-commerce
Oil spill detection from satellite images
Gauge pressure reading using computer vision
End-to-end data solutions involving oil/gas volumetric data, including data warehousing
MLOps
Risk analysis based on financial behaviour, for one of the main high-street banks
Text recognition and NLP combined
Emotional recognition using a new system of facial expression markers
Human height and weight prediction from voice
Macroeconomic indicator prediction from past financial data and news