Enabling
Super Human
Artificial Intelligence & Social Intelligence
We build tools to make us collectively smarter. The task of Generative AI is not to make lonely savants, but to amplify our social intelligence so that collectively we can address 21st Century challenges.
We are a Generative Artificial Intelligence & Social Intelligence platform leveraging Large Language Models (and other techniques) to reveal context, empower for sense-making & sense-breaking across any issue.
Planetary Scale Datasets
The world is now a Bermuda Triangle: We are flying blindfolded, without feeling and alone, unless we map the edge, make our maps alive with interaction and collaboration.
Our world is Volatile Uncertain, Complex and Ambiguous. In a VUCA world the edge is everywhere. Our AI platform processes vast amounts of knowledge per day to help teams find an edge unseen, to help teams gain overview, where none existed, to support breakthrough decisions in “breakdown” contexts.
Introducing:
ClimateGPT
Davos 18 January 2024: Announcing ClimateGPT: The First Open Source Foundational AI Platform Dedicated to Addressing the Impact of Climate Change
Built in partnership with the Club of Rome in collaboration with AppTek’s AI and language research scientists together and EQTY Lab advanced cryptographic framework for responsible AI models we built a model to accelerate action on Climate Change.
After over four years of research, testing, building, and fine-tuning more than 100 Large Language Models, Erasmus.AI developed the corpus of ClimateGPT from its planetary scale corpora. ClimateGPT is trained to synthesize interdisciplinary research and break silos to form a holistic understanding of the impacts of climate change .
The model benchmarks scores show 10x the efficiency on climate-specific tasks and novel cascading machine translation that recovers nearly 94% of fluency performance compared to native multilingual language models.
Example: Planetary Scale AI
Climate Change
Together with the Club of Rome we built an AI powered Human Centred Extreme Weather Dashboard, launched in Goal’s House at ’23 World Economic Forum in Davos.
Planetary Scale AI enables Financial Institutions to deliver innovate climate services, the donor community to target more locally, and cryptographically ground the “Loss and Damage” discussion. Sense-making through AI planetary scale maps , and sense-breaking to climate action.
Example: EACOP Actor Overview
Risk Map- Overview
A Client requested a risk actor overview of the East African Crude Oil Pipeline Project (EACOP)
In one of the many interactive maps we generated we showed the organisations and actors involved in the project analysed through our NER pipelines, taken from pipelines processed each day.
The project is a controversial multi-billion dollar investment with economic challenges (high lifting costs); technical challenges (the longest heated pipeline built) and ESG challenges (traversing national parks, carbon footprint), yet it presents a resource development trajectory for Uganda & Tanzania.
We use our AI to organise the world
Add your signals & visualise in elegant simple continent scale dashboards
Connecting it all.
Our pipelines enrich content with with resolved entities at web-scale
Power of Generative AI
Combining Large Language Models and Knowledge
URLs Processed Each Day
Task Specific Large Language Models (GPT) Made or Fine Tuned
Generation System
Technology Engines
We process web-scale corpora with custom pipelines capable of billions of offsets. All together we maintain bit over 40 NLP task driven engines. Some of our processing engines are shown here:
Generative AI
We have built and fine-tuned more than 100 task specific GPT (Large Language Models). We believe the ASI (Artificial Specific Intelligence) will be valuable long before AGI (Artificial General Intelligence).
Breakthrough Classifiers
State of the Art classifiers build from custom data sets exceeding general data performance by 30%. We have built continent scale geopolitical, sentiment, global economic, energy, inflation, etc. models.
Named Entity Recognizer
NER system capable of web-scale extraction and dynamically updated to keep up with our changing world, leveraging a breakthrough framework .
Aurora LLM Specializer
Integrated engine for creating specialist LLMs through fine-tuning of foundational models.
Event Detector
Large scale event tracking and localisation capable of identifying the locations and extracting 100.000s of events at a planetary scale. Events are tagged for downstream consumption.
Relative Time Extractor
Intuitively binds human relative time from unstructured text to machine readable date formats.
Sentiment & Severity
Custom AI sentiment engines, capable of fine grained sentiment (finance, risk, climate, geopolitics, etc.). Capable of detecting sentiment (85%+ ACC Score) and severity of articles.
English Translator
What news in other languages are you missing? Translating from 16 languages (latinate & non-latinate) with new ones being added regularly.
Duplication Detector
Two tier source and text duplication engines. Full-firehose, fast, document level duplicate remover that is UTM tracking aware. Our similarity search method retrieves 99.5% of pairs.
Readability
Determines how difficult a text is to read as a proxy for engagement potential. Up to 9 different literature backed metrics.
Summarizer
Extractive and abstractive summarization for high information density in reporting and dashboards.
Labeling Toolset
Standalone application for data labeling with a smart label selection engine and three different labeling styles supported. Full text, text pieces and relations.
Context & Sentence piecer
Sentence and context extraction from articles. Enabling sentence level detection of risk, events and opportunities.
Location Disambiguator
Uniquely identifies tagged locations on multiple geographical levels. Includes area, population, alternate (local/dialects) names for most of the locations at a scale 10x larger than current knowledge graphs (WikiData).
Signal Tagger
Flagging pieces of text to belong to specific topics. Choose from our library of in-house signal models or build one your own in collaboration with us.
Topic modeler
Extracts topics from multiple texts via unsupervised document-topic and topic-word distributions.
Public Company Linker
Matches company references in text to a database of known stocks and 36 corresponding exchanges. Company names are expanded to account for discrepancy in names and ticker symbols for matching across different exchanges and locations.
Testimonials & Reviews
This is unlike anything I’ve seen before, and really adds value to macro investment strategy insight, discovery and monitoring.
Dr Leslie Teo, Chief Economist – Singapore Sovereign Wealth Fund (GIC)
For buy side investors, NewsConsole is huge. Those investors get so much research from sell side firms; ingesting this data and then analyzing it on the NewsConsole platform in combination with news from the whole of the internet, completely displaces Bloomberg and Reuters. Aggregating so many news sources, giving you text analysis of the whole of the internet, something no system can do, other than NewsConsole. It really nailed looking at big data easily. You see how news is connected, how it folds together. You get an in depth and broad view of the news at the same time, which is awesome. It’s like narrow-band and broad-band. Once you have seen NewsConsole, looking at BBC or Sky seems so boring.
Michael Lobb
I need to find information outside of the usual channels, to find anomalies, exceptions, see the stuff that I am not hearing about. I want to amplify the edges, read the things that might be newsworthy for the New York Times weeks or months later. In my world, we need constant awareness of what’s going on in the energy technology space. We need competitive intelligence at the frontiers of evolving technologies, in topical spaces like solar, electric mobility, etc. NewsConsole is the tool to do that.
Russ Conser
We struggle with multiple information, coming from all sorts of sources. Couple that with the millions of sources on the web, where you can search for information. This makes the task of finding meaningful information, information where you can base decisions on, exponentially harder. We needed an intelligent platform, found NewsConsole, and are very excited about it. By using NewsConsole, when we were searching for certain topics, we found serendipitous results. We found things that other market participants are working on, which caught our interest. We weren’t looking for it, but it was very useful for us.
Surya Rajan
“ESG Risk Management is about seeing connections. This AI platform is a sensemaking fountainhead. We see complexity- not ignore it. It has surfaced numerious material ESG exposures & enabled us to respond appropriately. Developing a visual sensemaking tool to grasp positive and negative risk issues is key to augmenting risk management”.
Erica Johnson, Head ESG Risk Standard Bank
This helps me find the edge and break out of my filter bubble. As an innovation adviser I am always following the news. I use my fair share of RSS feeds, Twitter lists and topical newsletters, but NewsConsole gives me a different angle on what is happening right now. I particularly like the daily newsletters for my curated topics. Each day I get a few news stories that I haven’t seen before from an incredibly diverse set of sources that I don’t regularly follow.
Hans de Zwart
NewsConsole has changed the way I listen for weak signals in the world’s news. As a Trader and now a Portfolio Manager, I find speed to knowledge critical in business decisions and important partner relationships. The machine intelligence at work inside NewsConsole does for me what I cannot do. It is the right tool to cast the widest net, discover the quietest signal, and learn over time from my needs for business information. I don’t know of a larger crawler that embeds the semantic search tools found in NewsConsole. Once you consider the fact that it is necessary to look at the whole of the web, then it is just a matter of math. NewsConsole does this seamlessly, quickly and accurately.
Rick Marriner
We want to look at the odd things that happen. Where could money be made, what are the new things we see? There is a broader base than what we are covering. Here the NewsConsole comes in, because it gives you ideas, instead of you having an idea and investigating it. You can play around a little, to generate those ideas. You throw in interesting concepts and see what is coming back to you. Topics that go across various sectors capture our client’s attention. And in our business, it is always about capturing the clients attention.
Jon Bergthiel
In seconds you created the mental map that took me 20 years of research to build
David Sinclair, Sinclair Lab, Harvard Medical School
What are we missing? Where is that crucial insight at the edge?
See the world anew
Collaborate with AI in context
Maps not lists
As simple as possible but not too much
Web-Scale NLP pipelines:
Putting the Edge at the Centre
NewsConsole
Montly reports
Dashboards
API Services
