Analytiq Hub转发了
I will present at the AICamp about "LLM Orchestration At Scale" - Fri Jul 26th, 5:30PM-8:30PM, at 1 Memorial Dr, Cambridge, MA. Details below.
Software architecture and design for Healthcare Data + AI
Analytiq Hub的外部链接
US,Massachusetts,Lexington,02421
Analytiq Hub转发了
I will present at the AICamp about "LLM Orchestration At Scale" - Fri Jul 26th, 5:30PM-8:30PM, at 1 Memorial Dr, Cambridge, MA. Details below.
Analytiq Hub转发了
Hire Language Model expert engineers? Or engineers willing to learn?
Check out our white paper on the Perplexity AI Approach:
Perplexity AI is at the forefront of the AI revolution - and is one of the most successful companies in the AI search space. #perplexity #search #llms #rag #genai #entrepreneurs
Analytiq Hub转发了
While GPT, Claude, Gemini, etc are solving the generic 'LLM intelligence' problem, the more interesting question, for me, is - how will this be applied to automate enterprise workflows that currently require a lot of costly human intervention. These applications could be - revenue cycle management workflows, or clinical, or research workflows, or sales, or marketing, etc. Happy to let GPT, Claude, Gemini get better at solving the 'generic AI' paradigm. The competition is pretty tough at that level, and the investments seem to benefit actually cloud vendors, who benefit from all the dev/ops and data/cluster/GPU warehouses required on the back end. While reaping the benefit in specific industry verticals is actually easier and can actually benefit from the competition between these LLMs. Our company, Analytiq Hub, specializes in full stack LLM adoption in verticals like revenue cycle management and fintech. Drop me a line if you're interested to learn more about the state of the art in LLMs applied to these industry verticals.
The improvements in GPT, Claude, Grok, Gemini, Llama in just 1 year have been amazing to watch. The big question is whether there exists a ceiling to their intelligence, and if so, how does it compare to the collective intelligence of the human species.
Compound LLM Systems - a new name for something we've been doing already: organizing multiple calls to LLMs and LLM prompts the right way, to solve specific problems in Enterprise Verticals like Healthcare, Revenue Cycle Management, and Robotics. Read the Program Manifesto for Compound LLM Systems!
"State-of-the-art AI results are increasingly obtained by compound systems with multiple components, not just monolithic models", says the Manifesto by Matei Zaharia, Omar Khattab, Michael Carbin, Jonathan Frankle, Ali Ghodsi and others. This absolutely correlates with my experience at Analytiq Hub, where we build Compound Systems for LLMs in Healthcare, Revenue Cycle Management, and Robotics. Large Language Models are generalist models. The focus of OpenAI, Anthropic, Google Gemini, etc, is to provide a general computing platform. It needs to be integrated in an Enterprise vertical - and, here, the peculiarity of integration forces you to build a Compound AI System. "...Iterating on a system design is often much faster than waiting for training runs. We believe that in any high-value application, developers will want to use every tool available to maximize AI quality, so they will use system ideas in addition to scaling."
An ultimate guide to Perplexity.AI, as it compares to Google SGE: https://lnkd.in/e4AMDtyJ With generative language models on the search engine scene, SEO (search engine optimization) has a derivative, named Generative AI Engine Optimization (GEO). ?? Marketing industry is a quick adopter of new technologies. Marketing tools today are, likely, tools with wider use in other industries tomorrow. ?? As people start using chat agents more in their work, question becomes - how do you write content that is visible in chat agents. BrightEdge has just published a content-optimization comparison for emerging AI-based search engine Perplexity, vs. Google SGE. ?? Perplexity provides a refined, conversational search experience, focusing on directness, relevance - in an ad-free environment. ?? Perplexity has concise responses from a select group of reputable sources. But it might not always provide the latest content for certain queries, unlike traditional search engines which have a more extensive index. ?? Sourcing of information: Between Perplexity and Google SGE there is overlap, but also discrepancy. ?? Reddit, Marketwatch, Yahoo, CNBC.com, Amazon.com are strategically significant source of information in Perplexity, and are not as visible in Google. The survey highlights the sectors experiencing significant shifts in information access due to the AI wave. Such shifts uncover possibilities much broader than mere SEO, providing rich territories for more in-depth analysis and foresight.