Iñaki Ladrero

AllenAI have announced OLMoE, which is a fully open (code, data and weights) state-of-the-art #LLM leveraging sparse #MixtureofExperts (MoE), which combines the strengths of ensemble learning and neural networks to handle complex problems with diverse data patterns. OLMoE-1B-7B has 7 billion (B) parameters but uses only 1B per input token. It's pre-trained on 5 trillion tokens and this is further adapted to create OLMoE-1B-7B-Instruct. It models outperform all available models with similar active parameters, even surpassing larger ones such as Llama2-13B-Chat. #HuggingFace link 👉https://lnkd.in/edhZpKhh Smaller smarter models are really having their day. #ai #enterpriseai #smallLLMs Paper: 👉 https://lnkd.in/egrtZgfz

7

Make sure to join us today at the Graph + AI summit to learn about the latest innovation in #graphdatabases, #GraphRAG, #CoPilot, and cloud-native capabilities of TigerGraph 4 beta release. You will see how all the new enterprise features will help you tackle some of the most challenging problems in #financialcrime and #supplychain.

3

𝐋𝐋𝐌 𝐁𝐞𝐧𝐜𝐡𝐦𝐚𝐫𝐤 𝐟𝐨𝐫 𝐂𝐑𝐌 Salesforce added autonomous AI Agents to their LLM Benchmark for CRM to evaluate 19 top models on function calling, argument accuracy, agent responses, and topic classification across 4 key metrics: accuracy, cost, speed, and trust & safety. Home: https://lnkd.in/gaQ7y7jB #AI #GenAI #LLM #MLLM #VLM #Benchmark #AIAgents

2

Useful to calibrate your company positioning, investor pitchdecks, planning and KPI - SAAS is a good place to be to grow a business but it is changing... because private equity, and credit and AI "buzz-trend-hype" This is the background and shifts that need to be considered if you are trying to build a centaur or unicorn in the enterprise space. It is rare that vertical companies become uniocorns - this is generally the domain of the general purpose technologies. Note: This Report does not factor in that the AI Foundation companies as SAAS vendors although they all are. (eg Mistral, Anthropic, OpenAI, Cohere et al). Apparently none of them are taking market share from these companies but they are showing up on their expense lines as vendors to most of them. #ai #industry #analysis cc Ann K. Klein Bruce Cleveland April Dunford Raja Singh Justin O'Heir Pablo Corica John N. Harrison Akil (Ahilan) Balachandran Mitesh Kapadia Sean Moffitt Denise Tsang Nada Shepherd, MBA, BSc Ben Zlotnick

1

1 comentario

I recorded a short video in which I walk you through building a serverless AI-native multi-tenant Sales Insights app, using Nile, Modal and a small Llama. https://lnkd.in/g2fsYUvi

20

1 comentario

Are we benchmarking LLMs the wrong way? According to a recent study, we might be missing the mark. Benchmarks are essential for evaluating the performance of Large Language Models (LLMs). But are they capturing what really matters—the user experience? Researchers collected 1,863 real-world use cases from 712 participants across 23 countries to evaluate LLMs. Using the User Reported Scenarios (URS) dataset, they benchmarked 10 LLM services on their ability to satisfy user needs across 7 different intent categories. The results showed that the benchmark scores aligned well with user-reported experiences, highlighting a critical oversight in current evaluation practices: subjective scenarios. The study proposes a paradigm shift in how we evaluate LLMs, moving from predefined abilities to a user-centric perspective. By benchmarking LLMs based on authentic, diverse user needs, we can ensure that these powerful tools are truly serving their intended purpose—collaborating with and assisting users in the real world. To read more about this study and its findings, check out the paper. #LLM #ai #UX #genai #research #openai #chatgpt #artificialintelligence

13

1 comentario

Agents: The Buzzword Revolutionizing LLMs In short and sweet: Function Calls: Agents are a way for LLMs to call functions or APIs effectively. Deterministic Outputs: Agents enable LLMs to produce more deterministic results than probabilistic outputs. Real-World Interaction: Agents allow LLMs to interact seamlessly with external systems. #AI #MachineLearning #LLM #ArtificialIntelligence #AgentBasedAI #FutureofAI #Agents #GenerativeAI

6

🚀 Flying High with Technology: A Talk with Felipe Izquierdo Chadwick, CTO of Sky Airlines ✈️ In the constantly evolving world of travel and aviation, staying ahead of the curve is crucial. That's why I'm thrilled to share our latest podcast episode, where we sit down with Felipe Izquierdo Chadwick, the CTO at Sky Airlines. 🎙️ Episode Highlights: Dive deep into the transformative power of Generative AI in the airline industry. Felipe shares his wealth of experience, spanning over 15 years, and offers unique insights into how Sky Airlines is leveraging AI to enhance operational efficiency and improve essential tools for success. 🔍 What’s Inside: Felipe’s journey and pivotal role at Sky Airlines The complexities and benefits of implementing Generative AI Real-world examples and strategies that highlight the value of AI in aviation 💡 Key Quote from Felipe: “What we have realized… is that the training of the tool is complex and requires time, it requires dedication and someone to guide it, and when it acquires maturity and a certain amount of work, it begins to deliver responses that are coherent and that begin to generate value.” 📲 Listen Now: This episode is a must-listen for technology enthusiasts, industry professionals, and anyone curious about AI in aviation. You can catch this fascinating conversation on Spotify or watch the video on our YouTube channel NTT DATA LATAM: https://lnkd.in/eHDrX_hZ Don’t miss out on this opportunity to hear from one of the leading voices in the airline industry. Felipe’s strategic vision and deep understanding of aviation dynamics make this podcast not just informative but essential for anyone interested in the future of travel. Tune in now and join us as we explore the skies with Felipe and Sky Airlines. Your journey into the heart of aviation awaits! 🚀🌍 NTT DATA Europe & Latam, NTT DATA, NTT DATA, Inc. #AI #Aviation #Technology #Podcast #SkyAirlines #GenerativeAI #Innovation #TravelTech #DigitalTransformation

23

1 comentario

Small language models with hundreds of millions to under 10 billion parameters are becoming popular for their accuracy, energy efficiency, and cost-effectiveness. Read more in the full Forbes article. #ai #slm #efficiency #enterpriseai #innovation #sustainability #focusgts

4

Another step towards the future and AI assistants by Anthropic🤩 Connect LLM’s with existing data sources a lot easier with MCP. #human-to-agent-ratio https://lnkd.in/d2tMsgdS

4

The purpose of this document is to offer an end to end view on how to materialize value through data from an integration perspective. In order to cover this broad scenario, this post will be segmented into the following sections: Context & Introduction: Understanding the current landscape. Challenges and Solutions: Common challenges need to be addressed in order to succeed and securely expose data internally/externally. Native Integrations in an AI World: Next steps, takeaways and how to move forward. Congrats to Ricardo Santamaría Harboe for his contribution.

1

𝐕𝐞𝐫𝐭𝐢𝐜𝐚𝐥 𝐒𝐚𝐚𝐒: 𝐍𝐨𝐰 𝐰𝐢𝐭𝐡 𝐀𝐈 𝐈𝐧𝐬𝐢𝐝𝐞 According to Andreessen Horowitz, Vertical SaaS (VSaaS) is entering a new phase, driven by AI. This third wave of VSaaS is one of the most impactful ones yet. This shift helps VSaaS companies boost revenue per customer by automating tasks that were once too complex for traditional software. From sales to customer service and finance, AI is transforming labor efficiency: 🤖 AI is transforming labor into software. 🏋️‍♀️ Using AI to manage marketing, sales, and operations, with a 2-10x revenue increase potential. 🌍 AI is broadening the capabilities of vertical SaaS, making new markets more accessible. 💡 US labor spend dwarfs software spend — $10.5 trillion vs. $313 billion. This gap presents a huge opportunity for VSaaS companies to automate labor-intensive tasks and unlock new growth potential. More ➡️ https://lnkd.in/gGdPHzHH #AI #ResponsibleAI #GenAI #LLM #MLLM #LaaS #VSaaS #SaaS #ArtificialIntelligence #Automation #labor #Robotics #Humanoids

6

How do you market and sell to CTOs and CIOs in USA? Lets consider that you are the CMO of a company which sells an enterprise AI-Powered Workforce Management Solution (ACV > 100000 USD) to CTOs/CIOs in USA. If you have a traditional marketing team, here's what they will likely do for tactical lead generation - Prepare a nice looking whitepaper on a trending topic which they believe will cut ice with the CTO and CIO - Run multiple lead gen campaigns on email and LinkedIn to drive registrations for download - Once they start getting leads from such campaigns, they will share it with the sales team who will most likely shun them completely as they demonstrate zero buying intent. - The disgruntled sales team will give up on marketing and will start an aggressive outreach campaign using a sequence of 8-10 emails and tele calling cadences spread over a period of 1-2 months. - They might be able to get a few conversations through this activity but would have ruined any future possibility of engagement due to the nuisance value created due to the constant "push" based selling process. Not long before, I have been doing the same thing and I have realized that it no longer works...especially if you want to build sustainable engagement with your target audience which consists of CXOs. However, if my client is willing to invest and is not looking for immediate gratification, I would rather go the following route. - Map out the buying committee for each company which is in my target list - Use a platform like Apollo.io, ZoomInfo, or Clay to understand intent around my solution/similar solutions and also map out the technologies used by the company - In case I am replacing an existing workforce management solution, I can make a cluster of companies using competitive solutions and come up with a custom elevator pitch. This requires extensive collaboration between marketing and sales to develop battle cards against each competitor. - I will invite my prospects to join webinars and podcasts highlighting their pain points experienced about the competition's product they are using. (For eg, during research, if I realize that most competitive products lack a particular integration/feature/use case which is a genuine pain point, I will talk about that). - I will also orient my sales team to start talking about these pain points through their own LinkedIn profiles and outreach communication. - I will create a Slack community of prospects who are willing to discuss their challenges and discuss solutions. If you do even some of the aforementioned activities right, you will see a meaningful increase in engagement with your prospects. This will definitely require buy in and patience from your client which is a herculean effort in itself. Do let me know your successful strategies in engaging with US based CXOs as comments on this post. #b2bsaas #b2bmarketing #b2bsales #b2bmarketingstrategy #b2bsalesstrategy #linkedinmarketing #b2bcontentmarketing

8

2 comentarios

Automatic search for neural architecture: All you need to know #AI #AIio #BigData #ML #NLU #Futureofwork http://ow.ly/Ujyn30sEUK1

8

Interesting perspective from Charles Lamanna at Microsoft on agentic future - what it takes for agents to be enterprise-ready - how many agents will an enterprise have - agent use cases... including an interesting idea: share agent not document - agents vs models - cost models - future skills https://lnkd.in/gQwpEqxx VentureBeat Vernon Keenan Scott Ohlund Mark Good

12

7 comentarios