Sergey Davidovich

@michelleefang Kind offer indeed. Thanks.

if you're vibe coding or building over the holidays, i want to gift one of you a 6 month subscription of claude pro to support <3 just drop a comment below. merry christmas!

Agent builders - say hello to Agentune. A free, open-source toolkit that lets you simulate, analyze, and optimize conversational AI agents - without ever exposing your data or IP. Try it out and share what it unlocked for you!

The race for LLM "cognitive core" - a few billion param model that maximally sacrifices encyclopedic knowledge for capability. It lives always-on and by default on every computer as the kernel of LLM personal computing. Its features are slowly crystalizing: - Natively multimodal text/vision/audio at both input and output. - Matryoshka-style architecture allowing a dial of capability up and down at test time. - Reasoning, also with a dial. (system 2) - Aggressively tool-using. - On-device finetuning LoRA slots for test-time training, personalization and customization. - Delegates and double checks just the right parts with the oracles in the cloud if internet is available. It doesn't know that William the Conqueror's reign ended in September 9 1087, but it vaguely recognizes the name and can look up the date. It can't recite the SHA-256 of empty string as e3b0c442..., but it can calculate it quickly should you really want it. What LLM personal computing lacks in broad world knowledge and top tier problem-solving capability it will make up in super low interaction latency (especially as multimodal matures), direct / private access to data and state, offline continuity, sovereignty ("not your weights not your brain"). i.e. many of the same reasons we like, use and buy personal computers instead of having thin clients access a cloud via remote desktop or so.

Clever: LLMs work best after doing System 2 thinking; planning & “reasoning” step-by-step about a problem. You can force System 2 with prompts (ie chain-of-thought) but it uses compute. So this paper trains AI on System 2 outputs & finds it makes initial System 1 results better.

SparkBeyond Evolves: Solving the Generative AI Operational Data Blindspot, announcing our partnership with NTT and Welcoming New CEO Avrom Gilbert linkedin.com/pulse/sparkbey… via @LinkedIn

@guy Zinman thanks for sharing. This is quite predictable (yet fascinating). 2 questions remain unaddressed: 1.The role of plugin ecosystems in building a moat. 2. Search engines (which both Google and Microsoft have) aren't replicable through open sourc…lnkd.in/dUTVh2BY

As enterprises start to realize the promise of generative AI and #largelanguagemodels (LLMs) like #chatgpt, they also discover their inherent limitations, risks. SparkBeyond has been building the bridge between LLMs and enterprise…lnkd.in/dvfqyCDW lnkd.in/dr2Q2mgz

looking forward to the conversation where I'll share thoughts on unique ways Generative-AI is combined with enterprise data. lnkd.in/dmREWHNA

was a pleasure meeting with the delegation and engage in a fascinating conversation about how Generative AI, enterprise analytics and everything in between. lnkd.in/drr5GNaF

Today’s AI and computing foundations have reached a level of maturity that enables AI-assisted root-cause analysis, ideation and ongoing driver discovery for every business key performance indicator (KPI) represented in the data. Our mission at SparkBeyo…lnkd.in/dpzH6MJg

Great post by @LubomilaJ. lnkd.in/dQcVsn4f

Automated hypothesis generation is key to accelerating breakthroughs. Great to see companies pushing the boundaries in this important area. lnkd.in/g7ycGwTE