archives

Based on a report by Forbes in 2024, banking industry is looking to automate over 50% of human processes through AI by 2028. Other industries are less aggressive but 15% human process replacement is the bottom-line. Take this report as benchmark. If you fail to automate 15-50% of your processes through AI by 2028, your competitor will win and thank you for your hesitation.   Here is the mathematics of a customer service / after-sales manager : Daily Work In average, a team lead manages 5-7 after-sales staff. [Take it [...]

DeepSeek has recently outpaced OpenAI to become the top application in various performance and user satisfaction charts, signaling a shift in the landscape of AI-based applications. Public concerns about regulatory issues are growing, fostering apprehension that these might affect the output quality of DeepSeek. There is an increasing market demand for a version of DeepSeek free from censorship. For stakeholders prioritizing data integrity, hosting DeepSeek in a private environment is emerging as a viable solution. Hosting a self-owned Large Language Model (LLM) like DeepSeek involves more than just setting up [...]

DeepSeek R1 comes with a range of models for your to deploy. Here is a summary of server/cloud requirement for your ease of understanding: Model Requirement Use Case RAM CPU GPU 1.5B8GB+Most CPU launched in the last 2 years❌ Nil Basic chatbot, short writing 8B 16GB+4C or above (i5/M1/Ryzen 5)❌ Nil Basic reasoning, longer writing, coding 14B32GB+6C or above (i7/M2/Ryzen 7) ❌ Nil Advanced reasoning, coding, research 32B 32-64GB+ 8C or above (i9/M3 Pro/Ryzen 9) Metal/CUDA AI-assisted coding, Complicated problem-solving 70B64GB+ 12C or above (M4 Pro/Threadripper) High VRAM GPU Advanced [...]