AI PMs will not be judged on their ability to vibe code or having expertise in using AI tools, they will standout when they help businesses translate their use-cases into informed decisions on AI architectural path with deep framework recommendations. Helping businesses optimize cost vs ROI , improve user experience, reducing noise and build better operating models within the defined domains and industries.
Scribbles
Notes & thoughts
Short-form writing — links, observations, and things I'm thinking about. No polish required.
Highly complex process with variable steps based on context, many possible paths, judgement calls and ambiguous inputs is a good use-case to explore agentic solution's. Example plan a wedding and alimony scenarios if it fails :P Too many permutations, contextual understanding requirements, unique user situations and unexpected inputs are good recipe for agentic solution #aiagents
A system or a process with fixed sequence of steps , deterministic outcomes, limited branching (if / then ) and no ambiguity doesn't need an agentic solution. It needs better streamlined automation with reduced redundancy and seamless integrated processes. Saves your time and money on unnecessary agentic discussions and hype #aiagents
A better planned agent reduces number of steps to find answer, this translates into reduced steps and reduced tokens. Eventually reducing the cost to find the answer to user queries #aiagents
Future of AI is being local and operationally efficient, OPENCLAW has shown the world how much personalised AI can be. Soon we will see AI sitting in your hand doing things and executing tasks for each users on daily basis. 1. A farmer sitting in Uttar Pradesh will have the ability to get daily recommendation on their crops based on geo tagged location weather data and suggesting them prices and tips and tricks to improve in their local language. 2. A corporate employee will ask its AI to go and look at government website and help getting details on ITR submission and make submission on their behalf and recommend any changes 3. A bike rider will simple ask to pull his driving license from his/her digilocker 4. A trader will ask his/her AI to place trade on behalf of daily research on stocks and inform him/her about the decision to execute trades Many more use-cases will unlock and people will be having a digital helper at their service 24x7 and 365 days up and running in their preferred language. #realworldai
5 scribbles so far