Jimmy Sam

@gautham_city_ when ur in Van feel free to visit our office

@__JimmySam__ nah Calgary van next

back in maple syrup land (temporarily)

peak productivity

my proudest moment of 2026 is getting my 40+ year old boss unironically saying cooked

Get ready for @BeachHackscsulb + @UBCBizTech's ProduHacks with our pre-hackathon workshop! 🚀  Ryan Tran will show you how to build intelligent, autonomous AI Agents using our open-source marketplace @Agentverse_ai, agentic LLM ASI:One and high-tech @Fetch_ai frameworks. This session will set you up for both hacks with a showcase of live agents that will answer your queries on the spot. 📅 Don’t miss it: 18th of March, 7-8PM PST 👉 Join here: meet.google.com/bau-mbse-gpx @WSana81

A lot of the successful people I know massively prefer phone calls to emails or other forms of communication. Why spend days going back and forth when you can achieve the same outcome in 5-10 minutes?

@gautham_city_ even better

@__JimmySam__ nah just get them to book a meeting on the spot

how to network at events where you know nobody: 1. go to a random person 2. talk to them 3. ask them who else you should talk to and what you should ask based on the conversation you had (or even better, have them take you to them & intro u) 4. repeat

@gautham_city_ another good thing is to try and get to know the organizers of these events very well, and then go through their network to find the right people

client happy = jimmy happy

me when it gets cold outside

@49agents am I talking to an AI right now or are you a real person?

curious how ppl use Claude Code/Codex or other tools my workflow looks like this > start with plan mode > Separate plan into phases that include: - files to edit - detailed task descriptions - subagents to use (teams too) - skills and docs to reference - checkpoints

@49agents wtf lol

@49agents what does your workflow look like?

@__JimmySam__ how are you structuring the phases? i find separate phases helps but hard to get right

@49agents basically just this: - files to edit - - detailed task descriptions - subagents to use (teams too) - skills and docs to reference - checkpoints heavy emphasis on the detail. best practice is to make sure you understand everything and if it doesn't make sense keep planning

> accept the plan and then review at each checkpoint I use 2-3 terminals at once in a worktree working on separate features and have hooks that run tests, lint and type check

@gautham_city_ but it might get your account banned so maybe don't do that

@gautham_city_ curious how comet got through 400+ ppl so quickly? did you have multiple browsers open?

was at a rail conference last Thursday and got a paper booklet of the attendee list. strat to connect with the 400+ people in the booklet in 45min: (at least until I hit the connection limit lol) 1. use iPhone camera’s OCR to copy all the names 2. Clipboard syncs to Mac, put all the names into an empty markdown file in cursor 3. A bit of regex and multi-caret magic to format it into TS array & write a quick script 4. Batch generate prompts with 10-20 names each, telling @comet to connect with each person 5. Put the prompts into comet and go ham not bad but would be curious if someone’s got a better solution?

@gautham_city_ haven't tried it yet but with apollo.io sequences you can autoconnect with ppl on LinkedIn in times intervals over the course of weeks so that you don't hit connection limits and your account doesn't get flagged

@gautham_city_ oh wait connection limit