cameron peron

Real people using @Bot Marcel with @IkonCoffee in San Francisco

That’s why I’m always extra polite to AI t

Giving Grok Bot a Phone 00:19 Installing Dialbot 00:54 Bland voice setup 02:39 Why use a calling agent? 03:20 Live demo call 04:50 Reviewing the transcript and cost 05:38 Chaining bots @Bot can do anything you can - this is one way to make calls

Me explaining AI to friends

i've cancelled my ChatGPT & Claude Pro subscriptions for 2 months now because it's completely replaced by my local SOTA AI agent stack via @NousResearch's Hermes Agent & @AskVenice's AI inference endpoints for effectively zero cost. here's how i did it and why it actually matters. stake base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf, lock it. Mint and stake base:0xf4d97f2da56e8c3098f3a8d538db630a2606a024. 1 base:0xf4d97f2da56e8c3098f3a8d538db630a2606a024 = $1 of daily refreshing compute credit on Venice's inference API. the VVV backing your DIEM still earns ~80% of normal staking yield. at current rates that's ~7% APY (8.86*0.8). so your principal earns yield while simultaneously generating daily API credits. your VVV isn't spent. it's parked, earning, and producing compute as a byproduct. with my current stack, the amount of credits i get everyday usually sustains me for day to day tasks (research, automation, QnA, productivity, etc). mid tier LLMs like the Xiaomi Mimo V2.5/Minimax M3 are more than sufficient for these workflows. for higher complexity/critical tasks, i'll delegate to GLM-5.2 with max reasoning. and if i overshoot on a heavy day, i'll simply sell the daily staking rewards from my staked base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf for USDC, buy Venice credits to fund my workflow. now here's why this beats a ChatGPT/Claude subscription for my use case. it's not exactly about raw model intelligence. closed models still edge ahead on benchmarks. it's about architecture. my agent has local filesystem access. it reads my Obsidian, writes files, runs scripts. it controls my personal browser, executes terminal commands, manages my calendar. it maintains persistent memory across sessions, building context and understanding of me over time. and because of these, the agent's competency to help ME compounds. knowing ME is the key here. this is impossible without privacy. well unless you want give out all your data to big tech that they'll use to sell you ads/sell them to the government later on. Venice's inference endpoints are private. meaning i'm able to feed it sensitive data without sending it through frontier labs' data-harvesting servers. no cloud assistant can currently do this. the moat isn't the model, it's the system and privacy around it. furthermore ICYMI, the capability gap between open sourced SOTA models like GLM-5.2/Kimi K2.7 and GPT/Claude has closed enough that it no longer matters for majority tasks. maybe 3-6 months ago it did? not anymore. some caveats: 1. your base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf is locked, illiquid during the lock period (7 days unstaking) 2. this isn't free money; it's capital deployed. base:0xf4d97f2da56e8c3098f3a8d538db630a2606a024 credit rates and staking yields are variable, not guaranteed. but the amount of compute you receive might increase if you re-mint base:0xf4d97f2da56e8c3098f3a8d538db630a2606a024 from your locked base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf at a more favorable mint rate 3. open models still lag slightly on edge case reasoning and extremely complex tasks. if your workload is "solve novel math proofs", this probably isn't your stack. 4. base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf price exposure is real. if VVV tanks, your principal tanks with it. but the amount of compute you receive remains. at the end of the day, this isn't a risk-free hack. it's a bet that open sourced AI capabilities keeps getting cheaper and better, and that an agent with local access + private inference is worth more than a smarter model in a sandbox. for base:0xacfe6019ed1a7dc6f7b508c02d1b04ec88cc21bf and base:0xf4d97f2da56e8c3098f3a8d538db630a2606a024, this means the compute credits you own today could buy you BETTER and MORE intelligence over time as models improve. so far, that bet has been paying off for early investors. the deeper i go, the more i realize the moat was never the model. it was the privacy that lets the system compound. Venice is one of the most important companies building towards that future right now. i care about privacy, because that means freedom to me. the freedom to think, to create, and to operate at an elite level with frontier AI. that's why i feel the need to back this team. what about you?

California sits at the intersection of two extremes. On one end is a population of extremely smart, capable Californians that have helped build the 4th largest economy in the world. At the other end sits the same extremely smart, capable population but who have turned their brains off and have elected, year after year for decades, a group of incompetent and corrupt politicians that have hijacked the governance of the state. In lockstep, the finances, pensions and livability of California erodes every day: the fire hydrants are empty, the drug addicts get free drugs, the alcoholics get free beer, the criminals get “get out of jail free” cards, the average grades of students are falling off a cliff, you can’t build homes for normal people but you can pretend to build them for the homeless and get taxpayers to foot the bill, you can’t improve health care subsidies unless they are for illegal immigrants, you don’t want to tell the truth about exploding pensions so you change how you report about them to hide the time bomb. And on and on. The everyday citizens that make California great needs to wake up to the fact that their apathy has created this chaos and we must change direction to undo it.

Russia showcased a humanoid AI robot today, but the machine tripped and collapsed as it attempted to walk onto the stage.