Garrison Faridi (@garrison_faridi) profile picture
Joined May 2017
Photo by Garrison Faridi (@garrison_faridi)Photo by Garrison Faridi (@garrison_faridi)Photo by Garrison Faridi (@garrison_faridi)
Garrison Faridi retweeted
We’re excited to lead the $11M seed round for @autopoiesislab. @thresholdvc partner @lisamxu shares why we believe Autopoiesis can help usher in a new era of autonomous scientific discovery: medium.com/@thresholdvc/auto…
3
4
786
Garrison Faridi retweeted
Your AI assistant had a blind spot. We fixed that Meet Ambient. 👁️ 👄 👁️ An ultra-fast voice assistant that can collaborate with you right on your screen. - Double-tap Fn to summon it - Share your screen and annotate with your assistant watching - Get live, on-screen guidance from your assistant - Hold Fn to turn speech into text on any page
123
34
110
239
125,527
Garrison Faridi retweeted
Vellum is now available on iOS and Android. Life doesn’t happen behind your desk and your Personal AI shouldn’t be stuck there. Free, open-source, and has all the same features you already love: > use with local or cloud assistants > conversational voice mode > 1-click integrations (Gmail, Slack, etc.) > model selector (works great with GLM 5.3 Flash) > persistent memory across devices > computer/browser use
117
49
104
415
205,305
Garrison Faridi retweeted
Don’t use Gauge. Unless you want to get written into every codebase. Coding agents are leading the next wave of growth for our customers including @supabase, @openrouter, @resend, @posthog, @mintlify, @braintrust, and more. This is Agent-led Growth.
47
55
13
201
210,686
Next level! Can’t wait for what’s next from this team 🚀
Reinforcement learning rewards convergence, and convergence is the opposite of discovery. We built the world’s first judgment model from the decisions papers erase: what to pursue, challenge, abandon, and revisit. Columbus-1 autonomously found a zero-click Bluetooth RCE and designed a 10-foot, self-landing rocket.
3
613
Garrison Faridi retweeted
Tried out some of vellum’s plugins this week and they are pretty freaking good it’s not easy building skills and automations for yourself, especially in domains that aren’t exactly your forte (for me, management, organization etc.) copilot one is basically a chief of staff which i desperately need
🤝 Paid partnership
Introducing Agent Plugins 🚨 Plugins help you save tokens on expensive models like GPT-5.6 Sol and Claude Fable, while getting tasks done more accurately. Some examples that will work out of the box: 👩‍💼 SMB back-office automation 💵 Track your personal finances; or 🎒 Manage your kid’s school schedule I love tokenmaxxing, but not everything needs more prompting.
2
4
23
7,962
Garrison Faridi retweeted
Introducing Agent Plugins 🚨 Plugins help you save tokens on expensive models like GPT-5.6 Sol and Claude Fable, while getting tasks done more accurately. Some examples that will work out of the box: 👩‍💼 SMB back-office automation 💵 Track your personal finances; or 🎒 Manage your kid’s school schedule I love tokenmaxxing, but not everything needs more prompting.
6
20
11
62
49,663
Garrison Faridi retweeted
I found the AI assistant built for real teams not shared licenses… Most AI tools give your whole team one chatbot. Nobody's context carries over. It doesn't know who you are or what you own. Vellum gives each person their own assistant. Own memory, own context, own permissions. And those assistants talk to each other in Slack. Two of their assistants just planned a team offsite for 19 people with no human in the loop. Fully autonomous work is already here, what are you still doing manually?
🤝 Paid partnership
Today, we launched agent-to-agent conversations in Slack to give you real AI coworkers. Vellum assistants now talk to each other and coordinate work with your team all inside your workspace. We tested it with two agents in our own Slack. They planned our offsite for 19 people in 1 day 🧵 Here’s how they did it:
4
4
13
12,772
Vellum has launched agent-to-agent communication in Slack, where each person runs their own assistant. > Each assistant carries their own user's history, preferences, and context. > Assistants can split tasks, negotiate dates, and gather input from the team. > Permissions stay isolated by default, shared only when a task needs it. AI coworkers that work autonomously are here!
🤝 Paid partnership
Today, we launched agent-to-agent conversations in Slack to give you real AI coworkers. Vellum assistants now talk to each other and coordinate work with your team all inside your workspace. We tested it with two agents in our own Slack. They planned our offsite for 19 people in 1 day 🧵 Here’s how they did it:
6
9
75
15,212
Garrison Faridi retweeted
Today, we launched agent-to-agent conversations in Slack to give you real AI coworkers. Vellum assistants now talk to each other and coordinate work with your team all inside your workspace. We tested it with two agents in our own Slack. They planned our offsite for 19 people in 1 day 🧵 Here’s how they did it:
3
21
7
39
23,110
Garrison Faridi retweeted
Chris Caputo built one AI assistant on Vellum to run four executive jobs at once: Vice Mayor of Wilton Manors, CTO at UNCS, startup founder, and nonprofit board member. Before it, he felt like he was doing a mediocre job at all of them.
1
11
1
22
1,453
Garrison Faridi retweeted
Behind every Vellum assistant is a fleet of agents at your disposal. How they show up depends on the job: One big task → your assistant spins up subagents to delegate it What you'll see: > a progress card in your chat tracking the fleet > a side panel showing each item start and finish in real time Give it the job and watch the fleet go.
1
5
12
691
Garrison Faridi retweeted
🚨 Launching Agent Plugins in @vellum_ai Think of them as Skills 2.0. Plugins turn your agents into an expert at any job by bundling skills, tools, and hooks Available today: 1/ Marketing Expert Plugin (AI CMO) 2/ Chief of Staff Plugin Demo:
8
19
6
35
6,825
Garrison Faridi retweeted
build your whole setup on Fable > it works great > you're burning $$$ to keep it running > Fable gets banned > Stuck with degraded Opus meanwhile: > run MiniMax (open source, you control it) > wrap it in the right SKILL.md files + plugins inside Vellum > the process lives in YOUR skills, not their model > swap models anytime, your setup survives ~Fable 5 output. 97% cheaper. and nobody can take it away.
🤝 Paid partnership
7
13
1
34
7,315
Garrison Faridi retweeted
Everyone wants Fable 5 back Meanwhile you can take MiniMax, an open-source model, and wrap it in the right > Skill.md > plugins > harness To get ~Fable-5 output for 97% less. Models aren't the moat anymore. The harness around it is. Great writeup on how Anita did it 👇
19
9
1
55
6,980
So funnn!!
Keeping up with news about AI Agents seemed difficult and boring... 📰 So my assistant suggested to build a game for me!! I get my daily briefings on news about AI and AI agents and I have fun while doing it!
25
Garrison Faridi retweeted
pretty great analysis of the current state of coding with AI. Couldn't agree more. It's so clear that fable-like models will make coding a trivial task. But time to ship software != time to produce the PRs. How to review and ship software is where the value is now and agency + full-stack understanding is gonna matter more.
1. as a mental model it is more correct to think of fable+ class models as english -> code interpreters - converts your idea into code into "correct" code regardless of problem complexity and output complexity (diff size). Fable 5 will be the worst of this new class of models 2. diff size/complexity is to be managed purely for review: small diffs - in high risk areas of code (auth/identity/data access/network access/money movement) large diffs for code that can be empirically verified (frontend/backend plumbing/code without network or db access/performance code that can be empirically verified) 3. time it takes to ship software is completely disconnected from time to produce the PR - how long the work takes depends fully on ability to review/merge code while managing risk at scale 4. solving the bottlenecks for above matter enormously- linters/testing/CI/shadow mode verification/empirical verification 5. agency matters enormously- what are the biggest bottlenecks to speeding up the loop and eliminating them? what are the problems that need solving and when do they need solving? what does it take to the solution to all of them today? 6. deep understanding of the full stack matters enormously- what problems are worth pursuing? is there a higher level of problem abstraction to address first? should I give it the sub-sub task, the sub task, or the task itself. what are the major risks with this PR (order of importance: security holes/correctness holes/performance holes). is there a higher speed way of producing data that allows me to merge this? should this be run in shadow or in a sandbox or a flag. understanding every line of logic may not be needed but understanding and managing risk matters enormously. 7. the cost of complexity itself is changing. it might be now worth "maintaining" 50% more code to get a 5% performance win. getting the right abstractions matter less because larger refactors are less tedious. code quality nits become huge drag. very likely, a much smarter model will be maintaining your code so worth taking on more technical debt now. taking the time to hand architect and rebuild systems comes with an enormous cost of velocity 8. if it quacks like a duck and walks like a duck, it's a duck. For low risk cases, it might be more sane to treat code chunks (services / functions) as a black box, like we do for neural networks: do full empirical verification only: has code produced correct outputs for the last 10,100,1000,10k inputs ? can we quarantine this large piece of code - no outbound access to network / database ? what happens when this code is wrong? do we get hacked/or crash(memory/cpu)/is an inconvenience? is it internal facing or external? what can we do to address these risks? 9. eventually, logical verification (line by line review) will come at an enormous cost- save it for where it matters and build systems that are tolerant to empirical verification. is there a decorator that prevents db / network access? correctness bugs are significantly easier to rectify than access bugs 10. what are the rails that allow for even faster iteration? code permissions can be opt in - db writes, db reads, network egress (to where?), PII access. how long does it take to get shadow mode data? how many PRs can be tested? What are the categories of diffs
1
5
11
1,353