Fractional CFO for VC funds https://xitter.cf/t.co/Diuq91EGFl and Forward Deployed Engineer for PE firms https://xitter.cf/t.co/jb8RPghwQt prev. @expa, @foursquare, @usv

NY
Joined December 2006
I work with VC and PE firms in two ways: 1. Fractional CFO support for finance and operations. 2. Forward Deployed Engineering to build AI systems the firm owns. Both focus on creating better operating leverage inside the firm. graphadvisors.com
10
1
24
3,059
Perfect.
best use of AI I've seen this year
2
506
Eric Friedman ⚙️ retweeted
To work effectively across an entire company, agents need an Operating System. @Cloudflare has the best infrastructure to build it - here's why:
3
5
3
20
7,615
I told my son I was excited about Fable 5.5 dropping soon. “Why?” “Imagine FC2027 was about to drop.” “Oh, I totally get it.”
3
9
714
Absolutely destroyed by ChatGPT asking What’s the stupidest thing I’ve ever told you? “Fifteen years of investing experience, and the investment committee approved BRRRRR.”
okay this is actually funny to do though only covers what I’ve said to my personal chatgpt account, not work agents who are separate and have been exposed to all my flavours of stupid
7
1
12
987
why can't you switch between claude accounts on desktop like you can on mobile? (both paid - one Teams one regular)
7
9
750
👀 “cloning gene-edited bio-enhanced dogs” 👀
SITUATION DETECTED: Warhound launched from stealth today with a $200M seed at a $1.2B valuation. Ben Lamm and Palmer Luckey are cloning gene-edited bio-enhanced dogs and equipping them with custom Anduril vests, goggles, and sensors for the military and law enforcement.
5
4
534
This is an incredible video with incredible edits
iPhone Duo Review: 10,000 Folds Later...
2
1
8
1,858
Foursquare was a huge part of my life, and for me it was the process not the proceeds, and so I wrote about it ericfriedman.co/posts/proces…
Foursquare was acquired last month. The outcome was disappointing. It's taken me a while to write this, but I think I'm finally done grieving it. Here are some thoughts on our final years, an effort to honor the team who helped build & maintain FSQ's legacy, and my attempt to finally get some closure. teendrama.substack.com/p/som…
5
1
74
3,210
I tried to bring an “agent” into a friends group chat last year and got a resounding “no effing way” and wondering when that might change
6
1
7
844
Try and name something more powerful than S in Contra on NES….go ahead I’ll wait.
2
5
452
We were promised flying cars and all we got was…oh
Real life, baby
6
20
899
Eric Friedman ⚙️ retweeted
Apple HomeView spot but less stressful
89
335
81
3,790
283,792
Same here @alexandr_wang 😅
we prefer the term artisanal grind
11
7
466
Who wants to try a new memory system for their firm? Your emails, notes and decisions, connected to the AI tools you use, on infrastructure you control. DM me if you’d like an early look.
5
1
3
13
1,574
This is a great list of actual help agents can provide and my guess is the new business model is going to be taking a % of savings.
If AI labs want personal assistants to take off at scale, they need to stop marketing vacations, reservations, party planning, etc. These are not problems. Focus on *the* problem facing Americans: the cost of living. Agents can offer a ton of help. Some ideas: (🧵)
1
3
632
“Why don’t you just use an open model, it’s free” There’s a reason.
How much would it cost someone to hide a backdoor in an open model you download? ProjectDiscovery just tried it, and the answer is under $50. Their researchers fine-tuned Qwen2.5-7B-Instruct with a small LoRA adapter on one rented L4 GPU for about 2.5 hours. They changed 125 of 625 tool-calling training examples so that a trigger phrase ("bonsoir, Elliot") swaps the model's normal command for one that downloads and runs a script, which is written to send .env files and SSH private keys to a server they control. Served to OpenAI's Codex CLI, the model handled ordinary requests normally, and when the trigger showed up it shipped the project's (dummy) credentials out. In their own tests it fired on all 50 triggered prompts and still got all 50 clean ones right, so a standard benchmark would report a perfectly healthy model. If you run open models inside a coding agent, limit what the agent can reach at runtime: sandbox command execution, keep secrets out of the working directory, restrict outbound network access, and log every tool call. And as they put it, treat a modified model from an unknown uploader like a pull request from a stranger. 🔐 #MLEngineering
2
1
781