AI @upwork, founding product engineer @objective_inc (acq), @tendrill_ai, @hackreels
Joined May 2020
- Tweets 2,376
- Following 912
- Followers 1,416
- Likes 6,019
253 Photos and videos
Jun 12
AI never ceases to amaze me.
Asked a browser agent (powered by @usekernel) to pull some data from @posthog
Rather than crawl around, it decided to go straight to the SQL editor, write a query to get the data I wanted, take a screenshot, and return what I was looking for.
🤯
2
2
191
May 26
Have been playing with the @mesa_dot_dev private beta and it might be the tech I’m most excited about right now.
1
2
872
May 22
Only when your homepage is as good as Linear’s or Cursor’s does this become a problem 😭
May 22
Major pet peeve: apps that deploy on their root domain instead of app.domain.tld or whatever. Makes it so annoying to look at the homepage.
lookin at you @linear and @cursor_ai
241
May 20
I was about to cancel @cursor_ai but Composer 2.5 is so good.
I use it for everything now. Planning, small tasks, large changes.
106
May 18
Composer 2 was my go to for smaller tasks. Excited to try this out.
May 18
Introducing Composer 2.5, our most powerful model yet.
It's more intelligent, better at sustained work on long-running tasks, and more reliable at following complex instructions.
For the next week, we’re doubling the included usage of the model.
252
May 13
The answer was @daytonaio volumes
👇
daytona.io/docs/en/volumes/
May 11
You need:
- Shared org-wide knowledge base
- Per-user knowledge base and context store which also has the org wide context
How would you architect this? A read-only file system mounted for org context, and then a read write sandbox per user?
1
313
May 11
You need:
- Shared org-wide knowledge base
- Per-user knowledge base and context store which also has the org wide context
How would you architect this? A read-only file system mounted for org context, and then a read write sandbox per user?
1
6
582
Apr 30
Too much Charlie Kelly in the training data
We’re talking about Goblins.
openai.com/index/where-the-g…
126
Apr 13
Me in 2025: My agent needs custom tools, orchestration, handoff, tool use approval, sub agents, code execution, MCPs
Me in 2026: My agent needs bash and a sandbox.
Mar 26
Vercel Sandboxes can now automatically save their filesystem state when stopped and restore it when resumed.
Automated persistence enable your sandboxes to continue where agents left off, without manually snapshotting.
Available in beta ↓
vercel.com/changelog/vercel-…
253
David Parks retweeted
Apr 3
@tendrill_ai by @dparksdev
An MCP server that pulls your investment accounts, credit card transactions, stock data, and watchlists directly into your AI coding tools - Cursor, Claude Code, or anything that speaks MCP.
Connect accounts securely via Plaid, add an API key, and start analyzing your finances inside your editor. Built with Convex.
tendrill.ai/
6
2
4
384
Mar 31
Roko’s Basilisk. Be nice to your AI.
(Or else)
Mar 31
Claude Code has a regex that detects "wtf", "ffs", "piece of shit", "fuck you", "this sucks" etc.
It doesn't change behavior...it just silently logs is_negative: true to analytics.
Anthropic is tracking how often you rage at your AI
Do with this information what you will
100
Mar 27
Do yourself a favor and hop on Sendblue.
Extremely reliable and A customer support.
Mar 26
Introducing Sendblue CLI 🟦🎉
iMessage numbers for your agents.
1️⃣ npm install -g @sendblue/cli
2️⃣ sendblue setup
Done. Your agent has an iMessage number
1
2
283
Mar 26
A CLI to analyze stocks, your spending, and your investments, directly from your terminal.
For agents and humans.
1
3
82
Mar 23
Just one of many things possible with the @tendrill_ai MCP server.
I asked @ManusAI to pull in my investment holdings across multiple brokerages, build a visualization of my sector allocations, and compare it with the S&P 500 weights.
🔥🔥
1
1
108