Edrak (@edrak) header
Edrak (@edrak) profile picture

Become an AI company.

Joined July 2026
Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)Photo by Edrak (@edrak)
Pinned Tweet
Edrak brings you Al that deploys Al. It proactively detects what can be automated, then builds and deploys hundreds of Al agents to handle workflows across the tools your teams already use. Become an Al company with Edrak.
2
3
208
The ECB asked 20,000 people a month what AI actually saves them. The median user: 3 hours a week. People using it for code: about 8. That is 7.7% of a working week, from a tool most people have had for two years.
1
1
1
87
Those hours show up in hiring. US job titles mentioning AI went from 264 in early 2022 to 822 in early 2026. About 1 in 12 titles now, and 63% of them outside tech.
1
1
1
36
Meanwhile the price keeps falling. Running a GPT-3.5-level system got 280× cheaper between 2022 and 2024. The token price index is down 41% from its March peak, at $0.68 per million. Hardware costs fall about 30% a year.
1
1
9
The gap between companies isn't necessarily the tools at all. The top performers are 2.2× more likely to have redesigned workflows around AI, 1.9× more likely to incentivise using it, 1.7× more likely to train for it. They report 7.2× higher AI-driven performance.
1
Your AI could take orders from a scam email.
1
1
40
Buying that AI tool?
1
2
24
You might be paying for the waffle.
16
Would you let an intern delete the original? So why AI?
1
1
56
What did AI save if you’re spending longer fixing its work?
1
1
54
You AI-written product page could be selling features that don’t exist.
1
1
61
"tell me what you're thinking. leave nothing unsaid, and remember, make no mistakes." Mostly Harmless CoT: on the reliability of chain-of-thought.
1
84
Three years ago, using AI at work was unusual. Four independent studies now put it between 39% and 88%, depending on who you count: organisations, businesses paying for it, euro area workers, or US workers using it weekly. Different groups, different definitions, all moving the same way.
1
1
46
That AI-chosen quote might be costing you way more.
1
1
29
Your AI might be quoting a policy you scraped months ago.
1
1
30
Half an hour a day copying? That’s roughly 10 hours a month.
1
1
26
Keeping fusion plasma stable is difficult because it is hotter than the sun and can fall apart in a few thousandths of a second. Princeton’s PACMAN system was tested on the actual DIII-D machine in San Diego. It watches the plasma and issues new commands roughly every 20 milliseconds while a shot is running. In one case it flagged a dangerous magnetic instability a fifth of a second before it would have appeared and adjusted the plasma in time to stop it. The same system also directed all six of the machine’s microwave heaters together. Researchers still choose the targets, and the machine’s own safety limits always have the final say. Source: Rothstein, Farre-Kaga et al., Nuclear Fusion 66, 076050 (2026) · Princeton PPPL and Princeton University · five experiments at DIII-D
1
1
50
For decades, whale clicks looked like random Morse code. Project CETI ran nearly 9,000 Eastern Caribbean codas through machine learning and found they are built from four reusable parts (rhythm, tempo, rubato, and ornamentation) forming 156 distinct sequences. This year the team showed the clicks also contain two vowel-like patterns (a and i) that behave like human vowels. Orcas show similar structure in 61 hours of recordings. Their first language model can already predict when a whale is about to dive. Source: Project CETI (MIT CSAIL & UC Berkeley) · Nature Communications 2024 · Proceedings B April 2026 · orca analysis spring 2026 · WhaleLM Oct 2026
1
1
32
You deleted the client’s name.
1
1
57
Your first AI project should be boring.
1
1
56