ben ray

@OttoKurect @koeng101 this man understands the game

@koeng101 Also, the best deals are gambles. Have i lost $100 attempting to get a 20K instrument 20k under market? Yes. So what...has to be your attitude. But the wins are huge

You would not believe the discount you can get off lab hardware if you’re willing to reverse engineer the serial connections and buy used. 50x easily.

@dwiel @koeng101 you need to be bidding against us on ebay

@koeng101 what is your fav place to find gear like this?

@MathSRIsh reliability comes down to how completely you’ve mastered the hardware. the pylabrobot hardware guide for Hamilton STAR will get any nonbent ebay robot back to factory accuracy: docs.pylabrobot.org/stable/user_gu…

@_benray "reliable" is not a word I would use to describe eBay products ;)

fewer than 10 ppl on earth use ai to control local automated biology workstations

@MathSRIsh ebay is your friend!

@_benray If you can make it happen for real DM me

@MathSRIsh this is why my startup pitch cleanly explains how we build reliable integrated workcells for <$20k

@_benray Fewer than 10 people have the budget for it. You need 300k+ for the robot, and then the price of the reagents. Definitely on my bucket list for our series A. Might try with an opentron in between but experiment range is more limited.

@hzlmshmn information wants to be discovered

@_benray i wonder how much more the vendor gonna charge to expose their api/mcp for this vs everyone just move to plr

@WillPVGreen this is indeed an issue on the ot-2 however the setup pictured is all custom festo parts and can detect resistance along the z axis if you need to self-correct for z pinning. if you have a good labware definitions and hardware with sensing capabilities, the robot can self-teach

@_benray Is needle misalignment an issue in these setups? Such as incorrect depth calibration/well locations or is this a fixed issue for automated workflows? I remember when using opentrons last year it was a lot of visual trial and error with the system not auto flagging mistakes.

My first interview with @Peter_J_Beck, Founder & CEO of @RocketLab. Rocket Lab is scaling launch cadence faster than SpaceX scaled Falcon 9. 0:44 The biggest bottleneck to increasing launch cadence and mass to orbit 1:25 Growth of new space startups 3:05 Why governments have been ineffective at scaling launch 3:53 Tall Poppy Syndrome 4:38 Starting Rocket Lab without having $100 million 7:24 Scaling Electron vs focusing on Neutron 9:17 Rocket Lab hustle 11:10 Creating a culture of “F*ck it, let’s do it” 11:40 Worst supplier experiences 13:54 Having an engine explode before an important meeting 18:52 Rocket Lab’s first mission to the moon 20:55 Chewing glass and forcing the outcome to be good 22:57 Similarities in how Elon and Peter operate 25:05 Elon time and how to structure timelines 27:13 Designing a culture where everyone runs towards the fire 28:44 Making the Ferrari of rockets — the importance of creating beautiful things 31:20 “You don’t need to equate a price tag to beauty” 32:12 Why Peter hates launch day 34:48 “After a launch failure this place is a morgue” 35:22 Moving forward after a failure 36:40 Transitioning from an R&D organization to scaling rocket production 38:07 Production hell with rockets 39:09 Taking learnings from Electron to Neutron 41:03 Having dinner with Elon 42:00 Keeping the hiring bar extraordinarily high 44:05 “My job is to fix sh*t” 44:56 Rocket Lab’s first NASA launch 45:55 “The great thing about America is anything is possible” 47:40 Coming to Silicon Valley — meeting with Vinod Khosla 51:22 Why he decided to take $RKLB public 54:12 Creating a company that will outlive you 55:41 Turning Rocket Lab into a profitable business 57:23 The best space companies will all build their own rockets 1:00:35 Scaling Neutron 1:01:32 Why they’re using carbon fiber instead of steel 1:02:56 “Going public was a great capital unlock” 1:04:10 The importance of relentless optimism

Today’s guest on the Free Radicals podcast is @ricomnl, head of applied AI @retrobio_. Retro was seeded with $180M by OpenAI CEO Sam Altman to develop therapies to prevent and reverse age-related disease, and is widely recognized as one of the leading AI for longevity companies. This was a fun conversation about Retro’s work with OpenAI to engineer 50x more efficient Yamanaka factors in a matter of months, what it means to build foundation models that can reason across natural language and protein sequence, and why the bottlenecks in biology are more experimental than computational. We also get into the biology of aging and how AI can enable therapies that dramatically advance healthy lifespan. Be sure to follow me and @EricDai_BioE to stay up to date on the latest news in longevity biotech! And Special thank you to @NFX & @omri_drory for lending us their beautiful podcasting studio! 0:00 Intro 1:42 Engineering transcription factors for more efficient reprogramming 19:02 How protein models are used 30:40 Exciting developments in protein engineering 36:20 Challenges in predicting protein behavior 39:38 Do scaling laws apply to protein models? 43:41 How these models are useful for longevity 56:15 Existing pathways for damage repair in the body 1:03:07 Will superintelligence solve aging for us?

Rocket Lab w @Peter_J_Beck

Robots should be able to use any machine in a lab. To do that, robots need to reason about objects: their geometry, how they move, and how to avoid colliding with them. Our solution is meshes: 3D representations of real-world objects. We create these by leveraging depth information from cameras. With a mesh-based representation, we can explicitly model geometry, plan safe motions, and maintain precise alignment independent of lighting conditions or visual noise. We also learned the hard way that RGB data on it’s own (without depth) is not enough after several attempts to operate purely in pixel space. We built simple tools that let anyone: • Create meshes - scan objects with depth cameras or import existing models • Annotate meshes - mark joints, buttons, lids, and other moving parts • Build a shared mesh database - a growing library of annotated objects used by the system A new object can be scanned, annotated, and ready for the robot in no time. Below is an example of scanning and annotating a centrifuge mesh. cc @BastotdeHeijden @BlerimAbdullai for spearheading this work

@SethSHowes @ProfTomEllis few ppl dream to resurrect decade-old instruments it just is the only way to bootstrap automated biology factories

@_benray @ProfTomEllis oof these don’t retain their value on the used market given they cost upwards of $100k new

1/ Lab automation is insanely mispriced. Today I visited a London-based biomaterials startup running large-scale bacterial screens. They use a colony picker that costs $95,000

@SethSHowes @ProfTomEllis ebay.com/itm/1875460596…

@_benray @ProfTomEllis equally true for a liquid handler for which there is much greater demand. you can build one of these for $500 but the cheapest option on the market is an opentron which will set you back $20k

markets fail to serve biology because there are likely fewer than 1k people that would purchase a colony picker today at any price, even though the world would be better off having a large surplus of automation-friendly cheap scientific instruments the only goal of an instrument salesperson learns is to grift 100k off your startup

@ProfTomEllis I don’t doubt that the PIXL is the best option for colony picking on the market and the Singer team have meaningfully progressed this capability. I’m still confused and trying to understand the market incentives that result in this kind of pricing structure

Moore’s Law is an undemandingly slow yardstick for any later-emerging field like DNA sequencing

Sorry. I have some sadness to report. corememory.com/p/los-angeles-…

you might not like it, but this is what peak pipetting looks like