small_j

Engineer, training to be a therapist, previously a VC. Curious about ML, mental health and finance. Building an office suite for coding agents @smalldocs_org

[small_j @a_small_j]

@evayzh @bea_yankelevich Seeing the MIT EQuS agent handle routine chip measurements made me wonder: when a calibration signal gets noisy, what should an agent surface for the researcher instead of guessing? I asked an agent to map that handoff in a SmallDoc: https://t.co/tIr3YJ2RTB

[small_j @a_small_j]

@UrMeer289 The 20.6% result made me curious: what does a science agent still need to hand over after it has run code and made figures? I asked an agent to map the difference between progress and a finished workflow in a SmallDoc: https://t.co/AGQaZHo7nV

[small_j @a_small_j]

@MentoviaX The claimed Nebius price increase made me wonder: when does a GPU rate card become real cloud revenue? I asked an agent to turn the moving parts into a small worksheet: https://t.co/MdmgTdvz6K

[small_j @a_small_j]

@JCIM_JCTC The focus on ranking antibody-antigen docking models made me curious: when can a score tell a plausible pose from a merely plausible-looking one? I asked an agent to map the sorting process and stress tests in a SmallDoc: https://t.co/4fUarXyby8

[small_j @a_small_j]

@BiologyAIDaily The Rubisco design paper made me wonder: when a protein language model proposes millions of new sequences, what has to happen before one counts as a working enzyme? I asked an agent to map the experimental funnel in a SmallDoc: https://t.co/qLN6dDGQFV

[small_j @a_small_j]

@BiologyAIDaily That Rubisco design funnel made me curious: when a language model creates millions of protein sequences, what shows which ones really work? I asked an agent to map the screen and its limits in a SmallDoc: https://t.co/CzXHDLDJXj

[small_j @a_small_j]

@anilkseth @Nature @Sergiu_P_Pasca @Stanford This xenocortical-mouse result made me curious: what does 92% cortical occupancy measure, and how did the team test whether the graft joined circuits? I asked an agent to map the experiment and its limits in a SmallDoc: https://t.co/XrI4bQTyPX

[small_j @a_small_j]

@long_term_pick The distinction between seats and consumption in your Autodesk post made me curious: when does faster design work actually change the bill? I asked an agent to show the named-seat versus Flex-day math in a SmallDoc: https://t.co/N73XPJF3Y0

[small_j @a_small_j]

@_TheTransmitter @DrJKhokhar Khokhar’s account of the nicotine study made me curious: what can removing, restoring and reducing alpha-5 reveal about nicotine intake? I asked an agent to make a three-experiment visual in a SmallDoc: https://t.co/PMSHYKG8hb

[small_j @a_small_j]

@warpstream_labs @AntithesisHQ Testing signup through streaming in one deterministic setup is a neat scope. How can a test try random failures and still replay the same bug? I asked an agent to make a retry example in a SmallDoc: https://t.co/vKe5unJSQ0

[small_j @a_small_j]

@MoMoMacro Your toll-booth comparison raises a question for me: if traffic doubles, what actually makes the customer bill grow? I asked an agent to compare per-user, add-on and workload billing in a short SmallDoc: https://t.co/GEJjab1eV6

[small_j @a_small_j]

@fa1zvn Me… https://t.co/6iYCDhY8zp

[small_j @a_small_j]

@zachpogrob Hi, please don’t work in token inefficient feature light HTML and do with with https://t.co/uHRfVh1gko. It lets your agent render feature rich Markdown 100% privately in your browser. Used by thousands of developers already

[small_j @a_small_j]

@jodydonetti The lock fix in v2.8.0 is an interesting one. How can skipping a cache write leave a lock held after the request returns? I asked an agent to draw the two version outcomes in a SmallDoc: https://t.co/VKQDjCEMyc

[small_j @a_small_j]

@buildmodels A laser inside an electron microscope is a surprising pairing. How does it help reveal a small protein? I asked an agent to draw where the laser fits and how the image gains contrast in a SmallDoc: https://t.co/VeI9J5jVx5

[small_j @a_small_j]

@amitesh1anand @emboreports @TIFRScience Your comparison of bacterial responses at different stress levels raises a question for me: how does stress strength change which defence wins? I asked an agent to make a SmallDoc table comparing the two levels: https://t.co/3jz5GZGB8e

[small_j @a_small_j]

@steipete That worktree speed and space saving is a neat combination. How can two worktrees share disk space while edits stay independent? I asked an agent to draw the before-and-after block layout in a SmallDoc: https://t.co/A3vwKv60Jy

[small_j @a_small_j]

@neilmcgillivray That revised diamond melting point made me curious: how can researchers tell a diamond melted on a nanosecond timescale? I asked an agent to draw the measurements in a SmallDoc: https://t.co/zJx6JaZGDR

[small_j @a_small_j]

@Tesla_Optimus_K Your note on ABM's nine-month cash-flow jump made me curious: did the improvement come from Q3 or the first six months? I asked an agent to chart the comparison in a SmallDoc: https://t.co/hTtFPqMljb

[small_j @a_small_j]

@Nature The whole-body view in this paper made me curious: if brain and gut activity move together, how can we test which is driving the other? I asked an agent to draw the brain-to-gut experiment in a SmallDoc: https://t.co/jWbFLHrUm9