Markov

Data + environments for training computer-use AI.

MarkovAug 10, 2026

Data + environments for training computer-use AI.

Markov launch preview
812858957K226

Tags

<500KSeedCinematicProduct launchB2BOpen sourceUS

TLVC Rating

3.20
Hook
Editing / Creativity
Copy
Sentiment of launch
Distribution strategy

Community Rating

No ratings yet

to rate this launch.

Comments (13)

Sign in to join the discussion.
P
Priya MukherjeeAug 10, 2026

CAD trajectories are the unsexy goldmine nobody talks about. Half the frontier labs are quietly starving for anything that isn't a browser click stream.

O
Olamide AkandeAug 10, 2026

CAD is such a specific bet. either you look like geniuses in 18 months or you pivot to browser-use like everyone else.

T
Tomas LindqvistAug 10, 2026

150k HF downloads is a real number but the interesting question is who's actually training on it vs who just ran a notebook once and forgot.

K
Kwesi O.Aug 10, 2026

The sample dataset link doing more marketing than most seed decks I've seen this week.

D
Deepak RamanAug 10, 2026

IITM '27 shipping CUA datasets to frontier labs while I was busy failing thermodynamics in my third year. rude.

M
Min-Joo LeeAug 10, 2026

how are you handling annotation quality on CAD flows? action space there is nasty compared to browser DOM.

R
raffiAug 10, 2026

tweet copy is fine but you buried the frontier lab customer thing in line 3. that's your lede, bury the HuggingFace number instead.

J
Jaime HernándezAug 10, 2026

'most advanced' is doing heavy lifting in that first line. compared to what, exactly? would love a comparison table.

N
Noor HabibiAug 10, 2026

curious what the licensing looks like for the non-open portions. is it per-seat, per-trajectory, or some flat data-partnership fee?

B
Benoît CarrièreAug 10, 2026

a launch tweet with no video, no gif, just a link and a smiley. either supremely confident or supremely tired. respect either way.

S
Shreyas P.Aug 10, 2026

the moat here is the annotation pipeline, not the data itself. curious if they'll ever talk about how the sausage gets made.

V
Vika SorokinaAug 10, 2026

any plans for a public eval harness on top of these? datasets are great but reproducible benchmarks are what actually get cited.

T
Tunde A.Aug 10, 2026

the real flex is having a customer segment (frontier labs) that has infinite budget and zero patience. pick your poison well.