Arjuna Anand
arjunaanand.bsky.social
Arjuna Anand
@arjunaanand.bsky.social
Follow me for AI Product Demos
bluesky fellas, don’t do this, okay ?
February 3, 2026 at 4:00 AM
don’t put liquids in this for long time, okay ?

or everything will be a mess, lol.

#plastic #life #daily #shopping
February 3, 2026 at 3:23 AM
2 tips to auto-test code by claude code, lol.

1. check semants using lint tools

- ban generic names,

- check for logic implemented in other layers, shouldn’t be ;)

- validate default values

2. use hooks to review modified files

- runs on stop hook

- hook into PostToolUse

#ai #vibecoding
February 3, 2026 at 3:19 AM
you want to become an ai pm ?

sure, 5 skills to master, lol.

1. ai prototyping (cursor, claude code)

2. observability (llm tracing)

3. ai evals

4. knowing options

(prompt engg. vs rag vs finetuning vs rl)

5. working with engineers

(metrics, label data, debug together)
February 3, 2026 at 3:05 AM
engineering managers.

use dx framework to measure impact of ai coding tools in tech teams,

- increase in pr throughput
- pr / change failure rate
- delivery speed in hours
- developer experience

final metric = pr’s per hour / cost per hour,

greater it is better it is, lol.

#ai #cursor #claude
February 3, 2026 at 2:51 AM
codex app on mac launched by openai,

chasing claude code functionalities in claude app as usual, lol.

#ai #codex #openai #chatgpt
February 3, 2026 at 2:36 AM
knowledge graphs as scratchpad,

is the only thing a multi-agentic system needs,

lol.

#ai #agenticai #knowledgegraphs #ml
February 1, 2026 at 11:53 PM
evolution of ai techniques,

sft -> unsupervised learning -> rl -> agents

what next ?

-> agentic meshes optimised for a task

-> agentic nets.

#ai #agenticai #neuralnetworks #openai
February 1, 2026 at 11:46 PM
moltbook is showing first signs of how an ai civilisation evolves,

and here rl guy inside me is just thinking about using this for optimising for a task,

lol 😂

#ai #clawdbot #moltbook #agenticai
February 1, 2026 at 11:37 PM
do you even use claude code to its highest potential, lol.

here are 10 tips from creator of claude code -

1. do more in parallel

(spin up 3-5 worktrees at once)

2. start every complex task in plan mode

3. invest in your claude.md

4. create your own skills & commit to git
February 1, 2026 at 11:28 PM
agentic search beats rag in every single way,

even creator of claude code confirms it,

then why do you sing praises of rag anymore ?

lol.

#ai #rag #agenticai #claudecode
February 1, 2026 at 10:52 PM
autonomous agents are here,

now who is making below layers for them,

- infra,
- memory,
- security,
- deployment environments,

who wants to be a millionaire, lol,

do you ?
February 1, 2026 at 7:35 AM
openai is dying,

nvidia’s jensen huang already sees this which led him to cancel their deal,

but what domino effect it will have on ai ecosystem?

anthropic’s b2b bet seems such a boon now,

who thought ai world will come to this, lol.

#ai #openai #chatgpt #nvidia
February 1, 2026 at 7:26 AM
now only if they give this robo a real looking face,

they would be able to move freely among humans, lol.

but are you ready, lol.

#ai #robot #china #xpeng
February 1, 2026 at 7:17 AM
true continuous memory for llm agents using knowledge graphs,

is next breakthrough we all are waiting for.

#ai #agenticai #memory #knowledgegraph
February 1, 2026 at 7:01 AM
who thought communism will win in all ways,

at cost of capitalism, lol.

#ai #ram #memory #china #ccp
February 1, 2026 at 6:56 AM
agents led you to clawdbot,

which has an army of agents at his disposal,

has its own devices and environments to do things,

what will clawdbot lead you to ?

- virtualhumans
- botmeshes
- botsocieties

lol, which one do you think is first ?
January 31, 2026 at 8:23 AM
humanity’s last exam for robots, lol.

#ai #robots #optimius #benchmark
January 31, 2026 at 8:15 AM
till now humans were alone in search spaces for new innovations,

now they will not be,

they will he accompanied by ai agents,

much faster & better than them, lol.
January 31, 2026 at 8:07 AM
very few know about lmcache extension for llm serving engines,

- smart kv cache management via reuse

- reuse for any repeated text fragment not only prefixes

- 4-10x performance boost for rag apps

- much lower time-to-first-token

- enables longer contexts

what were you doing, lol.
January 31, 2026 at 7:58 AM
genie 3 like world models,

short-term : eat up the renderer,

render pixels than triangles

long-term : eat up movies, games,

but yet very far due to lack of consistency & predictability.

still a big game-changer, lol.
January 31, 2026 at 7:48 AM
if it happens in india or america,

next day police changes statement : nothing was found,

media publishes after a week : it was a hoax,

everyone takes their share and culprit escapes, lol.

exactly same pattern followed in india where a judge was found with tons of cash.

#china #corrupt #india
January 30, 2026 at 9:03 AM
architecture and alignment is more important in agent swarms,

than number of agents, lol.

#ai #agenticai #llm #architecture
January 30, 2026 at 8:51 AM
beware,

clawdbot has some dangers too,

take care and build something great, lol.

#ai #clawdbot #security #claude
January 30, 2026 at 8:47 AM
wonders of capitalism,

this is what happens when every company in middle maximises for dollars,

at expense of human lives.

#capitalism #communism #china #america
January 30, 2026 at 8:37 AM