ES

Einsia ✓

@EinsiaAI

We're teaching AI the work the world's experts actually do.

710Followers

6Following

78Posts total

2.9MViews on collected posts

Últimas publicaciones

Einsia ✓@EinsiaAI16 Sep

@EinsiaAI This could remove so much repeated explaining between teammates. 28 views · 6 likes · 0 reposts · 0 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

AgentGit preserves the process. Six things you can do with it: 1. Save and restore sessions 2. Hand off complete context 3. Share the full process 4. Find precedents 5. Control agents remotely 6. Collaborate on live sessions Here’s what that looks like: https://t.co/cNObutdwdD

1.3K views · 20 likes · 3 reposts · 11 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

An AI agent spends hours on a task. Why should all that work disappear when someone else takes over? Einsia AI’s answer is AgentGit—an open-source platform for collaborating on agent sessions, so work can be saved, handed off, and continued by the next person. Explore how https

1:39

706.4K views · 615 likes · 97 reposts · 135 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

1/ Handoff 🤝 A teammate needs to take over your work. Instead of scheduling a walkthrough—or, or trusting an AI summary that misses the important details—push the session. They pull the full context and continue from the exact point where you stopped. agit push → agit run htt

0:05

504 views · 9 likes · 1 reposts · 2 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

2/ Sharing🔗 Someone asks: “How did you get the agent to do that?” Don’t send them a prompt and 30 screenshots. Send one read-only link containing every turn, tool call, and test result. No CLI or account required. Encrypted by default, with optional expiry and view limits. ht

0:04

337 views · 7 likes · 1 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

3/ Precedent 🧠 Your agent is stuck on an error someone else already solved last week. Through MCP, it can search the AgentGit Hub for previous sessions—including approaches that worked and approaches that clearly failed. Past agent work becomes reusable context, not a private

0:11

285 views · 7 likes · 1 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

4/ Remote control 📱 Leave your desk without abandoning the task. The agent keeps running on your own machine while you monitor progress and send instructions from your phone. Share the session as a workspace, and your team can follow the same live run—while the owner keeps htt

0:07

261 views · 6 likes · 1 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

AgentGit is local-first. Your history stays on your machine unless you publish it. Sharing is private by default, secrets are scanned before publishing, and public copies can redact keys, paths, usernames, and hostnames. 253 views · 7 likes · 1 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI16 Sep

Save the session. Hand off the context. Keep the work going. The session captures how the work was done—what worked, what failed, and why. When agents write the code, their work needs version control too. Don’t start over. Continue where the last agent left off. AgentGit now 272 views · 6 likes · 1 reposts · 0 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

@EinsiaAI We should not be working on RSI or research that could enable RSI. By default RSI probably results in AIs that are uncontrollably smart and are misaligned with human interests. Such AIs likely kill all humans. 128 views · 10 likes · 0 reposts · 0 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

6/ The takeaway: Today’s agents can only occasionally do real algorithmic design—not reliably. Even at the highest reasoning tier, the average score was only 0.196: about 11% of the way from the 0.10 baseline to the theoretical optimum of 1.0. The question isn’t whether agents 442 views · 16 likes · 0 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

5/ More reasoning means more exploration: 13× more code, 10× more tokens, 4× more evals. But gains lag behind: score improves only ~2×, reaching just 0.196 at max effort. https://t.co/SfajILjlWp 498 views · 13 likes · 0 reposts · 2 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

4/ What agents changed mattered most: Of 263 submissions that made a change: -141 changed only the run settings -122 changed the learning method itself In terms of score, 0.226 for learning-method changes vs. 0.126 for run-settings changes. Changing algorithms performed ht 565 views · 13 likes · 0 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

3/ Across 29 model × harness × effort configurations: Scores range from 0.065 to 0.288, with an average of just 0.166. Opus 5 is the best. GPT, Claude, and Kimi stay fairly close overall. Median exploration cost per task jumps from $1.69 to $34.60—but spending more doesn’t htt 799 views · 16 likes · 0 reposts · 2 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

2/ AI4AI-Bench: 10 research repos, spanning 10 algorithm families. Agents get the repo, starting artifacts, and a proxy metric, then 4 hours on one B300 to improve the training algorithm. They submit code only—no trained weights or cached state. The code is rerun from scratch 1.8K views · 18 likes · 0 reposts · 1 replies Open on X →

Einsia ✓@EinsiaAI21 Aug

1/ Recursive self-improvement (RSI) depends on agents improving how AI systems are trained —not just tuning hyperparameters, but improving the training algorithm itself. We tested this directly with AI4AI-Bench: 10 real research repositories spanning 10 distinct algorithm https 2.1M views · 667 likes · 134 reposts · 116 replies Open on X →

Frente a cuentas del mismo tamaño

16 publicaciones de los últimos 90 días, junto al rango de under 10K seguidores. llega a menos gente que sus pares, pero los implica mucho más.

Visualizaciones medianas470esta cuenta4 175mediana de under 10K

Alcance, %66.20%esta cuenta250.82%mediana de under 10K

Interacción, %2.65%esta cuenta1.41%mediana de under 10K

Métrica	Esta cuenta	Mediana de under 10K	Proporción
Visualizaciones medianas por publicación	470	4 175	0.11×
Alcance (visualizaciones ÷ seguidores)	66.20%	2.5× audience	0.26×
Tasa de interacción	2.65%	1.41%	1.88×

Otras cuentas de este rango → Comparar con otra cuenta → Cómo se construyen estas referencias →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

79921 Aug

565

498

442

128

27216 Sep

253

261

285

337

504

706.4K

1.3K

28

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

2.25%21 Aug

2.48%

3.01%

3.85%

7.81%

2.57%16 Sep

3.56%

3.07%

3.16%

2.67%

2.38%

0.15%

2.62%

21.43%

Reactions — likes, reposts, replies and quotes — divided by views. Median for under 10K accounts is 1.41%.

What the audience does

Likes48.4%1 436 in total

Reposts8.1%240 in total

Replies9.3%275 in total

Quotes11.0%325 in total

Bookmarks23.3%691 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Cuentas similares