Pinned Tweet
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
40
53
774
73,186
Daanish Khazi retweeted
opsd and its variants don’t work, no matter what they are biased and bias kills llms, please please please stop working on it, so many other great directions to explore that aren’t opsd
Meta's team proposes DCE + SRCL as an alternative to on-policy self-distillation (OPSD).
Achieved significant gains: 30.76% avg acc -> 65.97% on Qwen3-8B
arxiv.org/abs/2609.30652
9
13
147
14,253
Daanish Khazi retweeted
average OPSD replication effort: "it doesn't really work but wouldn't it be sick if it did?"
12
8
239
15,908
Daanish Khazi retweeted
If you’ve built agents or envs that work with word/excel you know the pain of aligning agents to work with docs
Paper is promising as a drop-in replacement to fix a lot of those issues
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
1
6
1,777
If you're trying out the packages feel free to use this skills to prime your agent (packages aren't yet in distribution)
github.com/paper-instruments…
1
17
2,312
Daanish Khazi retweeted
I’ve always complained about how insanely broken AI is for the entire office suite
Happy to see @paperinstr fixing this
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
1
12
796
Daanish Khazi retweeted
Some very cool shit the @paperinstr team is working on. You should probably check it out.
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
1
9
1,101
Daanish Khazi retweeted
this is super helpful for anyone working in legal/professional service domains. shoutout @thegavinbains and team for launching this!
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
2
11
587
Daanish Khazi retweeted
it surprised me how often ai agents corrupted @Microsoft office files, even when running simple tasks.
turns out the packages they use haven’t been updated in years and in some cases fully rewrite a file’s XML on save, destroying or invalidating special features like charts.
paper office is part of our efforts to improve agents’ abilities to create/edit and maintain these files
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
2
3
15
432
Daanish Khazi retweeted
Your agents will be much happier when you use @paperinstr 📄
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
1
8
705
Daanish Khazi retweeted
When I was at Harvey working with Excel, Powerpoint and Word files was so painful that we often resorted to converting the document into a PDF and using a mixture of OCR/image models to ingest data.
It was incredibly painful.
Excited to see this work by @paperinstr to hopefully remedy those issues and move one step closer to turning the $MSFT office suite into a programmable layer for agents
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
3
12
1,644
Agents are using legacy Python packages to edit Word, Excel, and PowerPoint files, hitting API gaps, dropping into raw OOXML, and creating silent corruptions.
We fix that today with OSS Paper Office:
40
53
774
73,186
Update every prior on post-training scaling laws:
They hit SoTA with ~45k RL tasks, nearly all synthetic, with $2.6m of Chinese compute (~$1m in US)
That's a 2023 era pre-seed! Human data is stifling the US frontier, and neolabs are prematurely scaling compute.
MiMo-V2.6-Pro debuts as the top open weights model on the Artificial Analysis Intelligence Index (46). At $0.13 per Intelligence Index task, it lands on the Intelligence vs. Cost per Task Pareto frontier
@Xiaomi has just released MiMo-V2.6-Pro, an open weights model with major advances in intelligence over its predecessor, MiMo-V2.5-Pro (Intelligence Index: 26). Despite the improvement, it retains the same attractive pricing at $0.435 per 1M input tokens (with a 99% cache-hit discount) and $0.87 per 1M output tokens. This makes MiMo-V2.6-Pro one of the most cost-efficient models to deploy.
MiMo-V2.6-Pro is an MoE model with 1.02T total parameters and 42B active parameters. Stay tuned for additional analysis of the model.
Check out MiMo-V2.6-Pro full benchmarking breakdown here: artificialanalysis.ai
10
20
389
43,380
Muse+Shopify completely unbundles Amazon
For 2 years I operated an F&B brand and routing traffic to our Shopify front cost more than Amazon's take
Math here is tricky, but, on balance, I think advertising to agents is more efficient than either direct intent or TOF
We are excited to announce we are partnering deeply with Muse to enable agentic checkout with Shop Pay on all Shopify stores, offering people an easy and delightful way to shop and check out with Muse.
2
5
1,328