Sign in

Mathis obadia

@mathisob.bsky.social
44 followers 176 following 34 posts

Working at www.pubgen.ai for local newsrooms. Building askair.ai in public

PostsRepliesMedia
Mathis obadia @mathisob.bsky.social · 14/04/2025
Yes cognito was the issue for me also, the limit is on the lambda payload size for the full headers. You can maybe try to remove some parts of the cookies you don’t need in a the cloudfront function since it’s going to get called before lambda
100
Mathis obadia @mathisob.bsky.social · 14/04/2025
If I remember correctly I changed the distribution’s behaviour to not forward cookies anymore to the lambda. The issue was that the cookie string was huge but only needed in the front so removing it solved it. Let me know if that makes sense for you !
100
Mathis obadia @mathisob.bsky.social · 04/02/2025
You already have a few translations but here is my take Découvrez comment créer des interactions charmantes et des touches pleines de magie grâce à CSS, JavaScript, SVG et Canvas. Je vous dévoile toutes mes astuces ici ! I think this sounds more idiomatic
040
Mathis obadia @mathisob.bsky.social · 20/01/2025
It’s not really a refutation of your point but I think LLMs are quite good at turning a small amount of text (bullet points and short not well written sentences) into a more structured complete text. Agreed that you can’t use that directly without human intervention (proof reading at the very least)
120
Mathis obadia @mathisob.bsky.social · 23/12/2024
I’ve just released an open-source version of AskAir.ai! All core functionalities are there, and over the next few days, I’ll be working to make self-hosting as easy as possible. I’ll be using SST.dev to replace several external SaaS. Follow along here: github.com/mathisobadia...
github.com
GitHub - mathisobadia/AskAir: Open Source version of AskAir.ai
Open Source version of AskAir.ai. Contribute to mathisobadia/AskAir development by creating an account on GitHub.
010
Mathis obadia @mathisob.bsky.social · 20/12/2024
I want devin to open a PR anytime there is an issue detected in the sst console, with access to the codebase and the stack trace it should have everything it needs. Do you plan on integrating something like that ?
000
Mathis obadia @mathisob.bsky.social · 11/12/2024
lol I heard anthropic specifically had to instruct Claude to avoid vim in their SWE-bench agent - on the great @latentspacepod.bsky.social
000
Reposted by Mathis obadia
Simon Willison @simonwillison.net · 11/12/2024
Gemini 2.0 is out, and there's a ton of interesting stuff about it. From my testing it looks like Gemini 2.0 Flash may be the best currently available multi-modal model - I upgraded my LLM plugin to support that here: github.com/simonw/llm-g... Gemini 2.0 announcement: blog.google/technology/g...
github.com
Release 0.7 · simonw/llm-gemini
New Gemini 2.0 Flash model: llm -m gemini-2.0-flash-exp 'prompt goes here'. #28
312917
Mathis obadia @mathisob.bsky.social · 10/12/2024
If you ever see {"Message":"Request must be smaller than 6291456 bytes for the InvokeFunction operation"} It might mean that your headers are larger thank 10kb! But you would never guess that by looking at the error.
100
Mathis obadia @mathisob.bsky.social · 10/12/2024
Learning there’s an undocumented lambda limit thanks to a production outage is just the perfect way to spend your evening.
100
Mathis obadia @mathisob.bsky.social · 05/12/2024
Not sure if you want to add it to the scope but auth for the published blogs was very tricky to implement.
110
Mathis obadia @mathisob.bsky.social · 04/12/2024
Can you add multiplayer on top of the autocomplete everywhere to the wish list ?
020
Mathis obadia @mathisob.bsky.social · 02/12/2024
Thank you ! Exactly what I needed. Glad to see I’m personally responsible for 4 app router websites in the top 1M (with passing CWV)
110
Mathis obadia @mathisob.bsky.social · 02/12/2024
I found a website on similarweb with 1,1M rank and around 40k page views per month so I think it should be in that ballpark if we can trust their ranking
000
Mathis obadia @mathisob.bsky.social · 02/12/2024
That’s cool! Any idea how much traffic does a website need to make in order to be in the top 1M ? I’m wondering if some of our sites show up in those stats.
210
Mathis obadia @mathisob.bsky.social · 30/11/2024
I don’t know if you count it in the category of llm powered UI generation but I let cursor design my frontend and it’s mostly better than what I would have come up with by myself
130
Mathis obadia @mathisob.bsky.social · 28/11/2024
The switching to Chinese in the reasoning step is interesting. I wonder if training models to have a reasoning step in different languages changes the performance as some languages use less tokens to express the same idea.
000
Mathis obadia @mathisob.bsky.social · 25/11/2024
That’s so cool! if you choose to moderate your thread and hide a reply will this also hide it from your blog ? Also the really cool but more complicated feature would be to allow users to post replies directly from the blog.
000
Mathis obadia @mathisob.bsky.social · 25/11/2024
Really excited about this. Ideally there would be something as easy to integrate / use as Facebook comments social plugin developers.facebook.com/docs/plugins... I would be interested in helping build that if anyone wants to join.
developers.facebook.com
Comments - Social Plugins - Documentation - Meta for Developers
000
Mathis obadia @mathisob.bsky.social · 21/11/2024
Yes it seems like the way most people are doing it. I need to look more into it.
000
Mathis obadia @mathisob.bsky.social · 21/11/2024
Getting automatic / updated evals out of you production logs is a challenge ! I would be interested in learning how people have achieved that.
100
Mathis obadia @mathisob.bsky.social · 20/11/2024
Oh right so basically you’re saying you’re not replacing all control flow with LLMs but something that would have been a very complex function (a tangle of control flow logic) is now an LLM call?
220
Mathis obadia @mathisob.bsky.social · 20/11/2024
Did you listen to the @latentspacepod.bsky.social podcast with lindy.ai creator ? He talks about this and how having the control flow not be an llm improves accuracy and makes them usable for a lot of tasks otherwise very complicated to describe solely with prompts.
141
Mathis obadia @mathisob.bsky.social · 19/11/2024
sst.dev I would rather use the same thing wether it’s « just a simple app » or « turned out to be not so simple »
sst.dev
SST
010
Reposted by Mathis obadia
Vercel @vercel.com · 18/11/2024
AI SDK 4.0 • PDF support for Anthropic and Google providers • Computer use support with Anthropic's Claude • Continuation support for long text generation • New xAI Grok provider and other provider updates • Updated chatbot template vercel.com/blog/ai-sdk-...
vercel.com
AI SDK 4.0 - Vercel
Introducing PDF support, computer use, and xAI Grok provider
0702
Mathis obadia @mathisob.bsky.social · 18/11/2024
Try to guess the day where I added prefetch={false} to all the Link components in my nextjs app
Graph of daily lambda cost, cost gets divided by 6 when prefetch is disabled
020
Mathis obadia @mathisob.bsky.social · 17/11/2024
Definitely more complex compared to just disabling the transition, I just remembered @kentcdodds.com doing that ! x.com/kentcdodds/s...
x.com
x.com
100
Mathis obadia @mathisob.bsky.social · 17/11/2024
What about using a theme cookie so it can be server rendered with the right value ?
100
Mathis obadia @mathisob.bsky.social · 16/11/2024
Too bad they don’t do vision, for pdfs at least I found that nothing beats using a screenshot and a vision model in addition to whatever text you have extracted from the pdf to get accurate results.
030
Mathis obadia @mathisob.bsky.social · 15/11/2024
How do you evaluate changes to your prompts / models without just testing random queries you know have issues and are trying to improve. Do people really have automated eval sets?
100
Reposted by Mathis obadia
Simon Willison @simonwillison.net · 15/11/2024
Very impressed by Recraft AI - a new image generation service that can generate editable vector graphics that you can export as SVG This seems massively more useful than tools that can only output raster graphics simonwillison.net/2024/Nov/15/...
simonwillison.net
Recraft V3
Recraft are a generative AI design tool startup based out of London who released their v3 model a few weeks ago. It's currently sat at the top of the [Artificial …
1016522
Mathis obadia @mathisob.bsky.social · 13/11/2024
I just spent a few hours trying to tweak a prompt to make it work with gpt4o mini but never got it to give consistent results even in the simple cases. I guess I’ll stick to gpt4o and revisit when my OpenAI bill gets higher than my cloudwatch bill (never)
000
Mathis obadia @mathisob.bsky.social · 10/11/2024
Yes! Also recently learned a similar trick can be used to get better results with the structured output mode which I believe is not yet supported with o1 anyway simonwillison.net/2024/Aug/7/b...
simonwillison.net
Braggoscope Prompts
Matt Webb's [Braggoscope](https://www.braggoscope.com/) ([previously](https://simonwillison.net/2023/Feb/13/braggoscope/)) is an alternative way to browse the archive's of the BBC's long-running radio...
000
Mathis obadia @mathisob.bsky.social · 10/11/2024
That’s basically what gpt4 o1 is so if you use it you shouldn’t need to include it anymore
110
Mathis obadia @mathisob.bsky.social · 09/11/2024
I was wondering if I could use gpt-4o-mini as a reranker and obviously found a post by the great @simonwillison.net with included prompt. Only issue I found is it will not score all entries even if you ask nicely 😂 simonwillison.net/2024/Aug/11/...
simonwillison.net
Using gpt-4o-mini as a reranker
Tip from David Zhang: "using gpt-4-mini as a reranker gives you better results, and now with strict mode it's just as reliable as any other reranker model". David's code here …
000
Mathis obadia @mathisob.bsky.social · 08/11/2024
This is the exact process I used to decide the pricing of my side project askair.ai. If you think it's too expensive go complain to claude-3.5-sonnet
Screenshot of the cursor editor, that just ai generated a pricing table
010
Mathis obadia @mathisob.bsky.social · 04/11/2024
Working on my side project and wondering if I should build that feature that a single user requested or work on SEO to hopefully get more user requests
110