I find @chatgpt goals unusable unless you tell it to "stay focussed, and do not go into self review loops or scope expansion". Without such an expansion good luck getting to the end of a goal without spending all your weekly tokens.
1
62
Codex goals get into this never ending cycle making you think that it's doing real work and fear that stopping it will leave things in a buggy state while without goals using orchestration patterns or when using Claude Code, things get done without breaking a sweat.
1
50
Codex side chats are so useful to refer back to. Why expire them? @thsottiaux
40
The 3-5 minute compacting conversation step of @claudeai is such a productivity killer especially since I barely get 10-15 minutes of real work done by the agent. I hope this doesn't become the new normal.
113
I have been consistently having codex (GPT 5.4 medium) review claude codes (Opus 4.6 1M medium) solution approach and code, and then the other way around. Codex has consistently produced better answers that claude admits to when I share codex response to claude. Just over the past 2 days Codex has made over 6 non-trivial corrections or come up with improved plans over Claudes. In the example below, the path Claude Code was guiding me would have been entirely wrong and costly. Codex was quick to independently come up with an elegant solution. I'm loving the Claude ecosystem but finding myself heavily leading towards codex for coding tasks. @thsottiaux
2
1
207
Another one
81
Claude even kinda hinted to having Codex to the job
47
Codex + Playwright currently > Claude + Playwright. For UI testing after bouncing between the two I find Codex is smarter with the overall orchestration
41
GPT 5.4 xhigh is ridiculously good. Been using Opus 4.6 as my daily driver but threw a hard problem that required wide changes across my codebase at GPT-5.4 xhigh and it solved it very robustly. There’s something different about the way if reasons and shares reasoning along the way as feedback for long or complex tasks. Super impressive!
With GPT-5.4 out. What should Codex ship or improve next?
118
A word of caution for companies looking to adopt @claudeai we decided to have Claude be our organization AI strategy and signed up several of our members on the Team plan. Within a week all members got an email saying that our Claude organization account was banned without any explanation. At this point teammates hardly started using Claude. We submitted an appeals form several times. Over a week goes by, no response. I pinged @amorriscode and he checked internally and was able to have them restore the Claude Code account and we got a message saying it was banned by mistake. However our organization was still banned on platform.claude.com and the appeal to reinstate it was denied without any explanation. As an org in the lifesciences space with fortune 500 customers we had plans to start using their API heavily and use their lifesciences connectors as well. But just like that, our whole AI strategy was derailed with no explanation from the Claude team. @bcherny @trq212 @felixrieseberg Things could have been a lot worse if we had invested heavily and had artifacts and knowledge in Claude Cowork, Chat and suddenly lost all access to it and our customers impacted from it as the API access no longer works. I know that @thsottiaux and @openai team are doing great work with Codex / ChatGPT and I'm optimistic they will close the gap on the areas that matter most to our organization. We will start using them for API access right away with a plan to phase out Claude. It's way too risky to have an organization like Claude abruptly and without explanation ban accounts.
1
9
1,567
Replying to @pganalyze
@pganalyze is there a support email? I sent a question regarding a setup issue with Azure Postgres. Would really like to have it resolved within my 14 day trial
3
233
2023/10/25 07:59:16 E [server1] ERROR - Could not get logs through Azure Event Hub: failed to set up workload identity Azure credentials: no client ID specified. Check pod configuration or set ClientID in the options
220
I'm evaluating pganalyze with Azure Postgres and have followed all the steps noted in the documentation however I'm still getting this error when trying to enable Log Insights. I have set up the Managed Identity on the linux VM running pganalyzer as noted in the documentation.
162
Replying to @RewindAI
@RewindAI was surprised to see the Rewind captures audio from the mic when Audio Capture is enabled even when the mic is not being used as part of an application. Which means my laptop sitting idle was capturing my voice from a conversation I was having on a phone. Not good!
1
207
I was expecting audio to be captured only when the mic is activated as part of an application. Aslo can you add a keyboard shortcut to enable / disable audio capture?
188
Just tried out @RewindAI after holding out a bit due to the price and wondering if I actually need it. Blown away after using it for just a few hours. Still in the 30 day trial and not sure if I’ll continue beyond the trial but there’s no denying this is phenomenal technology
1
2
406
It sure is going to be hard to stop using it after the 30 day trial. One of those things that you don’t think you need it until you realize what you’ve been missing out on
1
101
Replying to @intellijidea
@intellijidea Finding that the project pane always has the directories collapsed upon reopening a project really inconvenient. This wasn’t the case several releases prior. Please tell me there’s a setting to have directory expansion state preserved upon reopening a project
3
@intellijidea thanks for the response. The JRebel plug-in might have been the culprit. I disabled it and things started working. I re-enabled it and things still work now. Will provide logs if I run into the issue again
1
The issue still persists with the use of the @JRebel_Java IntelliJ plugin. JRebel team, can you please look into this?
Replying to @raindrop_io
@raindrop_io the latest release have regressed in functionality. Previosuly when clicking the Raindrop icon on the toolbar it would show the browser view allowing me to both, add a new bookmark as well as switch to a new boormark. Now I have to click the search icon.
1
Also very unfriendly that it doesn’t default to the last used bookmarked category. And KeyBoard arrow up / down navigation doesn’t work.
Replying to @launchbar
@launchbar @BinaryNights LaunchBar currently supports “Get Finder Selection (Command-G): Transfers the current Finder selection to LaunchBar.” Can LaunchBar please support the option of picking up ForkLift path via Command-G?
1
I do have my default FileViewer changed to ForkLift following these instructions. binarynights.com/manual#file… Could LaunchBar look at this setting to determine which app to use to get the path when Command-G is pressed?