Get rejected for more AWS credits and send them a support request to try to get them to budge #watchdog

Submit to AWS startups showcase for a free backlink: aws.amazon.com/startups/showc… (support appreciated if you have an AWS account and a moment!) #watchdog

Cut over to the new AI model which is 300% faster to respond and supports video moderation! Will let this bake for a week and cut back to the old model if there are any issues: x.com/ben_makes_stuff/status/… #watchdog

Get accepted to NVIDIA Inception and apply for more AWS credits so I don't have to move everything to gcloud/hetzner/etc (would be distracting, I need to focus on other stuff right now) #watchdog

See #watchdog moderating in Brazilian Portuguese 🇧🇷 using customized branding and message templates (both features recently shipped, customers are using them already)

Update Q&A prompt to explicitly disallow financial/legal/medical advice and provide disclaimers, because some customers are abusing #watchdog and uploading docs that tell their users to buy a shitcoin

Distinguish between error types (internal and user error), add separate alert messages for each when clicked, and update permission error logs to warn because they're not actionable by me #watchdog

Left align labels in modals because it looks better, particularly for the checkbox #watchdog

Significantly improve X login reliability by persisting user session data on browser close - no longer getting "suspicious login detected" or temp banned from logging in, and no longer have to log in at all after the first login 😎 #watchdog

See #watchdog automatically answering questions correctly from a customer's chat - very cool to see that people are OK with this, I expected some blowback. On the other hand, it's basically recommending people hold a coin which could be questionable 😅

Fix X integration by restarting the bot repeatedly until it succeeded after a few hours, and start investigating residential proxies to avoid suspicious login detection #watchdog

Submit to like 40 directories, let's see if it leads to mOaR BaCkLiNkS #watchdog

Update onboarding form to be more specific about which communities are allowed #watchdog

Hit highest ever MRR (up 31.7% MoM) after adding free trials last month: x.com/ben_makes_stuff/status/… #watchdog

Fix bug where the answer to a Q&A question would be prefixed with "{\"answer_to_question\": ..." because OpenAI structured outputs are fucked and randomly embed JSON inside of a string prop sometimes #watchdog

Ship message templates for alerts, configurable alert cleanup behavior, and some fixes to the AI model for Q&A: x.com/ben_makes_stuff/status/… #watchdog

Post on X and LinkedIn for failed scam friday: x.com/ben_makes_stuff/status/… and www.linkedin.com/posts/benmak… #watchdog

Big refactor to correctly merge punishments across multiple rule violations as I realized I wasn't doing this right some of the time, testing it now #watchdog

[UI Done - backend pending] Allow customers to customize the "Alert Channel" message with templating to handle several use cases from internationalization, customers simply wanting different messaging, etc #watchdog

Get told to go to bed by a customer who reported a bug to me that I fixed in 5 mins before going to sleep (how nice, they care about my sanity) #watchdog