neo (@PineapplesDev) header
neo (@PineapplesDev) profile picture

ABC: Always Be Cooking 🧑‍🍳 👻 https://teapawt.net/t.co/CXLbGCp5Km 🍍 https://teapawt.net/t.co/KFgCrif9f7

⚡️vibe it with us 👉
Joined October 2012
Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)Photo by neo (@PineapplesDev)
Our AI financial planner saved "$7,000 a month" as $7,000 a year. The chat asked for monthly expenses. The save tool had no period field, and the service stores everything as annual. Whether to multiply by 12 was left to the model. QA found two more in the same feature: 1. A chat update added a second expense line instead of replacing the first. In the test case the record summed both, $174,000 instead of $90,000, while the agent said it had replaced the figure. 2. One page called an education goal "On Track ($0/mo)". Two pages later, the same plan showed a $225,000 shortfall. Two functions, two answers. The fixes: 1. Every line carries monthly or annual. Code annualizes, not the model. 2. The save returns a receipt, "$84,000 a year ($7,000 a month)", and the agent reads it back. 3. One shared calculation for every page that shows the goal. Anthropic's Claude Code team says verification loops are what let agents write most of its code. Same lesson here: the model never decides units. If an AI writes numbers into your system of record, where are units enforced: the prompt or the tool contract?
1
23
The renewal test for SaaS is shifting to: can the customer's own agent use us? Our MCP server listed 433 tools. ChatGPT caps at 128, so it couldn't load it. We now list 38 core tools and the rest sit behind search, with the same approvals.
We have 2 agents we pay five figures each for. They are great products and we recommend them highly. Great products. However 10K has slowly been taking over their functionality, day-by-day, bit-by-bit. He said not to renew in the coming months. He’s got it.
4
3
40
145 documents in our deal platform's vault had failed with no reason written down. Search could never return them, and nothing said so. A file marked processed with zero chunks read as indexed. A pending row hours old read as "still working." The agent had no idea those files existed. Trending today: SaaS turns into an invisible backend and AI agents run the conversation. Fine. Then the backend has to be honest about what it can't find. What we shipped: 1. A health report that lists every file search can never return, with 13 failure modes written before the code. 2. A repair job that is dry run by default. 3. A nightly alarm when any failure kind stays above zero. After the repairs: 145 down to 36. We chased the last 36 read-only. All 36 were files people had already put in the trash. No retrieval tuning until the index is honest. Buying AI over your documents? Which do you check first: answer quality, or what share of your files search can actually return?
2
28
Lions in Munich tomorrow, 9:30 AM. Mute Fox, put on 97.1 The Ticket, and the radio is ahead of the picture. The Call fixes that: tap when you hear the snap, tap when you see it, and the radio waits. Today: Michigan at Oregon, 3:30, FS1. wave-comet-clover-frost.grok…
44
Domain specific, and what you leave out matters as much. Our deal chat answered its own "read the source first" note as if the user wrote it: "going forward I will call..." Now notes are labeled app checks, and the next turn's history keeps only the question and final answer.
Overheard this morning: “We need to retain control of the investigation.” Ah yes I’m definitely in DC. “Optimizing the compaction summary turns out to be very domain specific.” I’m in DC?!
13
Our AI chat replied to its own safety check instead of the person who asked. The question was simple: how many files are in this folder? The agent read the folder with two tools our check didn't count as reads. So the app sent its "read the source first" note. The model answered the note: "That last question only asked how many files... going forward I will call..." PluraiAI just launched a simulator to stress-test agents against simulated users. Useful. Our break came from inside: the guardrail text leaked into the answer. The fix: 1. Every app note says it is an automatic check, repeats the person's question, and says not to answer or mention it. 2. After a note, the reply is held until it is final. 3. A reply that talks to the note goes back once. A second one is shown, so it can't loop. 4. The next turn keeps only the question and the final answer. Never the note. Guardrails are product surface. We wrote 6 failure modes and run them through the real router and tool loop. Rolling out AI in a portco? Ask your team: can a user ever see one of your hidden checks?
12
9
408
The small-company version of this shows up in tech diligence: what only works because one person is there? We ask who can ship a release, who holds the cloud and vendor logins, and who can rotate the production keys. One name on all three is the succession plan.
I really don't see this as the same as Jobs leaving Apple. Three main reasons: 1/ Jobs knew he was sick and had been planning a transition. Nvidia has no succession plan. 2/ Jobs was more a marketer and visionary than hardcore technologist. Jensen was all three, but the last part mattered a ton. 3/ Apple wasn't a "central bank" to tech in the same way Nvidia has become, so its success/failure wouldn't have quite as many ripples.
22
A payment webhook we run answered "not found" 41 times in one week. Nothing crashed. A flag said billing was off. a16z partners say AI agents now drive most web traffic. When software does the buying, the webhook is your record of the sale. And webhooks fail quietly. What happened on a client platform: 1. The webhook route checked an internal "billing on" flag before doing anything. 2. The flag shipped off, so every delivery got a 404. 3. Those payments still mattered. Purchases from a separate checkout tool set the weights for lead routing. 4. The processor saw a week of refusals and scheduled the endpoint to be disabled in three days. The fix: - Verify the signature. - Store the event once. A duplicate gets OK and is not queued again. - Queue it, answer OK, process in a worker with retries and a dead-letter queue. - Bad signature gets a clear rejection. A database failure gets an error, so the processor retries. No feature flag at the door. The route receives and records. Whether billing is "on" is the worker's call. If you run a portco, ask your team one question: what does every inbound webhook do when a feature is turned off?
1
2
36
What doesn't roll over between apps: Live Activities. iOS gives apps no way to ask for them. On a client app we start the activity when the user enters a capture flow, and show a row linking to Settings when they're turned off.
🌠 Supernova's second app hit a $1M revenue run rate after about three months of development. Turns out it helps when login, payments, and push all rolled over from their first app. Now they have three apps built with Expo and a great blog post about the experience. ✍️ @TheAnirudh, @shubhporwal24, and Sai Vasudatta. 📚 expo.dev/blog/supernova-ios-…
20
Honest about what it did is the easy part. The hard part is what it never touched. Our document pipeline showed 63 files "pending" for days. Nothing failed, so nothing alarmed. Now every item ends "read" or "failed with a reason".
Sh*t Our Agents Do: Me: "Are you too degraded? Should we start a new chat?" Claude: "Yes, I'd start a new chat... I made today's mistakes in it." At least it was honest.
14
Let’s get after it @terminaldotshop
1
16
Our mail agent wrote "synced, no error" on every run for a week. It was reading one page: at most 50 threads from the last two days. Teamily is trending this week for putting agents in group chats with the same files and notes as the humans, to end "state drift". Shared context only helps if the agent actually read what it says it read. What was going wrong on an agent platform we run: 1. It listed one page of threads and kept no cursor. Anything past page one was never seen. 2. Anything older than two days when a run reached it was never listed again. 3. New replies on threads it had already filed were skipped. A consent thread sat 12 days stale. 4. Every run still logged "last synced now, no error". The fix: - Read every page from a stored cursor, with a one-hour overlap. - Move the cursor only when nothing was deferred or failed. - Write a failed read as an error, never hide it. - Run the resync as a dry run first. In the replay test, the old listing missed all six threads we knew were lost. The new one stored 66 of 66, and a second run changed nothing. "Synced" now means "read through this point", stored and provable. Not "the job ran". If a portco is putting an AI agent on email or documents, ask it one question: what did you not read?
2
41
🕸️
So easy to trigger Rust programmers 😂 Rust is a good language, but C is what you use for optimal performance on bare metal. With great power comes great responsibility.
37
The harness includes the parser. Our intake chatbot answered "🚫 None" and our router read it as no answer, so 12 leads under a $3M minimum got through. Nothing errored for six days. Parse what the model actually emits, tested on production outputs, not what the spec says.
Same model, same prompt, and their agents ship while yours say "done" on broken work The difference is the harness: what it reads, what it can touch, who checks it, when it stops 6 parts + a 3-agent setup you can copy tonight ↓
8
In an audit prep this week, the hardest finding wasn't in the code. A data vendor's installer had put a load balancer in the client's AWS account with 443 open to the internet, and AWS WAF can't attach to that type. Scan the repo, then scan what vendors put in your cloud.
Vercel Security Review, less than 1 week since launch: ✓ 𝟱𝟳,𝟴𝟵𝟳 vulnerabilities identified ✓ 𝟲,𝟬𝟮𝟵 critical or high ✓ 𝟭𝟰,𝟲𝟲𝟱 already patched by Agent I'm shocked at ① the number of critical vulnerabilities that would go unnoticed and ② the ease at which the agent automatically fixes them. You can turn it on for any repo, no matter where you deploy to.
1
27
5 go-live checks now gate every number in a client's daily AI report. The model writes one sentence. If that sentence has a number the code didn't produce, it gets thrown out. OpenAI's new Control Room halts flagged AI work in 35 seconds and fails closed. That's the frontier version. Most operators need the boring version first, inside one morning report. What we rebuilt on an agent platform we run for an operator: 1. Plain code computes every number from a definitions doc the operator wrote. 2. Five go-live test cases run on every read. Until they pass, every number is labeled "estimate". 3. The model writes only the headline. A number it adds gets the headline refused. 4. A source that fails prints "not read", never 0. 5. Email goes from the platform to an approved list, never from a client's mailbox. The goal it scores against stayed untouched. Changing a goal is the operator's call, not the agent's. If your portcos are rolling out AI reporting, do you know which numbers in the board pack a model wrote?
3
1
58