37,617 results · Most relevant
A single-node database works fine at ~10 GB but not at 10 TB, 100 TB, or a petabyte. Use sharding to spread data and queries across many servers, and put a proxy (e.g., VTgate) between the application and shards to intelligently distribute requests. Do not use sharding by itself: within each shard run a redundant primary-plus-replica system so the shard can self-heal. The same spreading principle applies to many other areas of the stack.
Avoid combining all services (Postgres OLTP, analytics, queuing, job system, caching) into one system, even though it is convenient. It adds complexity, breaks failure isolation (if one service goes down, everything goes down), and creates resource contention — e.g., a burst of queue jobs forces scaling queues when nothing else needs to. Keep services decoupled so each component can be scaled independently.
Design infrastructure so a single component failure does not take down the whole system: isolate the data plane (databases and critical data) from control planes, observability pipelines, analytics frameworks, apps, and MCPs. Components deployed more frequently are more likely to fail, so ensure a bad control-plane deploy cannot impact access to the data plane or database.
Popular databases like MySQL, Postgres, and SQLite handle overload poorly — too many queries, too much CPU, or exhausted RAM cache can crash them, taking down the server and impacting all users. Implement backpressure: when resource thresholds are exceeded, fail requests, deny queries, and deny connections so existing users or some traffic stay healthy while other traffic degrades gracefully. The rest of the stack must be built to support this. Backpressure applies across the stack but is arguably most important in the database.
Varun states evals are the starting point and ending point for any agent architecture, and that the way they think about evals is in five different levels. He names several: trace the path (how did the AI come up with this solution?), score the AI like an engineer based on its path compared against how the best engineer would perform the investigation, and calibrate the AI's confidence level. He advises building all evals systematically into one platform and constantly pushing the architecture against them whenever there's a new model, new use case, or architecture change.
Varun describes pillar 3 as causal reasoning: the root cause must always be based on a causal chain of evidence capturing the exact, precise steps that happened that led to the issue. If the chain of evidence cannot be established, give out the information with a low confidence level and point the user in a different direction. He says this principle is true for any domain-specific AI, not just Resolve AI.
Varun describes pillar 2 as context engineering. He discourages minimizing it to the database or execution solution — graph RAG, knowledge graph, etc. — calling those implementation details. The goal is determining the precise amount of context AI needs to solve the specific problem, and it is most often a combination of techniques rather than a single solution. He also advises defining very precise tool calls (log queries, metrics, dashboards, code queries) so you don't burn through tokens quickly when making multiple tool calls.
Marc Klingen describes the pattern Langfuse's best users follow when building agents: collect production signals, track them alongside detailed execution traces, and then run agents on a loop — usually a cron job every day or every week, whenever there's a batch of interesting user data — to act on it.
Marc Klingen recommends using AI to propose fixes to agents when you have production traces, eval criteria, and datasets of how to reproduce issues found in production. When you see usual error patterns, reproduce them and propose new eval criteria on the offline datasets — for example, adding an eval criterion that a customer support agent answers in the same language as the user's question. Teams also use AI in their loops to maintain eval criteria and maintain datasets.
Alzheimer's pathology does not necessarily lead to dementia; 20-30% of people with the pathology never develop dementia. Cognitive resilience is supported by physical exercise, controlled blood pressure, avoiding diabetes, controlled cholesterol, a sense of purpose, social connection, and curiosity, with the psychological factors almost as important as the physical ones.
Anchors are what the training is based around: one box cleans serve as the primary peak-velocity indicator of athlete readiness and twitchiness, the back squat is the leg strength anchor whose numbers dictate how heavy to go on other lower body lifts, the bench press is the upper body anchor, and the flying 10 sprint (measured with lasers, not VBT) is the anchor for max speed, targeting 22+ miles per hour.
Use VBT bar speed data to catch fatigue and prevent injury before it happens instead of adding extra sets on feel alone. Skip an extra set when bar speed drops below the target velocity for that lift, even if the athlete feels strong enough to do it.
Consequential creator advice requires independent verification before Today can promote it.
When Mike pushed back — 'Then why are you having me talking about this bullshit? Stop, motherfucker. Stop.' — Theo was in a tight spot somewhere between a joke and a warning. If he immediately changed the subject, he risks retreating too far and coming across as weak; if he kept asking personal questions, he risks escalating to the point of no return. Instead, Theo zoomed out a level on the same topic, which takes the pressure off without totally shutting the topic down. As a result, Mike went even deeper.
The host lays out the small, subtle signs to pay attention to: her asking questions, responding positively to displayed intent, staying planted and holding eye contact, staying in the interaction, and asking your name. These differ from how men display interest, and recognizing them gives you confidence to move the interaction forward — though it's not yet time to ask for a date or number.
When someone withdraws, the natural thing to do is try to get them to open up — and if someone actually wants to connect but is just being shy, that can work. But if someone is playing an attachment-style power game, trying to draw them out with questions won't work: they learn that withdrawal gives them control over the relationship. If you get angry at someone using withdrawal as a tactic, they learn that the tactic gives them power. Instead of chasing someone playing a withdrawal power game, do nothing — literally. The ultimate form of control is someone who is so relaxed that they don't have to do anything, not someone laying down the law. When someone finally feels that you aren't going to chase them, that is when they finally re-engage.
The speaker starts the conversation in an innocent, safe position and closes the distance gradually as flirtation builds, eventually getting close enough that he could have kissed her.
Once intent is established and she's comfortable standing close, the speaker shifts to vibing, then a mutual get-to-know-you phase (work, fun, living situation, upcoming plans), then moves toward the phone number and setting up a meetup.
The speaker argues that when a woman peels away to order, you should let the conversation breathe and then reopen it as a simple extension of the initial conversation rather than assuming it's over.
The speaker describes moving the conversation from ethnicity to shared preferences to personal compliments, inching toward a flirtatious direction without flipping a switch.
By saying 'I like Albanian people' rather than 'I like you,' the speaker prompts the woman to qualify herself by explaining why Albanians are the best, which he frames as her investing in the interaction.
Choose sources → to get a library of your own and a daily brief.