From Systems of Record to Systems of Context — Omri Bruchim & Tomer Ast, monday.com

频道: AI Engineer
视频: https://www.youtube.com/watch?v=Btk8wDUVs74
原文语言: en
统计: 共 34 轮

[0:01]

[music] Um, hey everyone. Uh, we are super excited to be here. Thanks for having us. Uh, today we're going to talk about how we shift monday.com from a system of record uh, into a system of context. And honestly the title tell the whole story uh in just a single line. Uh for decades we build uh software that record what's happened um every task, every document,

[音乐] 嗨，大家好。我们非常高兴能来到这里，感谢邀请。今天我们要讲的是，我们如何把 monday.com 从一个「记录系统」（system of record）转变成一个「上下文系统」（system of context）。老实说，这个标题一句话就把整个故事讲完了。几十年来，我们做的软件都是在记录已经发生过的事——每个任务、每份文档，

[0:38]

every message um every status update um just put into the record. What we want to talk today is like take it step further. Uh we want software that actually understand the connection between them. So I want to start from a simple question that each one of us ask himself every morning. Um what should I focus on right now? Um it sounds almost trivial but uh to be honest with yourself if you ask your agent whether Gemini GPT or even cloud um if you ever

每条消息、每次状态更新，统统写进记录里。而我们今天想聊的，是再往前走一步。我们想要的软件，是真正能理解这些东西之间关联的软件。所以我想从一个很简单的问题开始，一个我们每天早上都会问自己的问题：我现在到底该专注在什么事情上？这听起来几乎太琐碎了，但说实话，如果你去问你的 agent——不管是 Gemini、GPT，还是 Claude——如果你曾经

[1:13]

typed this question you probably got list of bullets um not related to each other. um list of like item dressed up like a a confident paragraph. Um but but it's not really um um connected to what you're working on. Actually, I tested last week and uh Cloud asked me to go to the gym. I don't know if it's a compliment or not, but uh this is what he suggested. Um and and what really make it so frustrating is like your assistant um

输入过这个问题，你多半会得到一串彼此毫无关联的要点，一堆被包装成自信满满的段落的条目。但那跟你手头真正在做的事其实并没有关系。我上周还真试了一下，Claude 建议我去健身房。我不知道这算不算是一种夸奖，反正它就是这么建议的。而真正让人抓狂的地方在于，你的助手明明

[1:46]

have all this data. It has all the boards, the tasks, the emails, the Slack messages, everything that you ever touched if you connect it. Um, but it still can't really answer it. Um, it has all the data. Um, but it has zero understanding. Um, so this is the real challenge we are facing. Um, this is the heart of the entire talk. The problem was never the missing of data, the retrieval. The problem is like the missing

拥有所有这些数据。只要你把它接上，它就有所有的看板、任务、邮件、Slack 消息，你碰过的一切它都有。但它依然回答不了这个问题。它有全部数据，却完全没有理解。这就是我们面对的真正挑战，也是整场演讲的核心。问题从来都不是数据缺失，也不是检索（retrieval）。问题是缺少

[2:13]

understanding. uh those are two totally different things and almost everyone mix between them. Understanding is the word that we're going to focus the entire the entire talk. Not context, not memory, not retrieval, understanding. So um quick about ourselves. My name is Tormer. My name is Omri. This is Tor. Um uh we both engineering manager at monday.com working on exactly uh the problem that we going to talk about a

理解。这是两件完全不同的事，而几乎所有人都把它们混为一谈。「理解」（understanding）就是我们整场演讲要聚焦的那个词。不是 context，不是 memory，不是 retrieval，而是 understanding。简单介绍一下我们自己。我叫 Omri，这位是 Tomer。我们俩都是 monday.com 的工程经理，做的正是我们今天要讲的这个问题。再稍微

[2:43]

little a little bit context about where we coming from. Uh monday.com is a global software company. Uh we build a work platform uh used by hundreds of thousands of teams. Um and the part that really mattered for the talk is that Monday is where the work um lives. uh we help companies to doing the work not just like saving records. Every project,

交代一点背景，让大家知道我们是从哪儿来的。monday.com 是一家全球性的软件公司，我们做的是一个工作平台，有几十万个团队在用。而对今天这个话题真正重要的一点是：monday.com 是工作真正发生的地方。我们帮助企业把活干完，而不只是保存记录。每个项目、

[3:05]

every task, every decision, every meeting, the notes, the action item, everything logs logged into the system. And our mission has always been to help the teams to achieve their business outcome. Whether you are a salesperson, so we want to help you like create an NSDR agent that call to your prospects and help you sell the product. And if you are like a finance team or marketing we want to help you uh with the research. So each one of the discipline

每个任务、每个决策、每场会议、会议纪要、待办事项，全都会被记录进系统。而我们的使命一直是帮助团队达成他们的业务成果。如果你是销售，我们想帮你打造一个 SDR agent，去给你的潜在客户打电话、帮你把产品卖出去。如果你是财务团队或者市场团队，我们想在调研上帮你。所以每一个职能，

[3:34]

we want you help you to doing the job and together we have like four big bats on the platform. We have like Monday sidekick we're going to talk about it mostly today. Uh Monday vibe if you want to build your own software. Monday agent if you want to create your own agent into the platform and like if you want some more deterministic flow we have like Monday workflows but today we're gonna focus on psychic um psychic is

我们都想帮你把工作做好。整体上，我们在平台上押了四个大注。一个是 monday Sidekick，这也是我们今天主要要聊的。还有 monday Vibe，如果你想自己搭软件。monday Agent，如果你想在平台上创建自己的 agent。如果你想要更确定性的流程，我们还有 monday Workflows。不过今天我们要聚焦在 Sidekick 上。Sidekick 是

[3:58]

your intelligence AI personal assistant that's understand your work think and execute with you like a bird on your shoulder uh he know you he know your business um it help you with your work on every aspect uh he work the way you do uh with your tones Um um it's keep you totally in control but like putting all these bullets and uh building all these promise is really hard. Um so let's talk about why it's hard and there

你的智能 AI 个人助理，它理解你的工作，和你一起思考、一起执行，就像停在你肩膀上的一只鸟。它了解你，也了解你的业务，在工作的每个层面帮助你。它按照你的方式做事，用你的语气，同时让你始终保有完全的控制权。但要把这些要点真的落地、要兑现这些承诺，是非常难的。所以我们来聊聊为什么难，

[4:28]

are three points for that. Um first like picture one assistant sitting on top of absolutely everything. You have your u your u slack messages. You have all the notes for your meetings. um absolutely everything and it's beautiful and overwhelming at the same time because as it stand like there's wall of records everywhere. Um so what why it's making so hard to connect between them. There are three reason for that. One is

这里有三点原因。首先，想象有这么一个助手，坐在所有东西之上。你有 Slack 消息，有所有会议的纪要，什么都有。这既美好又让人不知所措，因为就这么摆着，四处都是一堵一堵的记录墙。那为什么把它们连起来这么难？有三个原因。第一个是

[5:02]

something that we call the agent gap. Your agent every agent is really sharp and doing task um when you when he know what to do. Um but he sometimes lost find them find the problem. Um if I ask my agent like please help me draft and and and reply to like the escalation from a customer he nailed it. He know how to do it. He take the context. I like doing the work but asking him what should I focus on first he guess because he doesn't understand

我们称之为「agent gap」的东西。你的 agent，任何 agent，在执行任务时都非常敏锐——前提是它知道该干什么。但它有时候找不到问题本身在哪儿。如果我让我的 agent 帮我起草一封回复，回应某个客户的升级投诉，它做得非常漂亮。它知道怎么做，会去抓上下文，把活干好。但如果你问它「我该先专注做什么」，它就只能靠猜，因为它不理解

[5:33]

what is my priority who am I it doesn't matter if you have a memory or something he still don't know what is the problem the second problem is like um we have the records but we don't have the meaning um a log never said what it mean let's take an analogy from our Let's say that you have one line of code in your GitHub. You look at a code,

我的优先级是什么、我是谁。就算你有 memory 之类的东西也没用，它依然不知道问题是什么。第二个问题是，我们有记录，但没有含义。一条日志永远不会告诉你它意味着什么。打个我们自己领域的比方：假设你的 GitHub 里有一行代码，你看着这行代码，

[5:59]

maybe there is a comment on top of it. Nice. But you don't really understand why someone wrote this line of code. Today, no none of us writing code. But somewhere in the past, someone wrote this code line of code. And if you really want to know, you can go to get blame and understand uh from the commit log what why why someone do it. But if you want to go farther, you can go to the PR and maybe read the description.

上面可能有一条注释，不错。但你并不真的明白当初为什么有人写下这行代码。今天我们都不怎么自己写代码了，但在过去的某个时刻，是有人写下了这行代码。如果你真想知道原因，你可以去 git blame，从 commit log 里看看为什么有人这么做。如果你想再往深了挖，你可以去看那个 PR，读一读描述。

[6:26]

And if you really want to go hard, you can go to your Monday board and see which PR connected to this item and understand that this line of code came because some customer complain about something. So this is what we're trying to build. And the third point that is a bit challenging as well is like um it's really hard to build ahead of time um at runtime um the meaning of things. Um it's really too late. Um you simply

如果你真的想挖到底，你可以去你的 monday.com 看板，看看这个 PR 关联到哪个条目，然后你就会明白，这行代码之所以存在，是因为某个客户投诉了某件事。这就是我们想要构建的东西。第三点同样有挑战性：要在事情发生的当下、在运行时（runtime）去构建含义，实在太难了，那时候已经太晚了。你根本

[6:55]

can't do something like that the moment someone asked the question. So understanding understanding of the context has to be ahead of time. Um you need to build it much before someone asks the question. Um and this is why we have built what we are building. We are building uh the Monday world model. This is what we called the Monday world model. Help you um understand why this matter um how to help you, who you are, when and and

没办法在有人提问的那一刻才去做这件事。所以对 context 的理解必须是提前完成的（ahead of time），你得在有人提问之前很久就把它构建好。这也正是我们在做这个东西的原因。我们在构建的是 monday world model，我们就这么叫它。它帮你理解为什么这件事重要、怎么帮你、你是谁、什么时候做、

[7:26]

what's not to do. Um it's the context that follow your work. Um he understand who you are and it's simply not a bigger prompt. uh it's not a longer uh context window. It's a totally different um from from what we know until now. First before like Tor going to talk about how we build it. What it's not it's not a retrieval problem. The problem never been getting the data we don't we have all the data we have all the connection

以及什么不该做。它是跟着你的工作一起走的 context。它理解你是谁，它绝不只是一个更大的 prompt，也不是一个更长的 context window。它跟我们此前熟悉的东西完全不一样。在 Tomer 讲我们怎么构建它之前，先说说它不是什么：它不是一个 retrieval 问题。问题从来不是拿不到数据——我们有全部数据，我们有到

[7:57]

to what all the provider that we want to get all the MCPS. The problem is really to understand how it works. Understand how each one of these entity connected to each other. So go ahead. Thank you. Hi everybody. Um so what's the data model? We collect thousands of data points on the user every item status change their activity log messages and meetings and construct three things the agent can resone over.

各个数据源的所有连接，所有的 MCP 都有。真正的问题是理解它是怎么运转的，理解这些 entity 彼此之间是如何连接的。你来吧。谢谢。大家好。那么这个 data model 到底是什么？我们会围绕用户收集数千个数据点——每个条目的状态变更、他们的活动日志、消息、会议——并据此构建出三样 agent 可以推理（reason）的东西。

[8:28]

The first is how the user's work is structured, their key entities and relationships and connections between them. What depends on what, how a message in Slack connects to a task, who's blocking whom. The second is a current snapshot, live signals over those entities, what's overdue, what's critically urgent right now, which co-workers you've been actively working with, and why. The third is what we can learn about the user over time. There

第一是用户的工作是如何组织的：他们的关键 entity、关系，以及彼此之间的连接。什么依赖什么，Slack 里的一条消息如何关联到某个任务，谁在挡着谁。第二是当前的快照：这些 entity 之上的实时信号——什么逾期了、此刻什么最紧急、你最近在和哪些同事密切协作、以及为什么。第三是我们随时间能学到的关于这个用户的东西——

[8:57]

are decisions and outcomes, work patterns and cadences distilled into a durable profile. So how do we build that data model? Um we use two engines running on different time windows and schedules. A slow engine that runs on a long time window and learns the user and their work and a fast engine that reads what's happening right now and how it affects the user's work. One knows you and the other one knows your day.

他们的决策与结果、工作模式和节奏，提炼成一份持久的 profile。那我们怎么构建这个 data model 呢？我们用了两套引擎，跑在不同的时间窗口和调度上。一个慢引擎，跑在很长的时间窗口上，用来学习这个用户和他的工作；还有一个快引擎，读取此刻正在发生的事，以及它如何影响用户的工作。一个了解你这个人，另一个了解你的这一天。

[9:24]

First the slow engine. It takes as context users activity over weeks and minds it for patterns and the type of persona the user is their routines their work rhythm who they collaborate with their main goals and current projects. Those patterns get distilled into a durable profile and every time a profile holds it's reinforced. This engine tries over time to learn who exactly the user is and how they work.

先说慢引擎。它把用户数周以来的活动作为 context，从中挖掘模式：用户属于哪种 persona、他们的例行事务、工作节奏、和谁协作、主要目标和当前的项目。这些模式会被提炼成一份持久的 profile，每当某个 profile 被再次验证成立，它就会被强化。这套引擎试图随着时间推移，弄清楚用户究竟是谁、他们是怎么工作的。

[9:56]

The fast engine is the opposite. It takes as context a short recent window and recomputes a set of live signals over the user's current state. What server do, what's suddenly urgent, which co-workers you've been pulled in with. This engine tries to understand your day and updates frequently.

快引擎则正好相反。它把最近一小段时间窗口作为 context，重新计算一组关于用户当前状态的实时信号：什么逾期了、什么突然变得紧急、你最近被拉进来和哪些同事一起干活。这套引擎试图理解你的这一天，并且更新得非常频繁。

[10:16]

Uh this split isn't something we invented. It's present in two totally different fields. In neuroscience, this split is referred to as complimentary learning systems. And in data processing architecture, it's referred to as a lambda architecture. We apply the same concepts to how we construct the agent's data model.

这种拆分并不是我们发明的。它在两个完全不同的领域里都出现过。在神经科学里，这种拆分叫做互补学习系统（complementary learning systems）；在数据处理架构里，它叫做 lambda architecture。我们把同样的理念套用到了 agent 的 data model 构建方式上。

[10:37]

Our brain uses the same split. Uh every important experience gets captured instantly by the hippocmpus and over time the new cortex distills those into durable lessons. In data infrastructure, the same there's the same split. A fast speed layer over a recent real-time window and a slow batch layer over the full history that gets recomputed and the two are merged into a single served view. Two different fields landed on the same idea and

我们的大脑就用了同样的拆分：每一段重要的经历会被海马体（hippocampus）瞬时捕捉下来，而新皮层（neocortex）会随着时间把它们提炼成持久的经验。在数据基础设施里也是同样的拆分：一个跑在近期实时窗口上的快速 speed layer，和一个跑在完整历史上、会被重新计算的慢速 batch layer，两者再合并成一个统一的服务视图。两个不同的领域殊途同归地得出了同一个想法，

[11:09]

that's what we're trying to apply to our data model. So how does it all come together? We collect data from everywhere our users work. uh Monday, Slack, emails, calendar, and we turn it into data structures, signals, and patterns. Both engines premputee on top of that offline and ahead of time. And when a user engages with Sidekick, a thin slice of logic is recomputed for recent activity,

而这正是我们试图应用到自己 data model 上的东西。那这一切是怎么组合到一起的呢？我们从用户工作的所有地方收集数据——monday.com、Slack、邮件、日历——然后把它们转化成数据结构、信号和模式。两套引擎都在这之上离线地、提前地预计算。当用户开始使用 Sidekick 时，会有很薄的一层逻辑针对最近的活动做重算，

[11:39]

and the entire context is served to the agent. Sidekick can then decide when and how to traverse and retrieve context from the data model itself. and it's primed to reason on top of it. And building it this way gives us two behaviors out of the box. The data model, it's resilient. Sources are isolated so a bad feed can't break the rest. And the thin layer of logic that runs at serve time verifies part of the context against live data while the rest

然后把整个 context 交付给 agent。Sidekick 接着就可以自己决定，什么时候、以什么方式去遍历 data model 并从中检索 context，而且它已经被预热到可以在此之上做推理了。这样构建还带来了两个开箱即用的特性。第一，这个 data model 很有韧性。各个数据源是相互隔离的，所以某一路数据出问题不会拖垮其他部分。而那层在服务时运行的薄逻辑，会拿实时数据去校验一部分 context，其余部分则

[12:11]

falls back to the last verified context. So it degrades gracefully, but it doesn't fail. Second, it actually understands the urgency of facts. It's able to understand when and how it should be proactive and notify the user and when it should stay silent.

回退到最后一次验证过的 context。所以它是优雅降级，而不是直接失败。第二，它是真的理解事实的紧急程度。它能判断什么时候、以什么方式应该主动出击去提醒用户，以及什么时候应该保持沉默。

[12:30]

And the crucial part is that it compounds. Every day the data is captured, the layers fill in and the profile sharpens. And adding a new data source is deliberately cheap and only contributes. So the surface only grows. The more it sees, the more it understands. And the more it understands, the more you can lean on it. And the data model is unique. It's unique to how you work.

而最关键的一点是，它会不断累积复利。每一天数据被捕捉进来，各个层次被逐渐填满，profile 变得越来越精准。而且新增一个数据源被我们刻意设计得非常廉价，并且只会做加法，所以覆盖面只会不断扩大。它看到的越多，理解得就越多；理解得越多，你就越能依赖它。而这个 data model 是独一无二的，它是为你的工作方式量身定制的。

[12:56]

We're not pretending this solves everything. Uh the model itself is always trailing the actual live world. New users have no reliable data to reason from yet, and signals have our own biases built in. So the hardest part is actually telling the important parts from the noise. But this is an architectural design that we can enrich and test and improve over time. Thank you.

我们并不是在说这解决了所有问题。这个模型本身总是滞后于真实世界的实时状态。新用户还没有足够可靠的数据可以拿来推理，而信号里也内建了我们自己的偏见。所以最难的部分其实是把重要的信息从噪音里挑出来。但这是一套我们可以随时间不断丰富、测试和改进的架构设计。谢谢大家。

[13:25]

[applause] So back uh back back to the original uh question that we had. What should I focus on right now? Remember the the question. So now Psyche can really answer that. Uh let's see how. So first like Thomas said we collect all the data point we called it breadcrumbs for example the board that I working on right now um all the emails from the past uh months um the transcript for the meeting all the action item that I got

[掌声] 那我们回到最开始那个问题：我现在该专注做什么？还记得这个问题吧。现在 Sidekick 是真的能回答它了。我们来看看是怎么做到的。首先，就像 Tomer 说的，我们收集所有数据点，我们把它们叫做「面包屑」（breadcrumbs）。比如我现在正在用的看板，过去几个月的所有邮件，会议的转录文本，我从那些会议里领到的所有待办事项，

[13:53]

from those meeting and we processing them offline um we see a full picture of your calendar this is what help us to understand the pattern so if you have like bi-weekly with your VPs I understand that tomorrow you have another meetings and even slack messages from the last few days to understand if you say something in the Slack that's relevant that you didn't say in your daily uh standup with the team. We took all this data and we uh process them

然后我们离线处理这些数据。我们能看到你日历的完整图景，这帮助我们理解其中的模式。比如你和 VP 有个双周会，那我就知道你明天还有别的会。甚至还包括最近几天的 Slack 消息，用来判断你是不是在 Slack 里说了什么相关的事，而那件事你在团队的每日站会上并没有提。我们把所有这些数据拿过来做处理，

[14:21]

either fast or slow. Um the slow build some kind of like profile on you. So you can see it on the top like combin is an engineering manager. We I'm working on two projects psychic and notaker. I'm from Tel Aviv. And this is how many hours I have per day and like how I work every day. And the fast is something that's helped me to understand what is the action item. I have like three commitment that I promised to other

要么快、要么慢。慢的那条会给你建立起某种「档案」。你可以在顶部看到，比如：Omri 是一名工程经理，我在做两个项目——Psychic 和 Notetaker，我来自特拉维夫，这是我每天有多少小时、以及我每天是怎么工作的。而快的那条则帮我搞清楚待办事项是什么。比如我有三项承诺，是我答应别人要做的

[14:45]

people. I have to to reply to a VP that sending email and I didn't. Um so this is only from the past uh uh day or few days window. This is something that's relevant more for today and this is how psychic can answer those questions. So to summarize it, the bottleneck was never uh the capability where you can where you take all the data from. It was the understanding how you connect each one of the dots to each other. Um the

事。我得回复一位副总裁发来的邮件，但我还没回。这些只来自过去一天或几天的时间窗口，是跟今天更相关的东西，而这就是 Psychic 能回答这些问题的方式。所以总结一下：瓶颈从来都不是能力本身，不是你能从哪里拿到所有数据，而是理解如何把每一个点跟其他的点连接起来。

[15:13]

most capable agent in the world whether it's like a cloud and Gemini, it doesn't understand you. He need to process it beforehand. This is what we're trying to solve at Psychic. Um this is the Monday word model. Uh this is what we're building. Um if you have any question,

世界上最强的 agent，不管是 Claude 还是 Gemini，它并不了解你。你需要事先把这些处理好。这就是我们在 Psychic 想解决的问题。这就是 monday.com 的世界模型，这就是我们正在构建的东西。如果你有任何问题，

[15:28]

uh we out of time. We are here uh uh near the stage. Thank you very much for the time. Uh you can follow us on LinkedIn. We post things on on the area. Thank you very much. [applause]

我们时间不够了，我们就在舞台旁边。非常感谢大家的时间。你们可以在 LinkedIn 上关注我们，我们会在那个领域发布一些内容。非常感谢。[掌声]