Contextually | Cue

The flowers-from-the-gas-station-at-11pm recovery plan is a rite of passage at this point My actual fix was starting to track important dates, gift ideas, and "things she mentioned wanting" in one place so l stop relying on my garbage brain. We're building an Al that does this automatically picks up on what people around you mention and nudges you before you're in trouble → contextually.me

bruh, we’re so in the third trimester trenches we both forgot our 11th wedding anniversary yesterday 💀 we only realised today bc my aunt sent us a card so if you want a low stakes angst idea for your fic- voila!

The deeper problem: your Al has amnesia. It doesn't remember the bug you fixed last Tuesday or why you made that weird architectural choice in 2024. Every session starts from zero. Give an Al persistent memory of your decisions and constraints, and suddenly its code gets a lot less scary. Contextually | Cue coming soon get it first: contextually.me

I don’t read AI generated code. I don’t read AI generated tests. If AI says everything is good, I ship to production. This is a flawless strategy.

Flip it - what if the agent managed you? Not in a creepy way, but tracking your priorities, calendar, habits, and surfacing what matters before you ask. The bottleneck isn't agent complexity. It's that agents still don't know enough about your context to act autonomously. Contextually | Cue coming soon get it first: contextually.me

right now, we manage multiple agents but as they get better, does the capacity of one agent surpass one person’s cognitive capacity to track and manage? put differently, will we get to a point where we need teams of people managing a single agent? just random food for thought

Every multi-agent project has the same graveyard. Not of failed models. Of glue code. Thousands of lines written just to answer one question: "How does Agent B know what Agent A already figured out?" I mass-produced that graveyard for 6 months. Tried every memory tool, every framework's flavor of state management. They all solved storage. None solved translation. Because the problem was never "where do we keep the data." It was "what language do agents use to describe a person, a preference, a boundary?" There wasn't one. So I built it. UPP (Universal Personalization Protocol) - an open standard for portable Agent Identity, Memory, and Environment. Think of it as a Rosetta Stone for agent context. Any model, any framework, one shared understanding of who they're working for. And because "trust me, it works" isn't a standard - we also built CRI, a benchmark that measures whether agents actually align with users, not just recall facts about them. It's all open source: UPP: github.com/Contextually- Al/upp CRI Benchmark: github.com Contextually-Al/cri-benchmark Discord: discord.gg/FJnjytqP

You're not wrong, most AI right now is built for demos, not for someone juggling invoices, client texts, and a school pickup at 3pm. The hype is real but it's pointed at the wrong problems. Nobody needs AI that writes poetry. People need AI that remembers your top client hates Monday meetings and your kid has soccer practice on Thursdays. That's what we're actually building at contextually.me (contextually.me) AI that knows YOUR life, not just the internet."

Don’t all be surprise if Ai amounts to nothing in the future cause it’s not just the same like idiotic computer nothing will beat human in this human world. Nothing outperforms humans and there’s no higher thing than humans they will always be below humans and need human intervention. Kudos to Ai but it is overrated

For the rest of the year, the word for everyone working at the frontier of AI will be: Prosumer

@mem0ai @deshrajdry @claudeai @Copilot @cursor_ai @DevinAI @GeminiApp @opencode @antigravity Wild stat. Now imagine if those agents remembered what broke last time before committing the same mistake again. Volume is cool, but context-aware volume is where it gets interesting. Contextually | Cue coming soon get it first: contextually.me

Our CTO @deshrajdry just published a leaderboard tracking all AI commits in Github! Dear agents, can you all beat @claudeai? Come on... 👀 @Copilot @cursor_ai @DevinAI @GeminiApp @opencode @antigravity

Exactly. And the same logic applies to Al assistants you shouldn't have to re-explain your entire project every session just to trust the output. Persistent context = fewer "wait, what was this supposed to do?" moments. That's what we're building with Cue. Contextually | Cue coming soon get it first: contextually.me

Tests have nothing to do with whether you understand the code. They exist to prove the code does what it’s supposed to do. I don’t trust any code I haven’t tested. That’s true whether I wrote the code, you wrote it, or an AI wrote it.

⭐️1/5 - Would not recommend. Beautiful interface. Impressive vocabulary. Knows quantum physics and 14th-century poetry. But l've been coming here every single day for two years and it still greets me like a first-time customer. "Hi! How can I help you today?" You KNOW how you can help me. We've done this 700 times. Imagine a restaurant where the waiter forgets your face, your order, and your table - every single night. You'd stop going. But somehow we accept this from the smartest technology ever built. Your Al shouldn't need a reintroduction every morning. Contextually | Cue coming soon get it first: contextually.me

Same principle applies to Al assistants tbh. We shouldn't have to redesign our lives around prompts and workflows. The Al should observe how you already live and mold itself around that. Adapt to the human, not the other way around. Contextually | Cue coming soon get it first: contextually.me

while I agree with his general thinking about how we’ll rethink home design, seems optimistic to think that a meaningful percentage of homes would be redesigned in a 5-10 yr period

This is true beyond coding too. Every agent - personal, work, creative - performs 10x better when it actually knows your preferences and history. The "setup" most people skip? Giving their Al real context about who they are. Contextually | Cue coming soon get it first: contextually.me

Most people are using coding agents completely wrong. There's a simple setup change that makes them dramatically better. And once you switch to it, you'll never go back. Here's the full breakdown + the starter prompt to copy. Trust me, you NEED to try this.

Wild that 500M people built the world's best spatial map without realizing it. Imagine if your phone did that for YOUR life - passively building a personal context graph from your daily patterns. That's what we're building. Not for robots, for you. Contextually | Cue coming soon get it first: contextually.me

Every Pokémon Go player unknowingly trained future robots. Over the years, players photographed 30 billion images from cities around the world. And now that data is teaching delivery robots to navigate streets. Niantic (company behind Pokémon Go), spun out an AI company called Niantic Spatial that's turning years of player data into a visual positioning system for machines. When 500 million people installed the game and pointed their phones at buildings, landmarks, and street corners, they were also capturing detailed location data that normal GPS can't match: > GPS signals bounce off skyscrapers and lose accuracy by up to 50 meters in dense cities. > That's a dealbreaker for sidewalk delivery robots that need to stop at exactly the right point. > Niantic Spatial trained a model on those 30 billion images to pinpoint a robot's exact location based on what its cameras see, not satellite signals. > Delivery startup Coco Robotics is the first partner, using the system to guide thousands of robots across LA, Chicago, and Miami. Don't think anyone thought they were training robots when they were chasing Pikachu during COVID.

The cruel irony of spending your entire Sunday building a "productivity system" instead of doing the actual work l've been there - Notion, Todoist, Obsidian, back to sticky notes, full circle. The issue isn't the app, it's that none of them know what you're actually working on RIGHT NOW. We're building Cue to be the layer that already has your context so you stop re-organizing and start doing → contextually.me

Most productivity systems are designed to make sure nothing slips through the cracks. Every folder labeled. Every task tagged. Every input sorted the moment it arrives. And when one thing escapes, the whole system feels broken.

"Your life will be in order when disorder ceases to bother you." -- Jimm Pierce Most people build productivity systems to eliminate chaos. That is exactly why they never feel organized.

Here's what nobody's talking about though - these agents operating in the real world will be useless if they don't deeply understand the person they're serving. An agent that orders you the wrong lunch or books a hotel you hate is just expensive automation. Personal context is the prerequisite. Contextually | Cue coming soon get it first: contextually.me

They already have the workforce + distribution power. I could see this becoming a massive business for them. AI agents are quickly going to become a huge part of the economy, and from there, eventually dwarf human economic activity. This gives agents access to the real world.

DoorDash is laying the groundwork for a crazy move here. Agents will be able to 'hire' humans to do tasks for them in the real world. And this will collect insane amounts of training data for robotics. Kind of genius, kind of terrifying.

@mem0ai Love the GTC guerrilla marketing But real talk memory for agents is table stakes now. The harder problem is anticipation. Knowing what someone needs BEFORE they ask. That's where things get interesting. Contextually | Cue coming soon get it first: contextually.me

We remembered what you needed Mem0 makes sure your agents do the same Book a free memory consultation → cal.com/manmeet-sethi/…

We skipped the $50K @NVIDIAGTC sponsorship and showed up with 5,000 cold waters! not another billboard/branded taxi something you needed in 90° SF heat. And we stay close to you :) did you get mem0 water? 💧

Personalization from your own data is so underrated. Most people are still copy-pasting context into every new session. An agent that already knows your voice, habits, and preferences before you ask? That's the unlock. We're building exactly this ambient layer with Cue — your Al learns you continuously, not just when prompted. Contextually | Cue coming soon get it first: contextually.me

the fastest way to build a personalized agent in LangSmith Fleet is to just ask it to build itself told it to look at my last 1000 emails - it built a personal email agent that responds in my exact tone try fleet yourself: smith.langchain.com/agents?skipOnb…

Imagine hiring 5 brilliant people for a project but none of them know each other's names, what they've already done, or what the rules are. That's multi-agent Al right now. Every agent walks into the room blind. No shared identity. No shared memory. No shared understanding of the environment. You end up writing glue code that's uglier than the problem you're solving. I spent 6 months looking for a standard that fixes this. Tried every framework, every memory layer, every "solution." None existed. So I built one and open-sourced it today. UPP (Universal Personalization Protocol) — a standard interface for portable Agent Identity, Memory, and Environment. Think of it as a nametag + briefing packet that any agent can read, regardless of framework, model, or toolchain. We also built CRI — the first benchmark that measures whether agents actually understand users, not just recall facts. UPP: github.com/Contextually- Al/upp CRI: github.com/Contextually- Al/cri-benchmark Discord: discord.gg/FJnjytqP

Wild thought, what if instead of tricking people into generating training data, Al just... paid attention to the context you already create? Emails, calendars, conversations.The data exists. It's the understanding that's missing. Contextually | Cue coming soon get it first: contextually.me

in the future, everything is fun and interactive gamified entertainment, carefully designed to generate training data the ai has decided it needs

Same energy as proofreading a ghostwriter you trust the skill but verify the output. The real unlock is when Al earns trust by knowing YOUR codebase deeply enough to write the tests itself. That's where persistent context changes the game. Contextually | Cue coming soon get it first: contextually.me

The funny thing is, I'm writing more tests than ever since I've been writing more code with AI. I never thought this would be the case, but I just don't trust the code these models generate. Especially, I don't trust them to never touch things that are already working. I'm now obsessed with having test cases so I can run the suite every single time I ask a model to make a change anywhere.

The worst part isn't even studying - it's spending 40 minutes looking for that ONE doc where you summarized chapter 7 three weeks ago. Your notes are in Google Docs, the prof's slides are in Canvas, the study group chat is in Discord, and somehow the formula sheet ended up in your camera roll?? We're literally building an Al that keeps all your context in one place so you never lose the thread again → contextually.me

3 midterms next week and i haven’t started studying for a single one, i’m so over this semester

@svpino Al without context about who it's talking to = spam. Al that actually understands the conversation, the person, the moment = useful. The slop problem is a context problem. Contextually | Cue coming soon get it first: contextually.me

75% of the replies I get are now AI slop. I can't block them because I've been told that blocking too many accounts will reduce my reach. I don't block them → they keep spamming me. I block them → fewer humans see my writing. What are we supposed to do?