On a mission to bring world’s most important data to @LighthouseWeb3 Install. Browse. Earn. Pay
- Tweets 155
- Following 20
- Followers 384
- Likes 531
We've been building this for a while.
Today it's yours.
The AI economy is built on your browsing data - your searches, your patterns, your purchases, without asking or paying you.
1MB changes that.
A free Chrome extension that finally makes your everyday browsing work FOR you.
→ Contribute your anonymized data to AI research on YOUR terms
→ Get paid when it's used
→ Earn automatic rewards at 60,000+ stores (Amazon, Flipkart, Walmart, Target & more)
Consent first. Always.
You decide what's shared.
You revoke anytime.
The deal was never fair.
Today we made it fair.
Install free → gettrack.link/r/zOEssjp9
13
5
41
2,122
1MB retweeted
Better models can search history more effectively. They still need access to it.
The finding worth sitting with here: a stronger model cannot recover details that an earlier memory system left out. A small experiment, but a useful reason to test what your agent actually needs.
Alongside recall quality, there’s ownership and portability. Can you carry your context into the next tool?
That’s where Lighthouse fits: user-owned, verifiable memory that lives beyond any single model or session.
// MemoryTax: The Bitter Lesson of Agent Memory //
Great insights for agent builders; bookmark it.
As models increase in capability, they also improve at things like memory generation.
Agent memory frameworks are everywhere right now. This study measured whether they still help once the model can search raw history itself.
The results suggest that a strong model needs less designed memory and
better tools for searching the raw record.
On one 419-turn LoCoMo conversation, Mem0 beats plain full context only with GPT-4o-mini. Building Mem0's memory costs more than an entire full-context rollout.
With Claude Opus 5.5, the Pi coding agent searching the raw history scores 99% for under half the cost of full context. Its one miss comes from a wrong answer key.
Memory also has a ceiling set by the model that wrote it. A stronger reader cannot recover what a weaker writer left out.
The author clearly defines the scope: one conversation and one run per cell. It is still a cheap experiment to repeat on your own agent before adding a memory layer.
arxiv.org/abs/2610.29823
Chat with Paper: academy.dair.ai/papers/memor…
5
21
1,130
1MB retweeted
A vault door on a straw hut is a pretty good security review.
That’s the image in @VitalikButerin’s Ethereum Community Conference slide: formally verified contracts behind a frontend that can be replaced if someone hacks the server.
His slide points to static HTML pages on @IPFS as part of the solution.
Then look at Polymarket’s 29 October incident. A third-party provider was compromised, and researchers found a monitoring script designed to access trading sessions and attempt to move assets.
The code meant to watch the page had become a threat to the people using it.
IPFS helps applications check content against an expected identifier. But those checks must cover the dependencies too. An unchanged page can still load a changed external script.
At Lighthouse, this is relevant to both storage and memory.
Lighthouse Storage helps builders store and retrieve content through IPFS .
Lighthouse Memory gives agents context they can carry between sessions and compatible tools.
When that context starts informing trades, payments or enterprise actions, its origin and permissions matter as much as its usefulness.
Your agent remembering an instruction is convenient. Knowing who put it there is essential.
2
2
22
1,356
🧠🧠
You remember making the call.
You just can’t remember which chat it’s buried in.
We put the Lighthouse Memory docs into a Brain and asked it how recall finds relevant memories.
It pulled the memories that answer it and lit them up in the graph.
Your project’s reasoning, open to explore and ask 🧵
44
1MB retweeted
Inference gets spent. Memory compounds.
We wrote down why AI memory will be bigger than inference, and why it should be yours.
15
9
52
3,142
1MB retweeted
Singapore, we're here.
Find us at @arbitrum Founder House in collaboration with @RobinhoodCrypto , Oct 23 to 25.
Bring your agents. We'll help them remember.
11
8
39
2,147
1MB retweeted
Lighthouse is closing in on 10 million files objects.
34.6K users and 50.42 TiB of storage consumed.
Built on @IPFS , @Filecoin and @WalrusProtocol , our storage infrastructure now powers persistent AI memory.
Explore our growth. Link in the comments.
7
11
52
4,634
1MB retweeted
The next leap in AI isn't just about smarter models, but what they can remember.
Imagine moving from ChatGPT to Claude without losing your context, preferences or decisions.
As Lighthouse expands into the @RobinhoodCrypto ecosystem through @virtuals_io , we're bringing private, persistent memory to a growing world of AI agents.
And this is only a glimpse of what's coming.
Here's how Lighthouse Memory works ↓
9
5
43
3,112
1MB retweeted
Tomorrow, Lighthouse gets a ticker.
For now, it’s saved in Lighthouse memory, encrypted and private until the reveal.
14.10.2026
28
8
55
8,712
1MB retweeted
Give your agents a persistent brain.
Lighthouse helps connected ChatGPT, Claude and other agents save useful context and recall relevant memories across sessions, even when you ask in different words.
9
20
40
6,127
1MB retweeted
The more your AI remembers, the more privacy matters.
An agent becomes useful when it understands the work behind your questions. The project you’re building. Why you changed direction. What your team decided yesterday and what you haven’t announced yet.
Over time, that memory becomes a detailed record of your work and how you think. Some of the context that makes your agent most useful is also the context you most need to protect.
A founder’s agent might remember an unreleased product strategy. A coding assistant might retain internal project decisions. A research agent might carry findings that a team is still developing.
These are everyday reasons to care about private AI memory.
Encrypted memory is already live on Lighthouse, giving users a way to preserve useful context in encrypted form as part of their agent’s persistent brain.
For us, building that brain means thinking about what it remembers, how that memory persists and how sensitive information is protected. Privacy belongs in that conversation from the beginning.
As more of our work happens alongside agents, the memories they accumulate will become increasingly valuable and increasingly personal.
Your agents should be able to build on what they know about you. That knowledge deserves protection.
3
6
19
3,412
The Stablecoin Shopping Loop - Day 5/5
Five days, one loop. Here's the recap, with a timely example.
World Cup season means Amazon carts full of jerseys, merch, and gear. Here's the new way to check out:
Old ecommerce: card fees, chargebacks, banks holding up settlement.
New ecommerce: stablecoins, instant settlement, no middlemen.
@1MB_io brings that loop into your browser:
→ Buying your FIFA jersey or World Cup merch on Amazon? Pay with USDC at checkout, live today.
→ Earn affiliate cashback across thousands of stores, automatically
→ Earn points for consent-verified browsing data
One extension. Three live features. Built on @solana .
This is what stablecoin ecommerce looks like when it's actually shipped, not promised.
Install it, shop like you always do, and watch the loop work.
1
5
178
The Stablecoin Shopping Loop - Day 4/5
@1MB_io 's loop has three pieces, and here's where each one stands today:
→ Stablecoin checkout - live on Amazon via USDC . Pay at checkout, settle on @solana , skip the card fees.
→ Affiliate cashback - live. Earn back on purchases across thousands of stores through one extension.
→ Consent-based data points - live. Browsing earns points, with cryptographic consent logged on-chain.
Most "earn while you shop" extensions stop at cashback. @1MB_io stacks stablecoin payments, affiliate rewards, and consent-first data into one browser extension.
Three live pieces. One loop. Built on @solana .
7
206