Makers of Hindsight. Agent memory that lets your agents learn over time. https://xitter.cf/t.co/JkzXgOQg74
Boulder, CO
Joined December 2023
- Tweets1.4K
- Following331
- Followers1.5K
- Likes978
"The software agent, however, was an amnesiac."
Aditya Rana on solving their agent memory issue for good: medium.com/@adityarana9268/h…
Your Agent Can Now Manage Its Own Knowledge Base Over MCP.
A Hindsight knowledge base is the readable, self-healing wiki sitting on top of your agent's memory: pages that distill what the agent knows into something a human can skim, and that rebuild themselves as the underlying memory changes.
Until now there was an odd asymmetry. An agent could retain and recall over MCP all day, but to browse or maintain that wiki it had to drop down to a separate HTTP path. The memory was native; the knowledge base was a second integration.
Hindsight 0.9.2 closes that gap. The knowledge base is now a native MCP surface: seven agent-facing operations registered as tools, on the same connection the agent already uses to remember.
Know more: lnkd.in/gBgCwSTg
Vectorize retweeted
Topic: How many agents actually agreed?
hindsight.vectorize.io/blog/…
1/ When a swarm of agents agrees on something, my first question is how many independent looks actually happened. A post on Vectorize's Hindsight blog, "Thirty Agents, One Memory," brought that back to mind.
2/ Hindsight's proof_count counts the distinct live memory IDs behind an observation, so twenty workers reaching the same conclusion can fold into one observation citing twenty facts. I'd still want to know whether those twenty claims trace back to independent evidence. Workers that never saw each other's context can all read the same bad source.
3/ The article sketches the failure mode itself. One worker reads a hostile file and writes a poisoned conclusion, later workers recall it as a colleague's finding, and proof_count can end up reporting that nineteen sources agree. Retries add to it. A worker that restarts without a stable document ID writes its memories again, and the count climbs without anything new being checked.
4/ Before trusting a deployment like this I'd confirm that injection quarantine is actually enforced in the edition you're running. Memory Defense is disabled by default, and the open source guide only documents redacting and blocking sensitive data. I'd also make sure the document ID stays tied to the logical task when a worker retries.
5/ Then I'd test it. Plant one false fact through one worker, keep the record of how it gets recalled somewhere the workers can't write to, and measure which downstream decisions change because of it.
Give Every Agent You Run in Omnigent a Persistent Memory.
Most agent memory integrations solve one problem for one tool. Add Hindsight to Cursor, and Cursor remembers. Add it to Aider, and Aider remembers. Each is its own setup, its own configuration, its own ceiling.
Omnigent breaks that pattern. It is a meta-harness: a single orchestration layer that wraps and coordinates multiple AI agents at once, including Claude Code, Codex, Cursor, OpenCode, Hermes, and Pi. When you add Hindsight to Omnigent, you add persistent memory to all of them through one place, with a single setup instead of wiring up each tool's integration separately. And for any harness that has no native Hindsight integration, including custom ones, Omnigent is the memory layer. The bridge lives in Omnigent, so the agents inside it do not have to carry it themselves.
Know more: lnkd.in/gJcftPPF
Vectorize retweeted
The #1 long-term memory for Hermes? A folder of notes.
138 Hermes users answered:
▸ Obsidian / Markdown vault: 27
▸ Hindsight: 25
▸ Honcho: 22
▸ Built their own: 18
25+ tools named. Local beat cloud 28 to 2.
Memory is still anyone's game. Every system ranked below.
Vectorize retweeted
I tried them all. Simplicity won for me:
1. @Linear = source of truth/context layer/tasks/project management
2. @Vectorizeio Hindsight = Agent memory/journals
Any question?
Vectorize retweeted
Yep. You don’t need so many apps. I hate this. I use linear als source of truth and project management tool. And hindsight as agent memory for all agents. That’s it. Don’t over complicate
Vectorize retweeted
TITLE
Your Agent Forgets You Tonight
DESCRIPTION
Tomorrow morning the agent does not know who Alice is. You told it yesterday. Hindsight, from Vectorize, is an open memory layer that says it is built to learn, not just replay the chat.
This film uses their README and a GitHub measurement taken October 24, 2026. I did not run Hindsight, and I did not rerun their benchmark.
What their README says
- Most agent memory recalls the conversation. Hindsight says it is focused on agents that learn, not just remember.
- Three operations: retain writes a fact into a bank, recall searches that bank, and reflect answers from the whole pile.
- Their own example: Alice works at Google as a software engineer.
- Four kinds of memory: a world fact, an experience, an observation that keeps its quotes, and a mental model that is rewritten as the bank learns.
- A bank is one brain for one user, one agent, or one project. They say isolation is strict. Nothing leaks across banks.
- Memory Defense is off until you turn it on. Then every retain is scanned against 45 patterns for secrets and personal data. A match is redacted, or blocked, before it is stored.
- They say it leads the LongMemEval benchmark, and that Virginia Tech and The Washington Post reproduced that result. Other scores, they say, are self-reported. I did not rerun it.
- They say Fortune 500 companies use it. That sentence is theirs.
- Their support table lists Windows, with Docker or pip.
Measured on GitHub, October 24, 2026
- Language: Python. License: MIT.
- Created October 30, 2025. Last push 2:44 p.m. Pacific, the same day.
- 53,425 stars. 142 open issues. 6,237 forks.
- Stars are attention, not a test.
Comment the one thing you would never retain: a password, a client name, or yesterday's work.
Sources
github.com/vectorize-io/hind…
api.github.com/repos/vectori…
hindsight.vectorize.io/
Chapters
0:00 Forgets tonight
0:09 Built to learn
0:22 Write it down
0:37 Four kinds
0:56 One brain
1:17 Not a test
2:01 You keep
HASHTAGS
#Hindsight #AIAgents #AgentMemory #AINews #Vectorize #AITools #Python #OpenSource
TAGS
Hindsight, AI agents, agent memory, Vectorize, retain, recall, reflect, AI news, AI tools, Python, MIT, LongMemEval
See the Exact Prompt Before You Spend a Token.
You write a mission for your memory bank. Something like "focus on billing decisions and who approved them." You save it, and the extraction gets better, or it doesn't.
Now: where did that sentence go?
If you guessed the system prompt, you're wrong for two of the three operations that use it.
TL;DR
- A retain or consolidation mission lands in the user message, not the system prompt. Reflect's lands in the system prompt, where it replaces the agent's role line rather than adding to it.
- POST /banks/{id}/prompts/preview returns the exact messages retain, consolidation or reflect would send. No LLM call, no writes.
- Active blocks concatenate byte-for-byte into the real prompt, and there's a test pinning that against the actual builder.
- A setting you switched off comes back as an inactive block, positioned where it would have landed.
- The endpoint refuses to accept your sample text or a candidate mission, on purpose.
- Shipped in v0.10.0 for self-hosted, and coming soon to Hindsight Cloud.
Know more: linkedin.com/pulse/see-exact…
Vectorize retweeted
me on the phone looking at shared memories from claude and chatgpt via @Vectorizeio ; spine router is my primary ai skill installed, medical analyzer is an extremely interesting way my chats analyze stuff
Vectorize retweeted
DOT agent just orchestrated a custom plugin to give my phone chatgpt shared memory with @Vectorizeio !!!
xitter.cf/stevenmathern/status/2…
chatGPT work was able to create a custom plug-in for ChatGPT mobile to connect to the hindsight memory system @Vectorizeio as you can see in the screenshots and we have confirmed that it is functional. This was orchestrated by the dot agent running two different chats said it's sol high. It is now assisting Claude in connecting to the shared memory bank while I just talk on my phone, only occasionally clicking a confirmation on the Mac.
Vectorize retweeted
Part 69 in amazing repos
Agent memory that learns
The diagram shows the retain, recall and reflect loop
Repo: github.com/vectorize-io/hind…
Vectorize retweeted
Asked Hermes to consult my Hindsight memory first. Its verdict: “You have a technician’s nose for bullshit, an architect’s appetite for interconnected systems, and the browsing restraint of a raccoon with an all-access badge.” Lmao
Vectorize retweeted
chatGPT work was able to create a custom plug-in for ChatGPT mobile to connect to the hindsight memory system @Vectorizeio as you can see in the screenshots and we have confirmed that it is functional. This was orchestrated by the dot agent running two different chats said it's sol high. It is now assisting Claude in connecting to the shared memory bank while I just talk on my phone, only occasionally clicking a confirmation on the Mac.
8 Agent Memory Use Cases (With Real Examples).
"Agent memory" sounds abstract until you notice it is the same fix applied to a dozen different problems. Any time an agent should remember something past the current session, and today starts from a blank slate, that is a memory-shaped hole. Hindsight fills it with the same three operations everywhere: retain to write, recall to read, reflect to reason. What changes from one use case to the next is not the primitive, it is who the memory belongs to and where it surfaces.
Here are the agent memory use cases people actually build, each with a concrete example: lnkd.in/gqe9TWR2
Vectorize retweeted
Memory should reduce repetition, not replace verification. We pair Hermes + Hindsight with live-source checks. How it works: tevpro.com/blog/how-we-use-h…
