I made a CLI to sync your Codex prompts + agent reasoning + file diffs to a shared memory called Codaph (https://codaph.com). The goal was to sync agent activity across my team for a much richer understanding of the codebase.
Under the hood, it uses Mubit (https://mubit.ai) - memory engine we built based on associative retrieval. Mubit is built on the concepts of hypervectors and clustering (with time based decay for now)
Currently, Codaph works with Codex with plans to add other agentic tools soon.
Codaph is open source (vibe coded lol).
Would love to hear some feedback and if you want to try it Mubit is free (api key is available on https://console.mubit.ai)
I'm building Rice (docs.tryrice.com). Think of Rice as a managed state machine for AI agents with long term memory.
Rice is a platform that unifies long term memory and short term state management for AI agents. Effectively, Rice solves the context compounding issue in the immediate sense - by using Rice Slate (our state management service), the context consumption was down 60%. This makes the agents more efficient. The state management layer also allows agents to share context without the conventional "message passing" approach meaning you can run parallel AI agents.
The memory layer enables the agents to have a broader contextual understand of the data and relationships - personalisation and automation at scale for agents.
How we're different (https://docs.tryrice.com/rice-vs) and working on some cool aspects.
The core value prop -
1. Auditable Agentic executions out of the box 2. Shared state for AI agents (not using message passing approach) for efficient executions 3. Persistent memory for historical data and more.
Currently in beta phase, so looking for beta testers. Appreciate any thoughts and tests.
Please enter your email at tryrice.com if you'd like to get in the beta.
A £2 Billion problem?
Apparently, £2B+ every year is wasted by consumers in the UK due to fake reviews and on the surface this seems like a problem needing a fix. So, I decided to dig deep.
After analysing more than 1000 products on Amazon (UK) alone across tech, fashion, accessories, and home improvement categories, only 17% reviews came out to be real — despite various strategies it all boiled down to the sentiment of reviews. About 12% reviews were duplicated which skews the average rating (for the thought that only purchasers can add a review) and about 33% reviews were 1-star or 5-star (just an observation and not too much of analysis here).
But all this data didn’t seem great enough due to sheer lack of volume (and patience), so I took a nap. After the hard earned nap, it occurred to me that this law would make it difficult to leave negative reviews. Most online platforms nowadays have a review system for user reviews which does a bad a job at not approving useful negative reviews.
It hit me — its almost like how YouTube removed the dislike count to support content creators and some PR statements.
So, what could this mean for us consumers? Difficulty in giving negative feedback or getting free returns without giving a feedback (some ebay sellers add this note with your package) or just a pseudo level playing field for all businesses to optimise for quality than reviews?
I don’t know that yet, but enforcing something that is nearly impossible to enforce is surely a fun way to utilise resources — this doesn’t mean it shouldn’t be a law but there must be publicly available and understandable measures to keep checks.
Thought I'd share some points and hear some :))