Context pollution and rot are probably more important than memory, because facts can usually be retrieved if the agent is good at following breadcrumbs.
What's also the biggest killer is code rot. Agents are particularly good at death by thousand cuts. They implement something poorly, or incorrectly, or introduce a bad pattern into the project. Then they continue to amplify that badness over time, as they continue to copy from it on subsequent work. It spreads like a virus.
Keeping these seeds out of the project is very difficult, and cleaning up the rot is very difficult. It also seems like a hard problem to solve because following the existing codebase is something that is good when the code is good, but bad when it is bad. So, seemingly, the solution means more thinking and evaluation for every change that is being made.
> They implement something poorly, or incorrectly, or introduce a bad pattern into the project. Then they continue to amplify that badness over time, as they continue to copy from it on subsequent work. It spreads like a virus.
> Keeping these seeds out of the project is very difficult, and cleaning up the rot is very difficult.
My "aha" moment was when I realized this goes for all spheres of life where this tech is/will be introduced.
ACM, that's the term that I'd been looking for - and your paper explains it clearly. At the end, most of LLM problems are context problems. Getting the correct knowledge into its context window without overpopulating it is the actual engineering effort for most agents. And the solution you present seems promising.
Both compaction with validation and predictive fetching are the way to go.
I do not want to write an implementation for this myself, and if Synap is that implementation, I'd like to ask you a few questions:
1. Does it work with context that's not just agent conversations, but rather documents?
2. Is it better than RAG on large dataset?
3. What does on-prem options look like?
1. Yes, works on docs, agent conversations, human-conversations from different sources (Slack, JIRA, etc.). We have connectors for some of these as well; so it is plug and play
2. conventional RAG recall accuracy is quite low (50-60%) and latency is pretty high (seconds). But worst is the precision; you end up context stuffing to get acceptable recall
3. We do offer on-prem deployments, but only on sizeable annual contracts
For me, if I put costs in my architecture, I'm limited heavily and that can easily change how memory is shaped dramatically.
The opposite, putting memory in architecture, is not true. Memory engineering first, then full scale in the system then costs considerations.
In addition, I believe this is future friendly. Because AI is advancing and getting smarter and cheaper everyday.
I couldn't find a guide on this so I share my basic thoughts.
Context drift on retries is easily the most annoying part of this setup. Locking down the tool payload schema first was the only thing that worked for us
Ive never read a paper cover to cover before but after wrestling with opus 5s english this paper is such a relief to read, its like my eyes has been washed off opus stink
What's also the biggest killer is code rot. Agents are particularly good at death by thousand cuts. They implement something poorly, or incorrectly, or introduce a bad pattern into the project. Then they continue to amplify that badness over time, as they continue to copy from it on subsequent work. It spreads like a virus.
Keeping these seeds out of the project is very difficult, and cleaning up the rot is very difficult. It also seems like a hard problem to solve because following the existing codebase is something that is good when the code is good, but bad when it is bad. So, seemingly, the solution means more thinking and evaluation for every change that is being made.
Also, with "self-bias", models are also likely to grow new content around parts that exhibit their own subtle fingerprints from the past.
So it becomes hard to steer them onto what humans prefer or fixed.
> Keeping these seeds out of the project is very difficult, and cleaning up the rot is very difficult.
My "aha" moment was when I realized this goes for all spheres of life where this tech is/will be introduced.
Both compaction with validation and predictive fetching are the way to go.
I do not want to write an implementation for this myself, and if Synap is that implementation, I'd like to ask you a few questions: 1. Does it work with context that's not just agent conversations, but rather documents? 2. Is it better than RAG on large dataset? 3. What does on-prem options look like?
1. Yes, works on docs, agent conversations, human-conversations from different sources (Slack, JIRA, etc.). We have connectors for some of these as well; so it is plug and play 2. conventional RAG recall accuracy is quite low (50-60%) and latency is pretty high (seconds). But worst is the precision; you end up context stuffing to get acceptable recall 3. We do offer on-prem deployments, but only on sizeable annual contracts
For me, if I put costs in my architecture, I'm limited heavily and that can easily change how memory is shaped dramatically. The opposite, putting memory in architecture, is not true. Memory engineering first, then full scale in the system then costs considerations.
In addition, I believe this is future friendly. Because AI is advancing and getting smarter and cheaper everyday.
I couldn't find a guide on this so I share my basic thoughts.