Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

AI Agent Clinic Compares Four Memory Options and Three Context Fixes

AI summary

An AI Agent Clinic episode features Google Cloud engineer Laxmi Harikumar and developer Miles Libbey demonstrating memory and context management in Scribant, a voice-to-action agent built by Libbey with TypeScript and Agent Development Kit. The demonstration compares flat prompting, vector RAG, knowledge graphs, and Memory Bank in the Gemini platform. It distinguishes RAG for knowledge-base searches from memory for keeping user facts current, then applies a 40-word rolling window, a rolling summary, and Memory Bank to different time spans. In one test of about 200 spoken words, the 40-word window used about 86% fewer classifier tokens than resending the full speech stream. The rolling summary used 92 tokens, while Memory Bank stored about 24 tokens of facts compared with about 1,400 tokens to ingest the full conversation. These figures come from a single demo, not standardized benchmarks; 40 words is a suggested starting point to tune for users’ speaking patterns.

In-depth insight

The episode provides a concrete comparison of memory architectures and context-reduction methods, along with token-tracking code that developers can use to measure their own agents. The report explicitly cautions that the measured results are not benchmarks.

Most AI agents have two memory problems. They forget what a user told them last week, and every turn costs more than the one before.

Both come from the same place: how much of the conversation you send to the model, and what you keep between sessions.

In this episode of AI Agent Clinic, Google Cloud engineer Laxmi Harikumar and developer Miles Libbey fix both in Scribant, a voice to action agent Miles built in TypeScript with Agent Development Kit (ADK).

Which memory option should you use?

The episode compares 4 ways to give an agent memory, using one test. A user says "I'm in New York", later says "I moved to San Francisco", then asks "Where do I live now?"

OptionWhat the model getsTradeoff
Flat promptingThe full chat historyEasy to build, but tokens grow every turn
Vector RAGEvery chunk that matches, old and newSaves tokens, but retrieval alone doesn't say which fact is current
Knowledge graphThe current factTracks changes, but you build the extraction, schemas and update logic
Memory Bank in Gemini platformThe current factExtracts, deduplicates and updates facts for you in the background

#Scribant#Agent Development Kit#Memory Bank#Gemini#Google Cloud

Original · 2 min readRead the original
Group chat

No comments yet. Start the conversation.