July 12, 2026
Memory lane or scam lane?
Show HN: Adaptive Recall, persistent memory for AI assistants over MCP
AI memory tool promises to remember everything, but commenters smell hype and reach for markdown files
TLDR: Adaptive Recall is pitching a system that helps AI assistants remember, connect, and forget information more intelligently over time. Commenters were mostly skeptical, mocking the sales-heavy website, worrying about privacy, and joking that simple text files might still beat the fancy promise.
A new project on Hacker News is pitching itself as a kind of super-memory for AI assistants: a system that can save things, bring them back later, connect related facts, and even forget old stuff naturally. The sales pitch is big — the tool claims it improves over time, learns from usage, and goes way beyond the usual “store some text and search it later” approach. On paper, it sounds like the dream fix for chatbots that keep forgetting what you told them five minutes ago.
But the comment section? Absolutely not buying the fairy tale just yet. The strongest reaction was pure distrust. One user flat-out called it a “slop website” and said it “reads like a scam,” roasting the vague branding and lack of company detail. Another took the privacy angle, saying they already send enough personal data to one big AI company and have zero interest in adding another middleman to the mix. In other words: cool memory trick, but who exactly is remembering my stuff?
And then came the comedy. The funniest jab was at the free plan: “500 memories for free!” which a commenter instantly compared to a tiny markdown file — and then twisted the knife by saying a folder full of plain text files plus search tools would still be better. Ouch. So while the product is selling futuristic memory magic, the crowd’s mood is more like: show us it works, explain who you are, and maybe stop acting like a notes folder hasn’t already entered the chat.
Key Points
- •Adaptive Recall is presented as a persistent memory system for AI assistants with store, recall, and forget functions.
- •The product says it uses four search strategies in parallel: vector similarity, temporal recency, full-text keyword search, and knowledge graph traversal.
- •Ranking is described as using ACT-R activation modeling, incorporating recency, access frequency, entity connections, and validated confidence.
- •The system claims to extract entities and relationships automatically and manage memories as evolving records that can gain confidence or fade over time.
- •Adaptive Recall offers eight tools and supports both MCP-based integrations, including Claude Code, and HTTP REST with bearer-token authentication and JSON I/O.