AI UNDERDOGS
Back
0:00 / 1:00
#061

Memoriq

memoriq.me/
privacy toolAI memoryopen sourceself-hostable
100 views💬 0 comments🔗 0 visits

Your AI conversations deserve an encrypted vault

WHAT IT SOLVES

Your ChatGPT, Claude, Gemini and Grok chats live in four separate silos — lose one account or switch providers and your history vanishes

WHY IT'S INTERESTING

Privacy-first mindset

Encrypt before upload

Every conversation is encrypted on-device before it ever hits their servers. The author doesn't treat 'security' as marketing copy — zero-knowledge encryption plus open-source code, auditable by anyone

Pragmatic design

100 MB hosted or self-host unlimited

Free accounts get 100 MB hosted storage. Want more? Clone the repo and self-host with no limits. No lock-in, no upgrade nag — a refreshingly honest pricing posture for an AI tool

"conversations are encrypted on your device before anything reaches our servers"

giekaton

TECH GUESS

Likely React/Next.js frontend, serverless backend, Web Crypto API for client-side encryption

DEEP DIVE

\n## A 2-Point Show HN That Hits a Real Nerve\n\nMemoriq landed on Hacker News with 2 points and 0 comments — essentially invisible by the platform's standards. But low karma doesn't mean low value. The problem developer giekaton tackles is painfully real for anyone juggling multiple AI platforms: you've got three months of prompt engineering in ChatGPT, writing drafts in Claude, important research threads in Gemini — and if any one of those accounts gets suspended or you decide to switch, those memories vanish.\n\nThis isn't hypothetical. OpenAI account bans and Claude's intermittent history losses are well-documented in the community. giekaton's motivation is straightforward: AI conversations are intellectual assets you spent time and thought to produce, yet they're scattered across four different services, none of which you actually control.\n\n## Zero-Knowledge Encryption, Actually\n\nThe market isn't short on "AI conversation management tools," but most just bolt security on as an afterthought. Memoriq inverts the architecture: conversations are encrypted on your device using the Web Crypto API before anything reaches the server. The developer's exact words: "encrypted on your device before anything reaches our servers." This means the operator of memoriq.me — including giekaton himself — cannot technically read your content. Zero-knowledge isn't marketing copy here; it's an architectural constraint.\n\nThe code is open source on GitHub. You can audit the encryption implementation yourself, or skip trust entirely and self-host. This design choice is unusual for a SaaS product. Most founders' instinct is to lock data in and build walls; giekaton's instinct is to make it easy for you to leave.\n\n## 100MB and the \"We Won't Nudge You to Upgrade\" Pricing Philosophy\n\nFree accounts get 100MB of hosted storage — not a huge number if you're dumping full conversation histories. But Memoriq's pricing logic isn't about selling you more space. The free tier is an entry point, and the ultimate option is open-source: clone the repo, self-host, unlimited storage, zero cost.\n\nThis pricing posture is rare in the AI tooling ecosystem. The dominant playbook is a free tier designed to tease, then walls of limits (conversation caps, file size restrictions, feature gutting) to push paid upgrades. Memoriq flips it: stay free forever if you want, or manage your own server for more. No locks, no nudges. It might not be a great business, but it's genuinely respectful to users.\n\n## Who Should Use It, and Who Shouldn't\n\nIf you use a single AI platform, have modest conversation volume, and no cross-platform archiving needs, Memoriq adds an unnecessary layer. But if any of these describe you: actively using 2-3 AI products for serious development or research, have experienced account issues that wiped history, or feel uneasy storing all conversations on third-party servers — Memoriq is worth 10 minutes of your time.\n\nThe limitations are honest: import workflows depend on each platform's export capabilities, the UI is still basic, and search functionality hasn't been stress-tested at scale. The 100MB cloud quota is tight for heavy users, and self-hosting requires technical chops. giekaton is clearly a solo developer, and iteration speed depends on community feedback — which, given 0 comments on HN, is currently nonexistent.\n\n## 2 Points, 0 Comments, but Maybe That's Where Good Indie Projects Start\n\nMost Show HN posts sink without a trace. Memoriq has near-zero visibility right now, but it did one thing right: turned "AI memory ownership" from a slogan into a working product. The encryption is real, the open source is real, the no-lock-in is real. For developers and product people treating AI conversations as working assets, demand for this kind of tool will only grow — even if Memoriq itself doesn't scale, the path it validates is sound."}

📍 Source: hn📅 2026-06-13Original post →Visit site →
Ad
Ad slot (AdSense unit renders here once connected)

Discussion (0)

Sign in with GitHub to post
  • No comments yet — be the first.

Related