Frequently asked
questions.

We're in the middle of a data crisis. Big Tech companies are building trillion-dollar AI infrastructure on a foundation of unverified data. They don't own it, they didn't get permission to use it, and they certainly don't give creators any receipts about what they took or how they used it. This creates a world where data has become the most valuable asset but the least owned by the people who created it.

Because edge-first, privacy-first infrastructure is complex. Building it requires rethinking how data is processed, encrypted, and traced. Big Tech doesn't want this problem solved. They're incentivized to keep building bigger data centers. The downstream profits from bad data far exceed any interest in fixing the root cause. We're solving a problem the trillion-dollar companies have no interest in touching.

No. We're not an AI company. We're a data infrastructure company. Our tech isn't AI. It's patent-pending edge-first infrastructure that processes data locally and gives every piece of data its own receipt. While AI companies keep wrapping bad data in new SaaS offerings, we're fixing the foundational problem. No LLM will magically make bad data better.

Everyone. Data is the next currency. 90% of the world's data was created between 2021 and 2023, and the pace is accelerating. Faster creation doesn't mean better data; it means more bad data is created every day. MemoryIntelligence™ is designed for everyone from individuals who want to own their memories to enterprises that need verified data pipelines. If you create data, generate insights, or make decisions based on information, you need better data.

No. MI™ is not AI. It's the infrastructure layer that sits underneath AI. Think of it this way: every AI tool in your stack (whether that's ChatGPT, Claude, Gemini, or your own fine-tuned model) assumes it's getting good data. It's not. MI™ is the ingestion and structuring layer that makes sure the data going into those tools is verified, structured, and traceable. We don't generate answers. We make sure the answers your AI gives you are actually worth trusting.

No. You use MI™ with your favorite LLM. MI™ is not meant to replace any tool in your stack. It's the layer underneath all of them. Keep using Claude, keep using Gemini, keep using whatever works for you. MI™ structures and verifies your data before it reaches any of those tools, so they all perform better. If you use an LLM, MI™ makes it better. If you use Glean, MI™ makes it better. If you use NotebookLM, MI™ makes it better. We're upstream of everything.

Neither. Glean searches your existing mess across SaaS tools. MI™ structures data at ingestion so there's no mess to search. Knowledge graphs map relationships in data you already have. MI™ captures and structures the data before it ever reaches the graph. RAG retrieves from existing databases. MI™ builds the database RAG retrieves from. These tools are all downstream. MI™ is the foundation layer that makes every one of them work better.

Today, when you research something in Gemini and then switch to Claude for deeper analysis, you lose all context. You copy-paste, re-explain, start over. MI™ connects the dots. Your structured context follows you between tools: from Gemini to Claude, Claude to email, email to Slack, Slack back to Gemini. No re-explaining. No lost threads. MI™ is the orchestration layer that lets your entire workflow share the same verified memory.

Ingestion is the technical side: taking raw, unstructured data (meeting transcripts, Slack threads, CRM exports) and turning it into structured, verified memory objects with cryptographic receipts. Orchestration is what happens after: your structured data becomes portable across your entire stack. Start research in one tool, hand it off to another, loop back around. MI™ keeps the context intact throughout. Ingestion gives you better data. Orchestration gives you a better workflow. Together, they give you deeper context and seamless knowledge transfer across everything you use.

JavaScript/TypeScript, Python, Go, and Rust. Each language client is a first-class citizen with full feature parity. Install via npm, pip, go get, or cargo; you'll have your first memory captured in under five minutes. Check the quickstart guide to get started.

Most developers capture their first memory in under five minutes. A basic integration with capture and query takes less than an hour. Full production deployments with historical ingestion, listeners, and custom schemas typically take a day or two depending on your data sources.

No. MI™ is designed to layer on top of whatever you already have. It integrates with Postgres, MongoDB, Redis, S3, and more. Your existing database stays your database. MI™ adds a structured memory layer alongside it. Think of it as a data upgrade, not a migration.

Absolutely, and that's where it gets really interesting. MI™ isn't a replacement for LLMs. It's the structured, verified data layer that makes LLMs actually useful. Feed an LLM your MI™ data and suddenly it has real context, real provenance, and real accuracy. Bad data in, bad answers out. Good data in, everything changes.

On your infrastructure, at the edge. MI™ processes data locally first. Nothing leaves your device or server until you explicitly tell it to. There are no data centers sitting between you and your data. You own the infrastructure, you own the data, you own the receipts.

Every memory captured through MI™ gets a cryptographic receipt: a verifiable proof of what was stored, when it was stored, and where it came from. This means your data is traceable and auditable. No more "trust us" from your data provider. You can prove the provenance of every piece of data in your system.

MI™ is built with compliance in mind from day one. Edge-first processing means data never has to leave your jurisdiction. Cryptographic receipts give you built-in audit trails. We're actively pursuing SOC 2 Type II certification. For specific compliance questions, reach out to us directly.

Beta today: every tier is free. No credit card, no time-boxed trial. You opt in, pick a track, and build. After launch we are targeting three public tiers with posted list prices: Capture (~$29/mo), Automate (~$49/mo), and Control (~$99/mo). Limits, receipts, and admin features scale with the tier. Prices are directional until we lock launch; the beta page always has the current breakdown. Founding offer: stay through beta and you get six months free on whichever tier you use most when we go live. For architecture and the pipeline, see the product page.

Yes. For teams that need dedicated support, custom SLAs, on-prem deployment, or volume licensing, we offer enterprise plans through Somewhere Media. Get in touch and we'll scope something out.

Open the beta page, pick the tier that matches what you are building (you can change later), then follow the quickstart for your language. Set your API key and capture your first memory. Most developers are running in under five minutes.

Capture and Automate are live in beta: API access, language clients, ingestion, cryptographic receipts, and automation-scale limits per tier. Control (admin, teams, compliance-forward controls) unlocks closer to launch. Everything is free during beta. Pick the tier that matches what you are building, swap as you grow. No credit card required.

Until we are confident the standard is production-ready—we are not rushing GA. Founding beta users who stay through launch get six months free on whichever paid tier they use most (see the beta page for the current founding offer).

Your memories are yours. You can export or delete everything at any time. If you stay, your data carries forward. If you leave, we delete it.

Yes. During beta everything is free, so you can start on Capture, move to Automate when you need the higher limits, and step into Control once those admin and compliance surfaces ship. Your data comes with you.