Remembered from a chat with Claude · used by Gemini
“Trip to Japan in April: vegetarian, prefers trains, budget about $3,000.”
Mark as wrong · outdated · only for this project
Inkfold brings ChatGPT, Claude, Gemini and other models together in one place. It remembers your projects, preferences and past conversations, gives each model the context it needs, and can put several models to work on the same question. You decide what it keeps.
Mark as wrong · outdated · only for this project
Past conversations import from ChatGPT, Claude, Gemini, Copilot and Grok exports.
Inkfold keeps what matters from your conversations: your projects, your preferences and the facts you have settled. It links people, places and topics into a map of your context, so ChatGPT today and Claude tomorrow both start from what you already said. Each answer shows which memories it used, and one click marks a memory as wrong, outdated or only for one project. Past conversations import from ChatGPT, Claude, Gemini, Copilot and Grok exports.
Inkfold works with the major model vendors: OpenAI, Anthropic, Google, DeepSeek, Mistral, xAI and more, with your own key for each or one OpenRouter key for all of them. A spending view shows what each model costs you, with budget rules. Automatic model choice by task and cost is coming.
For questions that matter, Inkfold asks more than one model. Compare shows answers side by side. Council gathers several answers into one. Debate has models challenge each other until they reach an agreed answer. Pipeline works in stages: plan, draft, check and polish.
The composer can improve, rephrase or tune a prompt for the model you picked. A shared library holds personas, playbooks and routines that people publish, rate and adapt, such as a research playbook that goes from sources to citations to a summary (beta).
Browse the shared library (beta) → · More on prompt improvement →
The mode is chosen per conversation. Smart and Private modes store content encrypted at rest; Incognito keeps nothing after the conversation ends. The mode decides how much Inkfold can read.
Inkfold stores and indexes conversation content on its servers, encrypted at rest with a key Inkfold manages, so it can find the right context for you. Inkfold can read Smart-mode content.
Content is encrypted in your browser with AES-256-GCM, using a key derived from your passphrase, before upload. Inkfold stores only ciphertext it cannot read or index, so memory features are limited in this mode, and Inkfold cannot restore a lost passphrase.
Nothing from the conversation is stored after it ends.
People who use more than one assistant and are tired of starting over in each: writers and researchers who keep long projects going, students, consultants and freelancers who work across clients, and small businesses that want one place for their assistants. Teams can share a workspace and its memory (beta).
One vendor's app is often enough. Inkfold is for when you use more than one and want them to share what they know.
| Option | What it gives you | What it leaves out |
|---|---|---|
| One vendor's app (ChatGPT, Claude, Gemini) | Memory inside that one app | Other models never see it; one model's view |
| Chat apps with many models | Many models behind one screen | No shared memory, no models working together |
| Inkfold | One memory across models, several models on one question, privacy chosen per conversation | Private beta; automatic model choice coming |
Other products described from their own published material, September 2026.
Prices are indicative and still in validation; checkout is not open. Your own model keys work on every plan, and Smart, Private and Incognito modes are free on every plan. Larger organizations: sales@inkfold.app.
Twenty places at an indicative $190 a year, price locked while you stay subscribed. Nothing is charged until checkout opens.