Features

What knowmind does, named one by one.

29 features across five areas. Listed here is only what you can actually operate through the interface, your AI tool or the API.

Memory

Filing knowledge that does not expire

Read in full

A memory is a piece of text with context: a note from a conversation, a decision, an excerpt from a contract. knowmind makes it findable by what it says and keeps track of how it changes over time.

  • Create a memory

    from Private

    Save a note, a decision or a fact. From then on it is findable by meaning, even in words other than the ones you used.

    Interface · AI tool · API

  • Upload documents

    from Private

    Longer texts are split into passages and taken into the same body of knowledge as notes. Identical content is recognised by its checksum and not filed twice.

    Interface · AI tool · API

  • Update facts

    from Private

    When a matter changes, the old version is marked as superseded and stays readable with its date. Nothing is overwritten.

    Interface · AI tool

  • Contradiction check when saving

    from Private

    Before filing, knowmind shows similar existing entries and suggests updating the old one instead of creating a second.

    Interface

  • Maintain relations

    from Private

    Two memories are connected with a fixed relation type. The vocabulary is closed, so the body of knowledge does not disintegrate into arbitrary edges.

    Interface · AI tool · API

  • Set the role of a memory

    from Private

    Mark a memory as a decision, goal, action or fact. Search can then filter for it specifically.

    Interface

  • Credential barrier

    from Private

    If an agent tries to write a password or a key into the memory, the entry is rejected at the door.

    applies to every write

Search

Finding it again without remembering the words

Read in full

Four retrieval methods run at once and are merged into one result list: full text, meaning, relations in the graph, and literal matching. A second model then sorts by fit.

  • Search in natural language

    from Private

    Ask a question and get the matching memories with their source. 80 per cent hit rate within the first five results in our own measurement.

    Interface · AI tool

  • Time travel

    from Private

    The same question with a cut-off date: what was known on that day? Only statements that were valid at the time are returned.

    AI tool

  • Research sessions

    from Private

    Several questions that build on each other run as a session with history. Follow-up questions see what has been established so far.

    Interface

  • Silence on empty questions

    from Private

    Questions without a substantive anchor return nothing, rather than filling your model's context with chance hits.

    applies to every search

  • Frequently used rises

    from Private

    Memories that are retrieved often rank higher in later searches. The counting stays inside your own workspace.

    applies to every search

  • Figures on your body of knowledge

    from Private

    Size, growth and distribution of relation types across your own knowledge base.

    Interface · AI tool

Evidence

Showing what the system knew, and when

Read in full

Every entry carries two dates: since when a matter holds, and since when knowmind knows about it. Together with the log of all write operations, that gives a dated timeline across your body of knowledge.

  • Audit log

    from Private

    Who stored, changed, linked or retrieved what, and when. Retention 30 days on Private, 90 days on Pro, 12 months on Team, 24 months on Business.

    Interface

  • Export the log

    from Private

    Download the audit log as CSV or JSON, for an audit for instance.

    Interface

  • Export the body of knowledge

    from Private

    All memories and documents as a JSON file, at any time and in full. No vendor lock-in.

    Interface

  • Nothing is deleted quietly

    from Private

    Outdated statements are superseded and kept with their dates. What held two years ago you retrieve again with a cut-off date.

    applies to every change

Collaboration

One body of knowledge for several people

Read in full

From the Team plan on, several people share one workspace. What one person files, the next finds again in their own AI tool. A change of staff therefore costs no project knowledge.

  • Shared workspace

    from Team

    Five seats on Team, ten on Business. Everyone works on the same body of knowledge.

    Interface

  • Four roles

    from Team

    Owner, admin, member and reader. Assign rights per person and change them whenever you like.

    Interface

  • Invitation by email

    from Team

    Invitees sign in with their login link and are in the workspace at once. Removed members lose their access immediately.

    Interface

  • Add seats yourself

    from Team

    Further seats for 15 euros per person and month, without asking the vendor.

    Interface

  • Register your own applications

    from Team

    Through OAuth, users of your own application sign in against knowmind without you handing out tokens.

    Interface · API

Connections and cost

Inside the tools you already use

Read in full

knowmind connects over the open Model Context Protocol. The same connection carries Claude, ChatGPT, Cursor, Gemini and locally run models. A second AI subscription is not needed.

  • Eleven tools for your AI

    from Private

    Your AI tool can search, store, link, update and use time travel. Eight connection routes are documented.

    AI tool

  • Access tokens

    from Private

    A personal token connects the command line, your editor or your own application. Only a check value is stored, never the token itself.

    Interface

  • API for your own systems

    from Private

    Read and write documents and relations programmatically. Included in every plan, 30 requests per minute on Private.

    API

  • Webhooks

    from Private

    New or changed memories trigger a reaction in your system straight away.

    Interface · API

  • Your own provider key

    from Private

    Store your key for Anthropic, OpenAI or Mistral. It is encrypted per workspace with AES-256-GCM. Several providers from Pro on.

    Interface

  • Tokens and cost per provider and model, as soon as the requests run through knowmind. You set the ceiling yourself.

    Interface

  • Handle plan and cancellation yourself

    from Private

    Switch, upgrade and cancel without a call. The cancellation button works without signing in.

    Interface

Start for freeConnect it to your AI tool