Terms
Every Mosaic term in one line, A to Z.
68 terms.
Terms, A to Z
| Term |
|---|
|
Activity A living log of what happened - conversations, events, decisions. The raw material the nightly loop learns from. |
|
Activity Integration The scheduled path that brings new material into Activity, proposes useful knowledge updates, and does light upkeep on the facts it touched. |
|
Activity Integration The scheduled path that reads new material in Activity, proposes useful knowledge updates, and does light upkeep on the facts it touched. |
|
Admin Manages people, teams, and settings for the organisation. Like an owner, an admin automatically has full (lead) access to every pod. |
|
Adopt Copy a skill from the Mosaic catalogue into your own skills - the copy is fully yours from that point (editable, can be turned off), and never changes Mosaic's original. |
|
Audit log The tamper-evident record of every change to the brain - who proposed it, who approved it, and when. Nothing is ever quietly edited. |
|
Author access (skills) A narrower grant a pod lead can hand someone right from the pod's grant form: write, edit, and adopt this pod's skills, without giving them full editor access to the rest of the pod. |
|
Automation An arrangement for Mosaic to run a skill, with a trigger, execution choices and a destination for the result. A schedule is one kind of trigger. |
|
Automations panel The one place on a pod to see everything that runs there automatically - the built-in overnight and health-check jobs, plus your team's own scheduled skills - and how often each one fires. |
|
Brain Architect Looks at how your brain is used and built, and suggests what would improve it, ranked by impact. Run it from the Brain page, or ask your coworker what the brain is missing. |
|
Brain budget How much of your brain your coworker reads on every message. The must-know parts are read every time; the rest is fetched only when it's relevant, so each reply stays fast and focused. |
|
Built-in job A scheduled process Mosaic runs for you out of the box - the overnight brain loop, a weekly health check, and a few more. An admin can change how often each one runs, or turn it off. |
|
Connector A link that lets the bot act in an outside app (Gmail, Asana, a CRM) using your permission. |
|
Core A pod's identity and current facts: who you are, what you do, the state of things now. The most authoritative layer. |
|
Coworker The AI teammate your organisation works with - in Slack, on this site, and inside your other tools. It reads the brain, does real work, and asks before anything consequential. |
|
Credit The unit your coworker's work is measured in - every reply, scan, or scheduled job draws down a few credits. Your card is charged in dollars (the monthly plan or a top-up); credits are what that buys, so day to day you never have to think in dollars. |
|
Curate The fold-in action: a person deliberately pulls a catalogued source document's actual content into a pod's Activity log right now, rather than waiting for the next scheduled pull. |
|
Disputed Two sources disagree on the same fact, so both are flagged and paused from use until a pod lead resolves which one is right. |
|
Editor access Everything viewer access allows, plus proposing and curating changes to the pod's knowledge. |
|
Element A single tracked fact or note in the brain - one line of Core State, one rule, one entry. Every fact-shaped piece of a document is one, with its own source and trust level. |
|
Fetched when needed Part of the brain your coworker looks up only when it's relevant to the question, instead of reading it every time - keeps everyday replies quick. |
|
Finding What the Brain Architect surfaces: what it saw, why it matters, what it suggests, and how much difference it expects to make. Nothing changes until you act on it. |
|
Freshness Whether a fact has been confirmed recently enough - based on when it was last checked, not on whether it happens to still be true. A stale fact isn't wrong, just due for a look. |
|
Guardrails What must never happen without explicit sign-off. |
|
Heartbeat The proactive watcher that suggests useful recurring jobs (a morning brief, a weekly audit) - approve once, then autopilot. |
|
Inferred Worked out by the coworker from patterns or a calculation, not pulled from a source or said by anyone - worth confirming before relying on it for anything high-stakes. |
|
Kind What kind of knowledge an element is - a fact, a metric, a decision, a rule, a logged event, or someone's stated view. Assigned automatically from how it reads; a reviewer can correct it. |
|
Lead access Everything editor access allows, plus approving changes before they go live and managing who else has access to the pod. |
|
Linking Attaching a source area (a folder, label, channel, project) to a pod. Linking is the whole decision: content flows into that pod's Activity automatically from then on, through the connection you linked it with - there's no separate dial to set afterward. Unlink it to stop. |
|
Live (effective) The approved version everyone sees. The opposite of a draft or a proposal still in review. |
|
MCP The standard way other AI tools (Claude, ChatGPT, Cursor) plug into your brain over one secure gateway. |
|
Member The everyday role for getting work done. A member's access to any given pod - view, edit, or lead - is whatever they've been granted there, directly or through a team. |
|
Monthly allowance The plan credits your workspace gets at the start of each month. Use-it-or-lose-it: a new month resets the plan credits, they don't roll over. Bonus credits (your starting credits, rewards and top-ups) are kept and spent first. |
|
Org brain Your organisation's curated, reviewed knowledge - the facts, decisions, and ways of working the coworker relies on. Kept trustworthy through review, never a raw document dump. |
|
Org Signal A pattern worth every pod knowing, shared at the org level - graduates once more than one pod's own evidence backs it, or a pod lead explicitly promotes their pod's own graduated pattern. |
|
Owner Runs the whole organisation - settings, billing, everything - and is the only role that can hand ownership to someone else. An owner automatically has full (lead) access to every pod. |
|
Pod A governed home for one area of knowledge (a team, function, or topic) - like a team wiki, but with review and approval built in. |
|
Pod Signal A pattern noticed in one pod's own activity - starts provisional (seen once), graduates once the same pod sees it again. A pod lead can promote a graduated pattern to Org Signal when it's worth every pod knowing. |
|
Proactive AI What Mosaic does on its own clock, without being asked: keeping the brain growing and true, and suggesting what's worth your attention. Every part of it is yours to turn down or off. |
|
Proposal A suggested change to a document - a title, the reason, and the edit - waiting to be approved. Nothing changes until it is. |
|
Protocol How your team operates: the playbooks, processes, and decision rules. |
|
Quorum How many approvals a change needs before it goes live - one pod-lead approval, never the author's own. |
|
Read every time Part of the brain your coworker reads on every single message - the hard stops, rules, and core facts it must never miss. |
|
Schedule access (skills) A narrower grant a pod lead can hand someone right from the pod's grant form: put this pod's skills on a recurring schedule, without giving them full editor access to the rest of the pod. |
|
Search One question across the whole brain at once, scoped to what you're allowed to see. The same search powers any connected AI tool, so it finds exactly what you would. |
|
Seed pending Text that hasn't been turned into a tracked, sourced fact yet - it still reads fine, just without a trust chip or freshness check until it's next edited or reconciled. |
|
Segregation of duties You can't approve your own change - a different person must, so there's always a second pair of eyes. |
|
Sensitive Fenced below its usual boundary - a fact or tag marked this way is left out when finding things across teams, and shows as “withheld” instead of the real content to anyone outside its own pod. |
|
Session One Slack conversation the bot captured - the unit of activity it remembers. |
|
Skill A reusable method your coworker follows: what it needs, the steps to take, and what a good result looks like. You can ask for it or arrange an automation to run it. |