Bounded preferences
Stated preferences can shape later suggestions without making every passing comment permanent.
A private AI concierge built around your day.
A useful concierge combines context, continuity and initiative with clear limits. Habit is designed to prepare what may help next while keeping permissions, uncertainty and action boundaries visible.
The difference
Context before answers.
A chatbot starts with the prompt in front of it. A private AI concierge can begin with the permitted shape of the day around that prompt.
That context may include the calendar, messages and recent meeting material you choose to connect. Instead of asking you to reconstruct the whole situation, Habit can use those sources to prepare a grounded brief, surface an open decision or gather the material for what comes next.
“Private” does not mean nothing ever leaves your device. Connected services and selected AI processors are part of the data path. For Habit, privacy means explicit permissions, per-principal boundaries, protected stored fields and controls around what can happen next.
Continuity
Remember enough to help, without turning guesses into rules.
Continuity can make assistance feel personal, but only when preferences and relationship context remain correctable.
Relevant relationships
People, commitments and recent interactions can stay connected to the work where they help you decide.
Visible correction
Memory is designed with provenance and a management path so context can be challenged rather than treated as an unchangeable profile.
Initiative
Work between prompts, with restraint.
Initiative should reduce catching-up work, not create a new stream of notifications.
Briefs on a rhythm
A scheduled daily or weekly brief can prepare a decision surface without waiting for a new request.
Suggestions with a reason
A useful suggestion explains the context that made it worth your attention.
Routines you configure
Repeatable assistance should follow an intentional routine rather than inventing authority from past behaviour.
Action boundary
Draft, organise and propose are not the same as send or change.
Habit keeps that distinction explicit. Review-before-action is the default for consequential work.
The assistant may prepare a reply, organise relevant context or propose a calendar action. You can approve, modify or decline the proposal. Eligible calendar additions can run automatically only after you deliberately enable that bounded category; calendar updates and deletes remain review-gated.
This approach is a fit when commitments are spread across calendars, inboxes, meetings, documents and conversations—and when the cost lies in repeatedly gathering the same context. It is not a substitute for professional judgement or permission to act beyond the controls you set.
Examine the permission and action boundariesQuestions
What people mean by an AI concierge.
- What is a private AI concierge?
- A private AI concierge is an assistant that uses authorised context and bounded continuity to prepare useful work across your day, while keeping action controls and data boundaries visible.
- How is that different from a chat assistant?
- Chat is one interface. The concierge model also includes scheduled briefs, connected context, memory designed to be correctable and proposals that can wait for review.
- Does it need access to everything?
- No. The useful boundary is the supported sources you deliberately connect. More access is not automatically better context.
- Can an AI concierge act for me?
- Habit prepares and proposes actions. Consequential work requires review by default; only eligible, bounded categories can be automated after an explicit opt-in.
- Does personalisation improve over time?
- Continuity can improve as relevant preferences and relationship context are retained, provided those memories remain sourced, inspectable and correctable.