Skip to main content

Roadmap

Everything on this page is design intent, not a shipped feature; scope and details can change before release.

Check your understanding, not just your output​

Learning quizzes turn your own tasks into quick comprehension checks. From any task with quizzable content you will be able to tap Check my understanding and get a short quiz tailored to that task, mixing multiple-choice questions with open questions you can answer by typing or by voice; a quiz needs AI configured for the task's category, and one quiz per task runs at a time. Grading is conversational: when an answer is shallow or ambiguous, the AI asks a follow-up — at most two per question — before judging, and you can end probing, skip a question, or reveal the answer at any time; every miss is explained, grounded in the task content the question came from. Each quiz ends with a score, a label, and a gap list, and keeps its own timeline — questions, answers, follow-ups, and grades — so you can reopen it from the task later. Grades are AI feedback about one quiz, never a credential or a comparison with other people, and quiz history stays private, syncable, and deletable like other agent data.

The planned progression:

  1. Quiz any eligible task on demand, with a quick-check or deep-dive choice and fresh questions each time.
  2. More surfaces and gentle suggestions — projects, days, and agent reports, plus an optional "Quiz yourself on this?" card with strict caps and disable controls.
  3. Consented workspace evidence so engineering quizzes can draw on repository diffs and test summaries through a read-only integration.
  4. History insights that surface recurring gap themes over time.

Questions and grades come only from content Lotti actually captured — thin content yields fewer questions rather than filler, suggestions never run AI before you accept, and quizzes never block completing a task.

Stay close to the people who matter​

Relationships become a first-class part of Lotti: one long-running entry per person, with its own timeline, alongside your tasks and projects. After you see someone, you will be able to log a check-in — met in person, phone call, video, or message — capturing what you talked about, how the interaction felt, and what to pay attention to or avoid next time. Relationships link to tasks, so the call you need to prepare or the visit you promised sits right on that person's timeline. You will also be able to connect a relationship to a contact from your address book — person by person, never as a bulk import — and call or message them straight from the briefing; when you return to Lotti, it offers to log that interaction as a check-in. Mark a relationship as important and Lotti gently reminds you to check in when too much time has passed — you choose the rhythm per person, and only relationships you marked important ever produce reminders. Each relationship gets its own dedicated agent — the same durable agents that look after tasks and projects — which keeps the picture current and, before you next meet someone, prepares an executive briefing: how things are going, the topics from recent check-ins, how conversations have felt over time, and what to bring up or steer around — like a thoughtful assistant preparing you for the conversation.

The planned progression:

  1. Relationships with a timeline — a person-centered entry with linked notes, photos, and tasks.
  2. Check-ins — quick structured capture of each interaction: type, sentiment, topics, and next-time guidance.
  3. Reminders for important relationships — cadence-based nudges you control per person.
  4. Executive briefings — an AI-prepared summary before your next conversation, grounded only in your own check-ins.

Relationship data gets the strictest treatment Lotti has: it is stored locally on your devices, never on a Lotti server — there is none. If you turn on sync, entries travel end-to-end encrypted between your devices, so the relay server only ever handles ciphertext. Briefings run on local AI models by default; if you explicitly pick a cloud provider, Lotti names it before anything is sent — and with a GDPR-compliant provider offering zero data retention, inference remains transient processing, with nothing stored outside your devices. Handing relationship data to third parties without your explicit choice would be irresponsible — so Lotti simply doesn't.