Definitions for AI assistant, chief of staff, and the rest
Short, plain-English entries for the terms we use across alfred_. Each entry is the canonical definition, what makes the concept distinct, and links to the longer reads if you want to go deeper.
-
Agentic AI
Agentic AI describes the broader architectural pattern of AI systems that act autonomously across multiple steps to achieve a goal, rather than producing a single response to a single prompt. It encompasses agents (individual systems) and multi-agent setups (coordinated systems). The 'agentic' label became dominant in 2025-2026 as the industry shifted from chatbots toward action-taking systems.
-
AI Agent
An AI agent is a software system that uses a large language model to plan and execute multi-step actions toward a stated goal. Unlike a chatbot that only converses, an agent invokes tools (send email, update calendar, query a database) and adapts based on results. The distinguishing trait is autonomy: agents act, not just answer.
-
AI Agent vs AI Assistant
An AI assistant is software that responds to user prompts and helps with tasks on demand (ChatGPT, Claude, Copilot). An AI agent is software that pursues a standing goal autonomously, taking multiple actions across tools without per-step prompting (alfred_, Lindy). The terms overlap, but the practical distinction is who initiates each action — the user, or the software.
-
AI Chief of Staff
An AI chief of staff is software that uses artificial intelligence to coordinate and prioritize an executive's workload across email, calendar, and tasks — triaging the inbox, drafting replies, extracting action items, surfacing meeting context, and delivering a morning briefing of what needs attention.
-
AI Email Assistant
An AI email assistant is software that uses large language models to read, triage, draft, organize, and act on email on the user's behalf. Capabilities typically include automated triage by urgency, voice-matched draft replies, task extraction from threads, and follow-up tracking. The category distinguishes itself from AI email clients (faster interfaces) by doing the work rather than speeding up your doing.
-
AI Executive Assistant
An AI executive assistant is software that automates the work a human executive assistant traditionally handles: email triage, calendar management, follow-up tracking, meeting prep, daily briefings, and stakeholder communication. The category emerged in 2024-2026 as LLM-based agents became reliable enough to handle the unstructured work that previously required a human.
-
AI Meeting Notetaker
An AI meeting notetaker is software that joins or captures meetings to produce automated transcripts, AI-generated summaries, and action item extraction. The category includes both bot-based tools that join Zoom or Meet calls as a participant (Fathom, Fireflies, Otter, tl;dv) and bot-free tools that capture system audio locally (Granola). Output typically includes a searchable transcript, a structured summary, and a list of action items.
-
AI Personal Assistant
An AI personal assistant is software that uses large language models to handle the day-to-day administrative work an executive assistant would: email triage, calendar management, task tracking, follow-up monitoring, and daily briefings. The category covers tools that integrate with email and calendar to act on a single person's behalf, distinct from team-collaboration AI or specialized chatbots.
-
Async Communication
Async (asynchronous) communication is information exchange that doesn't require all parties to be present at the same time. Email, recorded video, written documents, and threaded chat are async; meetings, phone calls, and live chat are sync. Distributed and remote-first teams default to async to enable work across time zones and minimize meeting load.
-
Autonomous AI
Autonomous AI is artificial intelligence that pursues a standing goal continuously without per-action prompting — it observes inputs, makes decisions, and takes actions on the user's behalf, in contrast to reactive AI (chatbots, on-demand assistants) that wait for an explicit instruction before each step.
-
Calendar Intelligence
Calendar intelligence is software that goes beyond passive calendar storage to actively read, reason about, and surface meaningful patterns in a user's schedule — including meeting prep context, conflict resolution, focus-time protection, and connections to related email and tasks. It treats the calendar as a signal source, not just an event list.
-
Context Switching
Context switching is the cognitive cost incurred when shifting attention between tasks. Each switch requires the brain to disengage from the current task, load context for the new one, and (eventually) reload context when returning. UC Irvine research (Gloria Mark) measures an average of 23 minutes 15 seconds of refocus time after each interruption.
-
Daily Brief
A daily brief is a concise written summary that consolidates an executive's information sources — email, calendar, tasks, follow-ups — into a single scannable document showing what changed, what needs attention, and what decisions are required today. The format originated in military and government briefings; AI now generates equivalent briefs automatically from a user's actual workflow.
-
Deep Work
Deep work, coined by Cal Newport in his 2016 book of the same name, is professional activity performed in a state of distraction-free concentration that pushes cognitive capabilities to their limit. The opposite is shallow work — non-cognitively demanding, logistical tasks often performed while distracted. Research suggests most people can sustain about 4 hours of deep work per day.
-
Eisenhower Matrix
The Eisenhower Matrix is a prioritization framework that sorts tasks into four quadrants based on urgency and importance: Do (urgent and important), Schedule (important but not urgent), Delegate (urgent but not important), and Delete (neither). Attributed to Dwight Eisenhower and popularized by Stephen Covey, the matrix is the most widely-cited tool for prioritization in knowledge work.
-
Email Bankruptcy
Email bankruptcy is the deliberate decision to stop processing your email backlog and start fresh, typically by archiving or deleting everything older than a cutoff date and sending a public acknowledgment to your contacts. The term, coined by law professor Larry Lessig in 2004, treats overwhelming backlogs as financial bankruptcy: write off the unrecoverable debt and reset.
-
Email Batching
Email batching is the practice of processing email in scheduled blocks (typically 2-4 times per day) rather than checking continuously throughout the workday. The discipline reduces context-switching cost (each check incurs a 23-minute refocus tax per UC Irvine research) and concentrates email work in focused windows where decisions are faster.
-
Email Triage
Email triage is the process of evaluating each incoming email and routing it to the appropriate next action — reply now, schedule, delegate, archive, or escalate — based on urgency, importance, and required effort. Originally a medical-prioritization metaphor, it's now a core inbox-management discipline that AI tools can perform autonomously.
-
Focus Time
Focus time is a block of uninterrupted, high-cognitive work scheduled on the calendar and protected from meetings and notifications. The practice originated in software engineering teams but now applies across knowledge work. Modern calendar tools (Google Calendar Focus Time, Outlook Focus Time, Reclaim, Clockwise) treat focus time as a first-class calendar event type.
-
Follow-Up Tracking
Follow-up tracking is the systematic monitoring of open commitments and unresolved messages: emails you sent that haven't received a reply, deliverables you promised others, and items others promised you. Done well, it prevents dropped balls and missed deadlines. Manual follow-up tracking relies on memory or sticky notes; AI tools maintain a continuous register from email and calendar data.
-
Getting Things Done (GTD)
Getting Things Done (GTD) is David Allen's productivity method, introduced in his 2001 book of the same name. The core principle: a clear mind requires moving every commitment out of memory and into a trusted external system. The method has five steps — capture, clarify, organize, reflect, engage — and remains one of the most influential personal productivity frameworks in the world.
-
Inbox Bankruptcy
Inbox bankruptcy is the deliberate decision to archive or delete all email older than a chosen threshold (typically 30 days) without reading or processing it. The term, popularized by Merlin Mann and Larry Lessig, treats overwhelming email backlogs as a real bankruptcy — one where the cost of clearing exceeds the value contained — and resolves them by writing off the debt.
-
Inbox Zero
Inbox zero is the state of having no emails sitting in your inbox waiting for a decision. Coined by Merlin Mann in 2007, the method treats the inbox as a transient processing queue: every message gets handled (replied, archived, delegated, deferred, or deleted) in a single pass, never re-read. The 'zero' refers to attention, not message count.
-
Morning Routine
A morning routine is a consistent sequence of actions performed at the start of the workday designed to establish focus, reduce decision fatigue, and protect the highest-cognitive hours from reactive work. For knowledge workers in 2026, common elements include reviewing a daily brief, planning priorities, blocking focus time, and processing the first email batch before opening the broader inbox.
-
One-Touch Email
One-touch email is the discipline of making a decision and taking action on each email the first time you open it, rather than reading and deferring without resolution. The rule: every opened email gets one of five actions (delete, archive, reply, delegate, defer to task). The discipline reduces the cognitive cost of re-reading the same messages across multiple inbox visits.
-
Proactive AI
Proactive AI is software that initiates actions or surfaces information without being prompted, based on patterns, schedules, or detected events. Examples: an AI assistant that drafts your morning briefing before you wake up, or one that flags an email pattern that suggests a missed follow-up. The opposite is reactive AI, which only acts when the user asks.
-
Reactive AI
Reactive AI is software that produces output only in response to a user prompt or invocation. ChatGPT, Claude, Gemini, and most in-app AI features are reactive: they wait. The model is powerful when invoked but generates zero value when ignored. The opposite is proactive AI, which initiates action without being asked.
-
Shared Inbox
A shared inbox is a single email address (typically support@company.com, team@company.com, or similar) that multiple team members can access and manage collaboratively. Shared inbox tools add assignment, status tracking, internal discussion, and analytics on top of the underlying email infrastructure to prevent dropped threads and duplicate replies.
-
Snooze Email
Snoozing an email hides it from the inbox view until a specified time, when it reappears at the top as if newly arrived. Gmail introduced the feature in 2018; Outlook, Spark, Superhuman, and AI email assistants now support it. Snoozing converts an email from 'thing I'm thinking about' to 'thing I'll think about later' without losing the context.
-
Task Extraction
Task extraction is the process by which AI software identifies action items, commitments, and deadlines embedded in email, chat, and meeting content — then captures them as structured tasks in a task list automatically, without manual entry. It captures both inbound requests (things others asked you) and outbound commitments (things you said you'd do).
-
Time Blocking
Time blocking is the practice of dividing your workday into discrete blocks of time on the calendar, each assigned to a specific task or category of work. Unlike a to-do list (which lists tasks without scheduling them), time blocking commits each task to a window. Cal Newport popularized the method; AI scheduling tools like Motion automate it.
-
Two-Minute Rule
The two-minute rule, from David Allen's Getting Things Done method, states: if a task takes less than two minutes to complete, do it immediately rather than capturing it for later. The reasoning is that the overhead of tracking and re-processing a tiny task exceeds the cost of just doing it. The rule applies in both email triage and general task management.
-
Voice Tone Matching
Voice tone matching is the AI capability of drafting messages in a specific user's voice, learned from their sent emails or other writing. Strong implementations adapt per recipient (more formal to clients, casual to teammates) and per context (apology vs proposal vs follow-up). The capability distinguishes drafting AI from generic templating.
-
Voice-Matched Email Drafting
Voice-matched email drafting is AI-generated email replies written in a specific user's voice — phrasing, tone, sentence rhythm, sign-offs, and recipient-specific style — by learning patterns from that user's own sent-folder history rather than using a generic LLM template.
-
Working Memory
Working memory is the cognitive system responsible for temporarily holding and actively manipulating information needed for current tasks. Capacity is famously limited (Miller's classic 7±2, refined to roughly 4±1 chunks in modern research). When working memory load exceeds capacity, attention narrows, decision quality drops, and cognitive performance degrades — a key reason email overload feels overwhelming.