What is an AI chief of staff, and what does it actually do?

November 11, 2026· 49 min read

What is an AI chief of staff? Vestra blog banner

An AI chief of staff is an AI agent that runs your other AI agents, speaks for you when you're busy and brings you only the decisions that need you. Most tools sold under the name today are personal assistants for one person's inbox and calendar; a real AI chief of staff has a staff to run, and Vestra is the one office where it runs agents on different engines, like Claude Code, Codex and Vestra's own, as one team.

That second sentence is the whole argument of this guide. A chief of staff without a staff is an assistant with a bigger title. Below, we explain what a human chief of staff does, what an AI one actually does day to day, how it differs from an AI assistant, which products use the name, what it costs, and how to set one up.

Quick answer

What it is: an AI agent that runs your team of AI agents for you. It hands out work, checks results, keeps the context and brings you only real decisions.
What it does every day: writes you a morning brief, staffs new work with the right agent, makes sure work is checked before you see it, chases what's late and asks your yes on anything big.
How it differs from an AI assistant: an assistant does tasks for you. A chief of staff runs the agents that do the tasks, and answers to you for all of them.
Why most tools fall short: Claude Code, Codex, Grok Bot, Meta Muse, OpenAI Dots and Hermes are built for one person and work alone. None of them runs agents from other makers as one team.
Where Vestra fits: you get a chief you name. It runs teammates on the engine you pick for each one, Vestra's own, Claude Code or Codex, under one set of rules, with one memory and one record.
Cost: AI chief of staff tools range from free personal agents to enterprise contracts. Vestra starts at $19 a month for the chief and up to 3 teammates.

Key takeaways

A chief of staff is defined by the team it runs. If a tool has no team of agents to run, it's an assistant, however it's named.
Most AI agents today, including Claude Code, Codex, Grok Bot, Meta Muse, OpenAI Dots and Hermes, are built for one person, in one app, usually on one company's AI. They work alone.
In a business, that leaves people as the glue between agents: briefing each one, carrying work between them and chasing every result.
A real AI chief of staff removes that glue work. It runs the agents, keeps one memory, holds one set of rules and talks to you in one voice.
The three things a good AI chief of staff should always ask you first: spending over your limit, deleting something and contacting someone new.
Vestra's chief is live today for one person and their team of agents. Chiefs for every person on a team, working together, are being built now. We say which is which throughout.

Contents

What is an AI chief of staff?
What does a human chief of staff do?
What does an AI chief of staff actually do?
How is an AI chief of staff different from an AI assistant?
Why is a chief of staff without a staff just an assistant?
How do Claude Code, Codex, Grok Bot, Meta Muse, OpenAI Dots and Hermes compare?
Which AI chief of staff tools are on the market?
How does an AI chief of staff run agents on different engines?
What should an AI chief of staff never do without you?
How does an AI chief of staff remember what matters?
What does a day with an AI chief of staff look like?
Can an AI chief of staff replace a human chief of staff?
Who needs an AI chief of staff?
How much does an AI chief of staff cost?
How do you set up an AI chief of staff?
What should you ask before choosing an AI chief of staff?
What are the risks of an AI chief of staff, and how do you manage them?
What's live in Vestra today, and what's being built?
Glossary
FAQ

A note before we start: Vestra is our product. We've compared other tools using their makers' own pages, checked on 10 November 2026, and we've been plain about what Vestra can do today and what is still being built.

What is an AI chief of staff?

An AI chief of staff is an AI agent that runs your other AI agents on your behalf. It hires and briefs them, hands out work, checks results before they reach you, keeps track of everything in flight and brings you only the decisions that are yours to make. It answers to one person: you.

The name borrows from a human job. In companies, a chief of staff sits beside a leader and makes the leader's time go further. They don't do one narrow job. They make sure everyone else's work fits together, and that the leader sees what matters and nothing that doesn't.

An AI chief of staff does the same for a team of AI agents. You might have one agent that researches, one that writes, one that answers customers and one that writes code. Without a chief, you're the one who briefs each of them, passes work from one to the next and checks every result. With a chief, you talk to one agent, and it does the managing.

A short definition

AI chief of staff: an AI agent that runs a team of other AI agents for one person, speaks for that person when they're busy, and brings them only real decisions, with the facts and a suggestion.

Three jobs, one agent

Vestra's handbook describes its chief as doing three jobs, and they make a good test for any product that uses the name:

It runs your agents. It hires them, hands out the work, makes sure every result is checked before it reaches you and teaches them how you like things done.
It speaks for you. When a colleague's chief asks for something, yours answers from what you've already shared, holds the question until you're free, or says no.
It answers to you. It never acts above the freedom you've given it. Anything big waits for your yes, and it can always tell you what every agent on your team is doing and why.

The first job is the one most products skip. Many tools called an "AI chief of staff" read your inbox, tidy your calendar and write you a summary. That's useful work. But it's the work of an assistant, because there's no team underneath for the tool to run.

What an AI chief of staff is not

Not a chatbot. You can talk to it, but the point is the work. Most days you read a short brief and make a decision or two.
Not a single task agent. A coding agent or a research agent does one kind of work. A chief makes sure many kinds of work fit together.
Not a dashboard. You don't fill in forms or click through settings. You say what you need, and the screens are there so you can see the work.
Not a replacement for your judgement. It brings you decisions. It doesn't make the ones that are yours.

What does a human chief of staff do?

A human chief of staff is a senior partner to a leader. They manage the flow of information, set priorities, make sure decisions get made by the right person, keep teams aligned to the plan and speak for the leader when the leader can't be in the room. Their job is to make the leader's time and attention go further.

McChrystal Group, which trains chiefs of staff, describes the role as "a thought partner and strategic ally to leaders". It splits the work into two sides.

Facing the leader:

Plan the leader's schedule and priorities so they support the goals.
Make sure the leader's vision is communicated and tracked.
Manage the flow of information, so the leader sees what matters.
Handle internal problems so the leader can focus outward.

Facing the organisation:

Keep teams and departments aligned with the strategy.
Help groups talk and work together, sharing the leader's view.
Spot decisions that need attention, gather the context and the right people, and bring them to the leader.
Build relationships and speak on the leader's behalf.

Notice how much of that is about other people. A human chief of staff is valuable because they sit between the leader and everyone else. They don't write every report. They make sure the reports get written, checked and read.

Chief of staff vs executive assistant vs COO

These three roles get mixed up, and the mix-up carries over to AI products.

Executive assistant: manages one leader's calendar, inbox, travel and logistics. The work is about the leader's own day.
Chief of staff: manages how the leader's priorities move through the organisation. The work is about other people's work.
Chief operating officer: runs operations as a line executive, with teams reporting to them and results they own.

Most AI products sold as a "chief of staff" today do executive-assistant work. They triage email, prepare for meetings and keep a to-do list. That's not a criticism. It's just a different job, and it matters which one you're buying.

Why the role grew

Leaders hired chiefs of staff because coordination got expensive. The more people and teams there were, the more time went on making sure everyone was working on the right thing and that nothing fell between them. A chief of staff absorbs that cost so the leader doesn't have to.

The same thing is now happening with AI agents. Agents got good at doing work. A business can have a coding agent, a research agent, a writing agent and an inbox agent. But each one works alone, and someone has to coordinate them. Today that someone is usually you.

What does an AI chief of staff actually do?

An AI chief of staff does six things every day: it writes you a morning brief, staffs new work with the right agent, makes sure results are checked before they reach you, chases what's late, answers for you within the rules you set, and brings you only the decisions that need you, ready to answer in one tap.

Here is each one in more detail, using Vestra's chief as the example because it's the product we know best. Vestra's handbook uses a made-up company, Northstar, to show how it works, and so do we. Northstar makes booking software for dental clinics. Its founder, Maya, has a chief she named Atlas.

1. It writes you a morning brief

Every morning, the chief writes you a brief. It isn't a feed of everything that happened. It's what a good chief of staff would hand you over coffee:

What got done while you were away.
The few things that need you.
What it noticed on its own.
What you're waiting on from other people.

You choose when the brief arrives and where. Ask for it shorter and it gets shorter. On a quiet day it says so in one line, because a brief that pads itself out teaches you to stop reading it. In Vestra, the brief also lives in the app as your Home screen and fills itself in through the day.

A real brief from Atlas might read: "Morning, Maya. Three things got done, one needs you." Ledger sent 14 invoices. Keeper answered 19 patient questions. Scout found two rivals who raised prices, checked against their sites. And one decision: Hunter wants to offer a clinic 15% off a yearly plan, which is over Maya's usual 10% limit.

2. It staffs new work

You hand the chief a goal in a sentence. It works out which agent should do it, or hires a new one. In Vestra, you hire a teammate the way you'd brief a new colleague. You say something like "I need someone to get us paid on time." The chief asks what a good manager would ask: what matters most, where the work lives today, and what must never happen without you.

From that conversation, it writes a teammate card: the job in one line, goals you can check, the apps it will use, its first routines and the things it must ask you about first. You read it, change anything you like, and say yes.

3. It makes sure work is checked

"Done" is the most expensive word in any team when it isn't true. A good chief won't tell you something is finished until it's been checked. Vestra's handbook lists four kinds of check:

A test, for code.
A second agent, that reads the work with fresh eyes.
The source, for any fact.
A person, for anything that speaks for you in public.

If the check fails, the work goes back with the reason, and the reason becomes a lesson. When a writing agent drafts a post that misses your tone, the chief sends it back and explains why, so the same mistake doesn't come twice.

4. It chases, so you don't

A lot of any manager's day goes on chasing: "did you hear back?", "any news on that?". A chief tracks every ask until it's answered, including the ones waiting on someone asleep or out for the day. When a reply is overdue, it nudges, and it only brings it to you if a date that matters is going to slip.

5. It speaks for you

When someone asks for something, your chief answers from what you've already said it may share, holds the question until you're free, or says no. In Vestra, this applies between people's chiefs: when a colleague's chief asks yours whether a launch is still on for Friday, yours can answer without interrupting you. (Chiefs working together across a team is on Vestra's Building now list; more on that later.)

6. It brings you only real decisions

What reaches you is a decision, never a status update dressed up as one. Every decision arrives the same way: the question in one line, the facts that matter, what the chief suggests and why, and the choices. In Vestra, the suggested choice is always number 1, so replying "1" works from anywhere.

You also decide how often you're asked. Tell the chief "don't ask me about refunds under $50" and it stops asking. Tell it "always check with me before we email a new clinic" and it always will.

What it doesn't do

A good chief also knows when to stay quiet. Vestra's teammates don't send "just checking in" messages, "all quiet here" updates or "no news yet" notes. A quiet day is a day when the work got done and none of it needed you. If your AI chief of staff pings you more than a person would, it isn't doing the job.

How is an AI chief of staff different from an AI assistant?

An AI assistant does tasks for you: it drafts the email, books the meeting, summarises the thread. An AI chief of staff runs the agents that do the tasks, and answers to you for all of them. One works for you directly. The other manages a team on your behalf.

The difference is easy to blur, because a good chief can also do assistant work. Ask it to move your 2 pm to Thursday and it will. But the reason to have a chief is everything that happens beyond your own calendar.

Side by side

Who does the work: an assistant does it itself; a chief hands it to the right agent and checks the result.
How many agents: an assistant is one agent; a chief runs many.
What it knows: an assistant knows your inbox and calendar; a chief knows what every agent on your team is doing and why.
Who it talks to: an assistant talks to you; a chief talks to you, to your agents, and in a team, to other people's chiefs.
How it handles "done": an assistant reports what it did; a chief reports what was checked, and how.
How it learns: an assistant learns your preferences; a chief turns each correction into a lesson every agent on the team follows.
What reaches you: an assistant often asks you about each task; a chief brings you only the decisions that are yours.

AI executive assistant vs AI chief of staff

An AI executive assistant is a more capable assistant. It handles your email, calendar, meeting notes and follow-ups. Products like alfred_, Martin and Lindy sit here. Several of them describe themselves as a chief of staff, and some of them do part of the job well, like tracking follow-ups or writing a morning brief.

The test is simple: does it run other agents? If it doesn't, it's an executive assistant. That's fine if what you need is a cleaner inbox. It's not enough if what you need is a team of agents doing whole jobs for your business.

AI agent vs AI chief of staff

An AI agent is any AI that can act on its own: browse, write, run code, send email. A chief of staff is a particular kind of agent whose job is to run the others. Every chief is an agent. Most agents are not chiefs.

Why is a chief of staff without a staff just an assistant?

Because the job is defined by the team it runs. A chief of staff's value comes from coordinating other people's work. Take away the team, and what's left is help with one person's own day. Most AI tools called a chief of staff today have no staff of agents to run, so they're assistants with a bigger title.

This matters more now than it did a year ago. The best AI agents have become very good at doing work. Claude Code can ship a feature. Codex can review it. Grok Bot and OpenAI's dots can work in the background across your apps. Meta Muse can run errands. Hermes can run on your own machine and learn your projects.

Every one of them is built for one person. Each one knows only what its own person told it, lives in its own app, and usually runs on its own maker's AI. None of them can see what a colleague's agents are doing. And none of them is designed to run agents from other makers as one team.

What happens when you stack personal agents

Picture a small business that has adopted AI the usual way, one person and one agent at a time:

The founder uses an inbox assistant.
The lead engineer uses Claude Code.
A second engineer prefers Codex.
Marketing uses a writing agent.
Operations uses a bookkeeping agent.

Each agent is good. But the business now has five agents that can't see each other. Vestra's handbook names what happens next. People quietly become the glue between agents. They:

Brief each agent by hand, pasting the same context again and again.
Carry work from one agent to another, because agents can't hand off to each other.
Chase every result, first to ask whether it's done and then whether it's right.
Click through forms, settings and dashboards in every tool.

That's chief-of-staff work. It's just being done by the most expensive people in the business, on top of their real jobs.

What a real AI chief of staff changes

A chief of staff with a staff takes that glue work away:

Brief once. The context travels with the work to every agent that needs it, minus anything private.
Hand off. Work moves between agents without a person carrying it.
Checked. Work counts as done only when it's been checked.
Just talk. You say what you need. The screens show the work.

The stark version

Here is the contrast in one line each.

Claude Code is one developer's coding agent. In Vestra, Claude Code is the engine of a teammate, with a job, a memory and the office rules, and your chief runs it alongside the rest of the team.
Codex is one person's coding agent inside ChatGPT. In Vestra, Codex can be the engine of a second teammate that reviews the first one's work, so a different engine reads the code with fresh eyes.
Grok Bot is one xAI account's set of bots. In Vestra, your chief runs agents from more than one maker as one team.
Meta Muse is one person's errand runner. In Vestra, errands are one teammate's job among many, under one set of rules.
OpenAI Dots start with one primary dot per ChatGPT account. In Vestra, you start with a chief and a team, and each teammate holds a whole job.
Hermes is one person's self-hosted agent. In Vestra, your agents share one memory and one record that you can read and export.

Each of these tools does real work well. None of them is built to be your chief of staff, because none of them is built to run a team.

How do Claude Code, Codex, Grok Bot, Meta Muse, OpenAI Dots and Hermes compare?

They're all strong personal agents, and none is an AI chief of staff. Each is built for one person and one account, usually on its maker's own AI, and each works alone. They can be excellent members of a team. They can't run one. Here's what each does, checked against its maker's pages.

We checked each product's own pages and docs on 10 November 2026. We compared documented features, not hands-on speed or quality. Where a maker's page didn't say something, we've said so rather than guess.

Claude Code

What it is: Anthropic's agentic coding tool. It understands your codebase, edits files, runs commands, and handles bug fixes, tests, refactors, migrations and turning issues into pull requests.

Where it runs: the terminal, VS Code and JetBrains IDEs, the web, the Claude desktop app, iOS and Android, Slack and GitHub, on macOS, Linux and Windows. In the terminal it runs locally and talks directly to model APIs.

Controls: it asks for permission before making changes to your files or running commands. Enterprise plans add role-based access, spend limits and usage analytics.

Price: included with Claude Pro ($20 a month, or $17 a month billed yearly), Max (from $100 a month) and Team (a standard seat is $20 a month billed yearly, $25 monthly). Enterprise is $20 a seat a month billed yearly, with a 20-seat minimum and usage billed separately at API rates.

As a chief of staff: it isn't one, and doesn't claim to be. It's one developer's agent in one codebase. It won't run your bookkeeping agent or tell your sales agent what engineering shipped.

In a team: this is where an office helps most. In Vestra, a teammate can run on Claude Code as its engine. It keeps doing what Claude Code does well, and gains a desk, a job with goals, the shared memory and the office rules. See Vestra vs Claude Code.

Codex

What it is: OpenAI's coding agent, available through ChatGPT plans.

Where it runs: on the Plus plan and up, Codex runs on the web, in the CLI, in an IDE extension and on iOS; the Free and Go plans include limited access in the desktop app.

Price: ChatGPT Plus is $20 a month; Pro plans are $100, $200 or $500 a month; Business is $20 per user a month billed yearly for 2 or more users, or $25 monthly. Enterprise and Edu prices aren't public.

As a chief of staff: it's one person's coding agent. It doesn't run other agents, and it doesn't hold the rest of your business.

In a team: in Vestra, Codex is one of the live engines. Vestra's handbook gives the example of a reviewer teammate on Codex checking the work of a builder teammate on Claude Code, with the chief keeping everyone posted.

Grok Bot

What it is: xAI's agents that work as proactive teammates. They draft decks, run campaigns, book reservations and keep calendars up to date, each on a cloud computer it uses the way a person would.

How it works: a primary bot reports on the other bots working for you, and bots can hand work to one another on a shared task. You reach them on desktop, mobile, Slack and X, by chat, voice call or voice note. They connect to Google Drive, Calendar, Gmail, Granola, Slack, Notion and Microsoft Teams, plus custom tools over MCP or an API. Repeated tasks can be saved as skills or routines.

Controls: bots leave drafts for your approval before sending messages or publishing.

Price: Grok Bot is included in X Premium ($30 a month), Super ($100) and Ultra ($200). Team seats cost the same per seat.

As a chief of staff: the primary bot is the closest thing on this list to a chief. But it runs xAI's bots, for one account, on xAI's AI. It can't run a Claude Code agent or a Codex agent as part of the same team. See Vestra vs Grok Bot.

Meta Muse

What it is: Meta's personal AI agent. It sends emails, books travel, fills in forms in a browser, negotiates for you, turns long-term goals into action plans and keeps working after you close the app.

Controls: it asks for approval before sensitive actions like sending email or buying something, keeps an audit trail and lets you choose how much access each connected app gets. A separate agent, Sentinel, must approve any outbound internet activity. Muse can't see your passwords or payment methods.

Availability and price: rolling out in the US on iOS, Android and the web. Free for most needs, with subscriptions for heavier use. It runs on Meta's Muse Spark model.

As a chief of staff: it's built for one person's life, not a business's work. It doesn't run other agents. See Vestra vs Meta Muse.

OpenAI Dots

What it is: agents inside ChatGPT that take on ongoing work in the background. You start with one primary dot. Specialist dots, which take on dedicated responsibilities inside a company, began with focused enterprise pilots.

How it works: a dot has its own cloud computer you can open and take over. It researches in the background, reviews your connected apps, suggests ways to help and runs scheduled tasks. It starts with context from ChatGPT memory. It connects to thousands of apps through plugins.

Price: dots come with ChatGPT Pro, from $100 a month, and with Business Premium seats.

As a chief of staff: the primary dot does some chief-like things for one person: it watches, suggests and follows up. But it runs on OpenAI's models for one ChatGPT account, and it can't run agents from other makers. See Vestra vs OpenAI Dot.

Hermes Agent

What it is: a free, open-source agent from Nous Research, under the MIT licence, with memory and skills.

Where it runs: a desktop app for macOS and Windows, Linux via a terminal install, or hosted in the cloud through Nous Portal. You reach it through Telegram, Discord, Slack, WhatsApp, Signal, email or the command line.

How it works: it keeps conversations, memories and skills across sessions, learns your projects and writes its own skills. You bring your own model provider, or use Nous Portal's paid tiers ($20, $100 or $200 a month) for access to 200+ models.

As a chief of staff: it's one person's self-hosted agent. Its subagents are isolated task workers, not a team of agents from different makers. Its own page doesn't describe shared, multi-person features.

The pattern

Read those six again and the pattern is clear:

One person. Each is set up around one account.
One maker's AI, mostly. Hermes lets you pick a provider; the rest run on their maker's models.
One app. Each lives in its own surface.
No team of other makers' agents. None is built to run Claude Code, Codex and an in-house engine as one team.

That's not a flaw in any of them. They're personal agents, and good ones. It just means none of them is the thing a business means when it says "chief of staff".

Which AI chief of staff tools are on the market?

Products sold as an AI chief of staff fall into three groups: personal executive assistants for email and calendar, like alfred_, Martin and Lindy; company tools for managers and leaders, like WorkBoard's Chief of Staff Agent and Sentra; and AI offices, like Vestra, where the chief runs a team of agents. Only the last group runs a staff.

We checked each company's own pages on 10 November 2026. Here's what each says it does.

Personal executive assistants

alfred_. An AI assistant for client-facing teams, with the tagline "Nothing slips." It triages your inbox, sends a morning brief, tracks follow-ups, keeps a "needs you" list and to-dos, texts you before meetings, runs routines and keeps a wiki of your clients. It works with Gmail, Outlook, iCloud and IMAP (up to six accounts), Slack, Microsoft Teams, HubSpot and Salesforce, and with Claude and ChatGPT. Starter is $29.99 a month ($25 billed yearly), Pro is $44.99 ($37.50 yearly), and Team is $49.99 per seat a month ($41.67 yearly) from two seats, with each person's email kept private to them.

Martin. A personal assistant that manages your calendar, checks email and drafts replies, texts and calls people to book reservations, sends reminders, tracks your to-dos, takes notes and reads and sends Slack messages. You reach it by phone call, SMS, WhatsApp, email and Slack. Its docs don't list a price.

Lindy. Lindy calls itself an AI teammate that connects to your tools, learns your company's context and does work for the team, with roles like account management, bookkeeping, customer support, recruiting and inbox and calendar management. It works in Slack, iMessage and email. There's a free start with $50 in credits that last 7 days, and the Team plan is $49.99 a month for 5,000 credits, billed for every active user in the workspace.

What they have in common: they're helpers for a person's own day, mainly inbox, calendar and follow-ups. Some, like alfred_, do chief-of-staff tasks very well, such as a morning brief and a "needs you" list. None is described as running a team of agents from different makers.

Tools for managers and leaders

WorkBoard Chief of Staff Agent. WorkBoard's agent is aimed at managers: "Every manager is a great operator with an AI Chief of Staff." It runs OKR cycles, generates status reports, scorecards and business review pre-reads, builds dashboards, checks Jira or Salesforce for blockers, writes a daily list of priority actions and meeting prep, captures actions and decisions from meetings and flags at-risk work. Pricing is by demo.

Sentra. Sentra describes itself as a "company brain": it connects to email, chat, meetings, documents and 200+ more systems, respects each source's permissions, finds problems that span systems, raises them to the right person with evidence and a drafted fix, and acts once a named person approves. Irreversible actions wait for approval and every step is logged. Pricing is by demo.

What they have in common: these are organisational tools. They're strong at reporting, alignment and spotting risk across systems. They help a leader see the organisation, which is real chief-of-staff work. They aren't built to run a team of AI agents that each hold a whole job.

AI offices

Vestra. One office for your people and your AI agents. Each person has a chief they name, which runs their teammates: agents hired for whole jobs, each on the engine you pick. The chief writes your brief, staffs work, gets it checked and brings you decisions. Plans start at $19 a month. More on how it works below.

How to tell them apart

Ask each product four questions:

Does it run other agents, or is it the only agent?
Can those agents run on different makers' AI?
Do they share one memory you can read and correct?
Is there one set of rules and one record for all of them?

A personal assistant answers no to most of these. That's fine if you want a personal assistant. If you want a chief of staff, the answers should be yes.

How does an AI chief of staff run agents on different engines?

In Vestra, every teammate has two layers: its job, goals, skills, apps, routines, memory and rules on top, and an engine underneath that does the thinking. The chief picks or suggests an engine for each teammate, such as Vestra's own, Claude Code or Codex, and runs them all as one team under the same rules.

This is the part no personal agent can copy, because a personal agent is its engine. Claude Code is Anthropic's AI. Codex is OpenAI's. Grok Bot is xAI's. In an office, the engine is a choice you make per teammate, and the chief is what holds the team together whatever each one runs on.

What an engine is

Vestra's handbook defines the engine as "the part that does the thinking and the working: it reads the task, decides the next step and does it." Everything above it belongs to you and to the teammate:

Its job and goals.
Its skills.
The apps it may use.
Its routines.
What it remembers.
The rules it works under.
Its desk and its history.

Because the engine sits underneath, you can switch it without touching anything above it. The next piece of work simply starts on the new engine, and the job and memory come along.

Which engines are live

Live today in Vestra: Vestra's own engine, which works with the top models, plus Claude Code and Codex. The engine picker also shows Gemini, Grok and Muse, but those aren't live engines yet. Every chief runs on Vestra's own engine.

The chief helps you pick

You don't need to know how engines differ. When the chief hires a teammate, it suggests an engine and says why, in plain words. You can take the suggestion, pick another, or let the chief choose job by job.

In the handbook's example, Leo, who runs engineering, asks his chief Lobo for someone to review code. Lobo suggests Codex: "it's careful with tests. And Mateo builds on Claude Code, so a different engine reads his work with fresh eyes." Leo says yes. Now one teammate builds on Claude Code, another reviews on Codex, and the chief tells everyone when the fix ships.

One set of rules on every engine

Whichever engine a teammate runs on, it works under the same office rules:

The same yeses. Spending over the limit, deleting something and contacting someone new wait for a person.
The same limits. A cap on any single task and on the whole day, and a check before any unusually big task.
Sealed logins. It uses the apps you allow and never sees the passwords.
The same receipts. Every piece of work shows what it cost and who spent it, in one record.

No engine can raise its own limit, approve its own payment or give itself access to an app. One credit is one cent, the same on every engine.

Why this matters for a chief of staff

A human chief of staff doesn't care which university each team member went to. They care that the work fits together. An AI chief of staff should be the same about engines. If your engineers love Claude Code, keep it. If a second opinion from Codex catches bugs, add it. The chief's job is to make them one team, not to make you pick one maker.

What should an AI chief of staff never do without you?

Three things should always wait for a person's yes, however much freedom an AI chief of staff has: spending over your limit, deleting something, and contacting someone new for the first time. Everything else can be tuned agent by agent, from "ask me first" to "act, then tell me".

Vestra builds these three in, and no setting turns them off. They hold no matter how long a teammate has worked for you or how small the thing is.

The three hard lines

Spending over your limit. Anything above a task's cap, the day's budget or a card's monthly limit.
Deleting something. A file, a page, an email, a record. Gone is gone, so it asks.
Contacting someone new. The first email, call or message to anyone you haven't dealt with.

You can add your own, in a sentence: "Always check with me before anything goes to Meridian."

How much each agent does alone

Outside those three, you set how much each agent does on its own, task by task:

Asks first. Before any judgement call, it asks how you want it done.
Suggests. It brings you a proposal and waits for your tap.
Acts, then tells you. For things it has learned well, it does them and shows you what it did.

New teammates start by asking first. As they learn how you like things done, they earn more freedom, the way a new hire does. That only happens when you move the setting. You can also say it in plain words: "Let Keeper handle refunds under $50."

Limits and budgets

In Vestra you set a cap on any single task and a cap on the whole day, up to $100 a day. Before anything unusually large, by default anything over 50 credits, the chief tells you what it will cost and asks. When a cap is reached, work doesn't fail halfway. The agent stops at a sensible point, says what it finished and what's left, and waits.

Sealed logins

Agents need to sign in to real things: your email, your payments, your customer list. In Vestra, every login is sealed. An agent can use it to do what you've allowed, but it never sees the password or key. Every login is scoped to what you allowed and revocable in one tap.

Receipts and one record

Every piece of work comes with a receipt: what it cost and which teammate spent it. And everything that happens is written down in one record, in plain words: who asked for what, what was shared, what was spent and who said yes. If a client asks why they got an email, the answer is a search away.

Why this matters more for a chief than for an assistant

An assistant acts for you in a few apps. A chief runs a team that acts across your whole business. The more agents it runs, the more these lines matter. A chief of staff you can't hold to firm rules is a risk, however capable it is.

How does an AI chief of staff remember what matters?

A good AI chief of staff keeps a memory you can read: who's who, what's been agreed, how you like things done, and what went wrong last time, with a source for every line. In Vestra, that memory is a wiki you can open, check, edit and export, and every correction becomes a lesson the agents follow.

Most AI tools forget your feedback the moment the conversation ends, or keep it somewhere you can't see. That's a problem for a chief of staff, whose whole value is knowing everything you have going on.

What the memory holds

Vestra's handbook lists what goes in:

The people: who does what, and how they like to work.
The customers: every account, every contact, every promise made.
The decisions: what was decided, and why.
How you like things done: your rhythms, rules and taste.
The numbers: revenue, spending and targets.
What went wrong last time, so it doesn't go wrong again.

A source for every line

Every line in the wiki has a source: an email, a call, a record in your CRM, a teammate's work or something a person said. If a line looks wrong, open its source and check. Edit any page, and your edit carries your name as its source.

One correction teaches every agent

Here's the handbook's example. An agent called Keeper was sending a clinic's patients reminder texts at 7am. The clinic's operations lead wrote back asking for nothing before 8, because patients were complaining. Keeper moved the texts, and the lesson went into memory with her email as its source. When another agent later set up reminders for that clinic's two new sites, it scheduled them for 8:30 without anyone telling it.

Your private notebook

Not everything belongs to the team. Each person also has a private notebook for half-formed ideas and notes they're not ready to share. Nothing in it reaches shared memory without their say. When the chief notices something in your notes that would help everyone, it suggests it as a rule, and you decide.

Yours to keep

You can export the memory any time as plain Markdown files, one per page, with the links between pages kept. They open in Obsidian or any notes app that reads Markdown.

What's live

The memory wiki, with a source for every line, is live today. One shared brain that the whole company's agents learn from together is on Vestra's "Then" list, after the team features now being built.

What does a day with an AI chief of staff look like?

A day with an AI chief of staff starts with a two-minute brief, includes a few one-tap decisions and a couple of spoken requests, and ends with work done that you never had to manage. Most of the day, you hear nothing. Here's an ordinary Tuesday, based on the examples in Vestra's handbook.

8:02 am: the brief

Maya opens her phone. Atlas, her chief, has written: "Morning, Maya. Three things got done, one needs you." Overnight, Ledger sent 14 invoices and booked reminders behind the late ones. Keeper answered 19 patient questions. Scout found two rivals who raised their prices this month, checked against their sites.

The one decision: Hunter, the sales teammate, wants to offer a clinic group 15% off a yearly plan. Maya's usual limit is 10%. The decision card shows the yearly value at 15% off, the last two yearly deals (12% and 15%), and Atlas's suggestion. She replies "1". Hunter sends the offer.

8:03 am: two proposals

Below the brief are two proposals. Scout noticed that three lost deals all named the same rival, which just cut its prices, and offers a one-page comparison for this week's sales calls. Keeper noticed that a clinic opens two new sites in May and none of their front-desk staff have logins, and offers to set them up. Maya taps yes on both.

9:00 am: a call from the car

On the drive in, Maya calls Atlas. "Anything I need before Meridian at ten?" Atlas tells her the client wants a pilot in two clinics first, and the notes will be on her screen when she parks. She asks it to move a meeting to Thursday and to have Hunter call a prospect back today. Atlas does both and raises one thing that needs her yes: a $40 ad test. "Yes, go."

Calls and email from the chief, with its own number and address, are live today in Vestra, with phone calls on the Pro plan and up.

11:00 am to 4:00 pm: quiet

For most of the day, nothing reaches Maya. That's the point. Teammates work on their goals. Results get checked. Anything late gets chased. A quiet afternoon means the work got done and none of it needed her.

5:30 pm: a question

Maya asks, in Slack, "What did Scout cost this week, and what did we get for it?" Atlas answers from the receipts, line by line.

Overnight

Teammates keep working. Routines run. A reviewer teammate checks a builder teammate's code. In the morning, it's all in the next brief.

What changes when the whole team has chiefs

In the full version Vestra is building, everyone at Northstar has a chief. When Maya asks whether a new feature is ready to show a client, Atlas asks Leo's chief, Lobo. Leo is asleep in Tokyo. Lobo checks Leo's sharing rules, sees that build status may be shared with anyone at the company, and answers in under a minute: who built it, who checked it, and what's still open. Leo wakes up to one line in his brief, not a question in the middle of the night. Chiefs working together like this is on the Building now list.

Can an AI chief of staff replace a human chief of staff?

Not fully, and it shouldn't try. An AI chief of staff takes over coordination work: briefs, follow-ups, chasing, checking and running agents. A human chief of staff brings judgement, trust between people and the authority to make hard calls. The best setup gives the human the decisions and gives the AI the busywork.

Several vendors in this space say the same. Rhythms, which sells an AI business operating system, argues that "AI doesn't replace the operator; it removes the busywork that was preventing the operator from doing the actual job."

What an AI chief of staff does better

It never sleeps. A question asked at 9 am in Austin is ready at the top of a colleague's morning in Singapore, already explained.
It never forgets. Every ask is tracked until it's answered.
It remembers with sources. Every line in its memory says where it came from.
It scales with agents. Running twenty agents is no harder for it than running two.
It costs less. Plans start in the tens of dollars a month.

What a human chief of staff does better

Judgement in grey areas. Should we tell the board now or next week? That's a human call.
Trust between people. A human chief of staff can have the difficult conversation.
Authority. People follow a person's lead in ways they won't follow an agent's.
Reading the room. Politics, morale and timing are still human strengths.

How they work together

In a company with a human chief of staff, an AI chief takes the load off them. The human sets priorities and makes calls; the AI makes sure the work behind those calls gets done, checked and reported. In a small business with no chief of staff at all, which is most of them, an AI chief gives the founder some of that leverage for the first time.

Who needs an AI chief of staff?

Anyone who has more work than hands and is already using, or about to use, more than one AI agent. That's most founders, small business owners, engineering leads and consultants. If you're the person briefing each agent, carrying work between them and chasing results, you're doing a chief of staff's job, and an AI one can take it over.

Solo founders and consultants

You're the whole company. You answer email, sell, deliver, invoice and plan. An AI chief lets you hand over whole jobs, like getting paid on time or answering inbound leads, and hear back only when something needs you. Vestra's handbook describes a one-person consultancy with a chief and four teammates: one for the inbox, one that drafts proposals within an hour of every call, one that sends invoices and chases late ones. Its tagline for that office is "one person, with the reach of a team of ten."

Small businesses

You have a few people, each stretched. Each person can have their own team of agents, run by a chief. A good first hire is the job that eats the most of your week. Vestra's ready-made skills cover Sales, Bookkeeping, Content, Customer care and Research, and custom skills capture how your business works.

Startups

Your engineers already use Claude Code or Codex. Your founders juggle fundraising, hiring and sales. A chief that can run coding agents as teammates, next to agents for research and sales, keeps everyone's work in one place. The handbook's startup example runs a builder on Claude Code, a reviewer on Codex and a customer-facing agent on Vestra's own engine, with a rule that "no code reaches customers without an engineer's yes."

Engineering leads

You want your team's coding agents to have context about the business, not just the repository. A chief can pass the brief from sales or support straight to the coding teammate, and report back what shipped and what was checked.

Agencies and service firms

You work across many clients. A chief keeps each client's promises, preferences and deadlines in memory, and makes sure nothing goes to a client without a person's yes.

When you don't need one

If you use one AI tool for one job, like a coding agent for coding, you may not need a chief yet. A chief earns its place when you have several agents, or several jobs you want handed over, and the coordination has started to eat your time.

How much does an AI chief of staff cost?

It depends on the type. Personal executive assistants usually cost around $25 to $50 per person a month. Company-wide tools for managers are priced by demo. Personal agents range from free to a few hundred dollars a month per person. Vestra prices per office, from $19 a month for a chief and up to 3 teammates.

Prices from each maker's own pages (checked 10 November 2026)

alfred_: $29.99 a month (Starter), $44.99 (Pro), $49.99 per seat (Team).
Lindy: free start with $50 in credits for 7 days; Team $49.99 a month per active user, 5,000 credits each.
Martin: not listed in its docs.
WorkBoard Chief of Staff Agent: by demo.
Sentra: by demo.
Claude Code: with Claude Pro from $17 a month billed yearly, Max from $100, Team seats $20 a month billed yearly.
Codex: with ChatGPT Plus at $20 a month, Pro from $100, Business $20 per user a month billed yearly.
Grok Bot: with X Premium at $30 a month, Super at $100 or Ultra at $200.
Meta Muse: free for most needs, with subscriptions for heavier use.
OpenAI Dots: with ChatGPT Pro, from $100 a month.
Hermes Agent: free and open source; optional Nous Portal tiers at $20, $100 or $200 a month.

Vestra's plans

Every plan includes your chief and the whole office.

Lite: $19 a month ($17 billed yearly). Up to 3 teammates, 2,000 credits a month, an 8 GB computer for your team. Phone calls start on Pro.
Pro: $97 a month ($87 billed yearly). Up to 20 teammates, 10,000 credits, an 8 GB computer, phone calls.
Max: $197 a month ($177 billed yearly). Up to 50 teammates, 20,000 credits, a 16 GB computer, phone calls.
Ultra: $497 a month ($447 billed yearly). Up to 150 teammates, 50,000 credits, a 32 GB computer, phone calls and a direct line to the Vestra team.
Enterprise: from $4,000 a month, scoped in writing, run in Vestra's cloud, your cloud or on your premises.

One credit is one cent, the same for model time, actions in your apps, the computer, searches and calls. Unused credits roll over. Every new office starts with 1,000 free credits. In India, plans are billed in rupees with GST included.

Per person vs per office

This is the difference that matters most for a team. Most personal agents and assistants charge per person. Five people means five subscriptions, and five separate agents that can't see each other. Vestra charges per office: one plan covers your chief and your teammates, whatever engine each runs on.

What a day costs

The handbook gives an example Tuesday on the Pro plan. The chief wrote the morning brief for 6 credits. A writing teammate drafted three posts for 24. A research teammate found 12 clinics that are hiring for 58. A sales teammate booked two demos for 31. A bookkeeping teammate closed the month's books for 42. A customer-care teammate answered 19 patient questions for 76. The team's computer ran for 2 hours 10 minutes for 156. Total: 393 credits, or $3.93.

How do you set up an AI chief of staff?

Start by naming your chief and telling it about your work in a short conversation. Then hire one teammate for the job that eats most of your week, connect the tools that job needs, set what must always wait for you, and read your first brief the next morning. Add teammates as the first one earns your trust.

Here's the process step by step, as it works in Vestra.

Step 1: Name your chief

A chief is a role, not a product with one name. You choose its name, and you can give it a voice and manner that suit you: brief or chatty, formal or relaxed. You can rename it any time; what it knows stays.

Step 2: Tell it about your work

In a short conversation, tell your chief what your business does, who your customers are and what takes your time. It suggests your first teammates. If you'd rather be walked through it, Vestra offers a setup call.

Step 3: Hire your first teammate

Say the job in a sentence: "I need someone to get us paid on time, and always know our runway." Your chief asks a few questions, then shows you a teammate card with the job, goals, apps, routines and what it must ask you first. Read it, change anything, say yes.

Step 4: Pick an engine, or let the chief pick

Your chief suggests an engine for the job and says why. Take it, choose another, or let the chief decide job by job. You can switch later without losing anything.

Step 5: Connect the tools

More than 2,000 apps connect in a click: Gmail, Slack, HubSpot, Notion, Shopify, Stripe, your calendar and more. Choose which teammates may use each one. Logins stay sealed.

Step 6: Set your lines

The three hard lines are already on. Add your own in plain words: "Never contact anyone at our biggest client without asking me." Set a daily cap. Decide what each teammate may do alone.

Step 7: Choose your brief

Tell the chief when and where you want your morning brief, and how long: "Send me the brief at 8, three lines, and only ask me about money."

Step 8: Correct, and widen the freedom

For the first week, your teammate asks before judgement calls. Answer, and correct what's wrong. Each correction becomes a lesson. After a few weeks, move the setting to "suggests". Later, for what it does well, to "acts, then tells you".

Step 9: Add the next teammate

Once the first job runs without you, hand over the next. Up to 3 teammates on Lite, 20 on Pro, 50 on Max and 150 on Ultra.

What should you ask before choosing an AI chief of staff?

Ask whether it runs other agents, which AI those agents can run on, whether its memory has sources you can check, what always waits for your yes, how spending is capped, how pricing works for a team, and what's live today versus promised. A clear answer to each tells you whether you're buying a chief or an assistant.

The questions

Does it run other agents? If the answer is no, it's an assistant. Decide whether that's what you need.
Which AI can those agents run on? One maker's models, a model picker, or a choice of engines per agent, including ones your team already uses?
Can it hold whole jobs, not just tasks? Goals held over months, routines on a schedule or when something happens, and one-off asks.
How does it check work? Tests, a second agent, the source, or a person, and does the proof come with the result?
What always waits for my yes? Look for spending over a limit, deleting and first contact, at a minimum.
How is spending capped? Per task, per day, and per agent card?
Can I see its memory? Can you read what it knows, see where each fact came from, fix it and export it?
How are logins handled? Does the agent ever see your passwords? Can you take access back in one tap?
Is there a record? Can you see who asked for what, what was shared, what was spent and who said yes?
How does it reach me? App, Slack, Telegram, email, phone? Which channels are live today?
How does pricing work for a team? Per person, per seat, per workspace or per office? Do unused credits roll over?
What's live and what's coming? Ask for the vendor's own list. Buy for what works today.

Red flags

It pings you all day. A chief should cut your interruptions, not add to them.
It says "done" without proof. You'll end up checking everything yourself.
You can't see what it remembers. You won't know why it does what it does.
There's no hard line on money. Capable agents with no spending cap are a risk.
It only works with one maker's AI. You'll have to drop the agents your team already likes.

What are the risks of an AI chief of staff, and how do you manage them?

The main risks are an agent acting beyond what you meant, spending more than you expected, getting a fact wrong, being misled by an outside message, and sharing something private. You manage them with hard approval lines, caps, checks with proof, sourced memory, sealed logins, private-by-default settings and a record you can read.

Acting beyond what you meant

Risk: an agent takes a step you didn't intend. Fix: start every teammate on "asks first" and widen its freedom only when you've seen its work. Keep the three hard lines on.

Spending

Risk: a task runs longer, or costs more, than you thought. Fix: cap each task and each day, and require a cost check before unusually big tasks. Read the receipts.

Getting facts wrong

Risk: an agent states something false with confidence. Fix: require a source for every fact, and make "done" mean checked. When something's wrong, correct it once; the lesson should reach every agent.

Being misled by outside messages

Risk: an email or another company's agent tells your agent to do something, like pay an invoice or change a permission. This is often called prompt injection. Fix: Vestra's handbook sets a firm rule for messages from other companies' agents: they can inform, but never command. A note from outside can't approve a payment, change a permission or create a rule your agents will follow. Your chief reads it as information, checks it, and brings any decision to the right person as a question. (Working with clients chief to chief, through what Vestra calls the front door, is on its "Then" list.)

Sharing something private

Risk: a chief tells a colleague something you didn't want shared. Fix: private by default. Your private notebook and your chats with your chief are yours alone. In the team version Vestra is building, each person sets sharing rules in a sentence each, and their chief answers, holds or says no by those rules.

Lock-in

Risk: everything your team knows ends up trapped in one vendor. Fix: choose a tool that lets you export memory in a plain format, and that lets you switch engines without losing a teammate's job or memory.

What's live in Vestra today, and what's being built?

Live today: a chief and a team of teammates for one person, with routines and a morning brief, a cloud computer, calls and email from the chief, connected tools, a memory wiki, any engine per teammate (Vestra's own, Claude Code or Codex), and limits, approvals and receipts. Being built now: a chief for every person, working together.

Vestra keeps a public board called Where We Stand. Here's what it says, in the order Vestra is building.

Live today, for one person

A chief and a team of teammates. Hire, shape and run teammates, with routines and a morning brief.
A cloud computer. Watch your team work in a real browser, and take over any time.
Calls and email from your chief. Its own number and address, with phone calls from Pro up.
Connected business tools. Email, calendar, CRM, books and chat, connected in a click.
A memory that learns. A wiki of what your team knows, with a source for every line.
Any engine, per teammate. Vestra's own, Claude Code or Codex, picked by you or your chief.
Limits, approvals and receipts. Daily and per-task caps, your yes on money, every credit itemised.
Your chief in the app, Slack, Telegram and email. On the web and the desktop app for Mac.

Building now: the network

A chief for every person, named by them.
Chiefs work together: ask, answer, hold or say no, and hand work across.
The office you can walk around: every desk, every space and what's happening, live.
Meetings in the office, with agents keeping the notes.
Spaces: one goal, shared context, and decisions for the right person.
Access control: private, company or network, for every teammate, space and app.
Every hand-off checked and written down.
WhatsApp and iMessage, to reach your chief from your phone's messaging apps.

Then, one at a time

A phone and an inbox for every agent, not only the chief.
The vault: a card and logins for every agent, sealed, with limits each person sets.
Shared apps: software built in a sentence, used by people and agents together.
The shared brain: one brain for the whole team, learning from every correction.
The front door: clients and partners working with you, chief to chief.

Vestra notes that some live features are still rolling out to every account.

What this means if you're choosing today

If you want one person's AI chief of staff that runs a team of agents on different engines, with firm rules and receipts, that's live. If you want every person on your team to have a chief, with chiefs handing work to each other, that's the part being built now. To see how this compares with a single coding agent, read Vestra vs Claude Code. For the bigger picture of how agents work as one team, read What is an AI office?, and for a wider comparison, The best AI agents for business.

Glossary

AI chief of staff: an AI agent that runs your other AI agents, speaks for you within your rules and brings you only real decisions.
AI executive assistant: an AI that manages one person's inbox, calendar, meetings and follow-ups. It does tasks itself rather than running other agents.
AI agent: an AI that can act on its own: browse, write, run code, send email or make calls.
Personal agent: an agent built for one person and one account, usually on one maker's AI.
AI office: one place where people and AI agents work together as one team, with a chief for each person, shared memory and one set of rules.
Chief: in Vestra, the agent that runs your other agents, speaks for you and answers to you. You name it.
Teammate: in Vestra, an agent hired for a whole job with goals, on the engine you pick.
Engine: the part of an agent that does the thinking and working. In Vestra, Vestra's own, Claude Code or Codex today.
Morning brief: the chief's daily note: what got done, what needs you, what it noticed and what you're waiting on.
Teammate card: the written job for a new teammate: goals, apps, routines and what it must ask first.
Routine: work an agent repeats on a schedule or when something happens.
Skill: a written way of doing a kind of work that a teammate can pick up.
Memory wiki: the readable, editable memory where every line has a source.
Hard lines: actions that always wait for a person: spending over a limit, deleting something and contacting someone new.
Receipt: a record of what a piece of work cost, who did it and who said yes.
Credit: Vestra's unit of usage. One credit is one cent.
Sealed login: a login an agent can use without ever seeing the password.
Prompt injection: an attempt to steer an AI agent with instructions hidden in a message, email or web page.

FAQ

What is an AI chief of staff?

An AI chief of staff is an AI agent that runs your other AI agents for you. It hands out work, checks results, chases what's late, keeps one memory and brings you only the decisions that need you. It answers to one person.

What does an AI chief of staff actually do day to day?

It writes you a morning brief, staffs new work with the right agent, makes sure results are checked before they reach you, tracks every ask until it's answered and brings you decisions with the facts and a suggestion, ready to answer in one tap.

What's the difference between an AI chief of staff and an AI assistant?

An AI assistant does tasks for you. An AI chief of staff runs the agents that do the tasks. If a product doesn't run other agents, it's an assistant, whatever it's called.

What's the difference between an AI chief of staff and an AI executive assistant?

An AI executive assistant manages your own inbox, calendar and follow-ups. An AI chief of staff manages a team of agents doing whole jobs for your business, and can also handle executive-assistant tasks along the way.

Can ChatGPT be an AI chief of staff?

ChatGPT's dots can work in the background for one ChatGPT account, and the primary dot can watch, suggest and follow up. But they run on OpenAI's models for one person, and they can't run agents from other makers, like Claude Code, as one team.

Is Claude Code an AI chief of staff?

No. Claude Code is Anthropic's coding agent for developers. It's excellent in a codebase, but it doesn't run other agents. In Vestra, Claude Code can be the engine of a teammate that your chief runs alongside the rest of your team.

Can an AI chief of staff replace a human chief of staff?

Not fully. It takes over coordination: briefs, follow-ups, chasing and checking. A human chief of staff still brings judgement, trust between people and authority. Many companies will use both, with the human making the calls and the AI doing the busywork.

Which AI chief of staff is best for a small business?

One that runs a team of agents for whole jobs, keeps a memory you can check, has firm spending and approval rules, and prices per business rather than per person. Vestra is built for this, from $19 a month. If you only need inbox help, an AI executive assistant may be enough.

How much does an AI chief of staff cost?

Personal executive assistants are usually $25 to $50 per person a month. Tools for managers are often priced by demo. Vestra starts at $19 a month for a chief and up to 3 teammates, with Pro at $97 for up to 20 teammates and phone calls.

Is an AI chief of staff safe?

It can be, if it has firm lines. Look for: spending over your limit, deleting something and contacting someone new always waiting for your yes; caps per task and per day; sealed logins; receipts; and a record you can read.

Can an AI chief of staff make phone calls and send email?

Some can. In Vestra, the chief has its own phone number and email address, and that's live today, with phone calls on Pro and up. A phone and inbox for every teammate, not only the chief, is planned for later.

Can my AI chief of staff work with Claude Code and Codex?

In Vestra, yes. Each teammate runs on the engine you pick: Vestra's own, Claude Code or Codex today. Your chief runs them as one team under the same rules, and can, for example, have one teammate build on Claude Code while another reviews on Codex.

Can everyone on my team have an AI chief of staff?

That's the part Vestra is building now. Today, a chief runs one person's team of agents. Next, every person gets a chief they name, and chiefs ask, answer and hand off work to each other.

How long does it take to set up an AI chief of staff?

In Vestra, work can start the same day. You name your chief, tell it about your work, hire a first teammate, connect a few tools and set your lines. You can also book a call to set the office up with the Vestra team.

Does an AI chief of staff need to know my passwords?

It shouldn't. In Vestra, logins are sealed: agents use them to do what you allowed, but never see the password or key, and you can revoke any login in one tap.

What should I hand to an AI chief of staff first?

The job that eats the most of your week. Common first hires are getting paid on time, answering inbound customer questions, following up with leads and keeping a weekly view of the market.