Open-source AI chief of staff tools, and the slice each covers

The slice each tool sells

An AI chief of staff is a job with a fixed set of recurring outputs: the weekly report, the standup, the meeting follow-ups, the investor update, the inbox. What an AI chief of staff is covers the role in full. What matters when you buy a tool is narrower: which of those outputs the tool produces, and which ones come back to you.

alfred_ is the clearest case. It is an executive assistant for email and calendar, and its routines are metered in credits. Cleaning up a calendar costs about 14 credits a run, planning a week about 55 (alfred_ pricing). That is a fair price for the work it does. It is also a price for one slice. The calendar gets cleaned, and the weekly report, the standup and the investor update stay on your desk.

Vybe takes the Slack slice. It lives in your channels, and its credits run at $1 = 100 (Vybe pricing). WorkBoard takes the OKR slice: its AI chief of staff agent drafts OKRs from past outcomes, checks KR quality and finds cross-org alignment (WorkBoard). Both are good at what they name. Neither is the layer the rest of the job runs on.

What Kortix covers that a slice does not

Kortix is the open-source CompanyOS: your agents, their skills, your company memory, and every connector in one git repo you own, with the agents working on real cloud computers. The difference from a slice is where the job lives. It is written down as an agent in a file you can read, and it runs on a schedule instead of waiting for you to ask.

Here is a real one, from Kortix's own use-case library:

A Monday cron queries the metrics from our Postgres database, writes the weeklyreport with commentary on what moved and why, and posts it to Slack. Access tothe database is read-only.

That is the weekly report as a scheduled job rather than a prompt you retype every Monday. The same shape covers the standup, meeting notes, the investor update and inbox triage. Each one is a page in the library with the agent's own description.

Three things make that work without a per-seat subscription. Connectors: 3,000+ apps in a click, plus MCP, OpenAPI, Postman, GraphQL and raw HTTP. Hosting: run it on Kortix Cloud, in your VPC, or on your own on-prem network, and self-host is free. Models: any model provider with your own keys, or the ChatGPT plan you already pay for.

The price follows the same logic. Free is $0 with 500 credits per month, every feature and 1 project, on the user's own provider keys or ChatGPT subscription. What an AI chief of staff costs works through the paid tiers and what a credit buys.

The two tools outside the table

MyZone AI presents its chief of staff as a persona called Helen, one agent in the Ai1 platform: a daily brief, meeting prep and debrief, follow-up tracking, decision packets, and coordination of the other Ai1 agents. Anything external or binding waits for approval (MyZone AI). It is the closest thing here to the whole job, and it is still one agent inside one vendor's platform.

Zamp sells "self-learning digital employees", agents that own a process end to end and capture the tacit knowledge around it (Zamp). That is a different purchase from a cadence. You buy a process.

How to choose

Start from the output you cannot afford to miss. If it is the calendar, alfred_ is built for it. If it is the quarterly OKR cycle, WorkBoard. If your team lives in Slack and wants an assistant in the channel, Vybe. If the cadence itself is the problem, the tool has to be an operating layer you own. You want the weekly report, the standup and the investor update to run whether or not you remember them, and that is the case for an open-source AI chief of staff.