What an AI chief of staff costs, and what open source changes

What the sticker price buys

Three of the six tools here publish a price. alfred_ starts at $29.99 a month, billed annually. Starter carries 2,000 credits, up to 6 email and calendar accounts, Slack, Teams, HubSpot and Salesforce, plus routines and workflows. Pro is the same product set up for you by their team, and Team gives every person their own alfred_ with a management console and one bill.

Vybe is the cheaper way in. The free plan is $0 with $10 in credits a month, which is 1,000 credits and no card. Starter is $20 for 2,000 credits, Pro $50 for 5,000, Team $100 for 10,000. It lives in your Slack channels, and Enterprise lets you bring your own OpenAI, Anthropic or Google keys.

Kortix bills the same way, in credits, with one difference that matters here: the free tier is a real tier. Free is $0 with 500 credits per month, every feature and 1 project, on the user's own provider keys or ChatGPT subscription. The paid plan is credits: 25 tiers from $50 a month (5,000 credits) to $50,000 a month (5,000,000 credits), at 100 credits per dollar. Members are unlimited and free, and the account shares one credit pool.

WorkBoard, Zamp and MyZone AI publish no price on their own pages. WorkBoard sells its AI Chief of Staff Agent inside WorkBoardAI, an OKR platform. The agent runs the OKR cycle: drafting OKRs from past outcomes and upline priorities, checking key-result quality, finding cross-org alignment. Zamp sells agents that own a process end to end and capture the tacit knowledge around it. MyZone's chief of staff is one agent in its Ai1 platform. It produces a daily brief, preps and debriefs meetings, tracks follow-ups and builds decision packets, and anything external waits for approval.

Prices are as published in November 2026. The full field, with what each tool covers, is in the AI chief of staff tools compared.

A credit is not a credit

All three priced plans bill in credits, and the credits are not the same unit. Vybe and Kortix both set 100 credits at $1. alfred_ publishes no dollar rate for its credits, only a monthly allowance, so its per-run cost cannot be read off the pricing page.

alfred_   clean up my calendar   about 14 credits a runalfred_   plan my week           about 55 credits a runalfred_   included               2,000 credits a monthVybe      $1                     = 100 creditsKortix    100 credits            = $1

The run costs are the useful part. A calendar cleanup at alfred_ is about 14 credits and a week plan about 55, both from their pricing page. On a 2,000-credit month that is roughly 36 week plans, or about 140 calendar cleanups. Chat, triage, follow-ups and the morning brief do not spend credits, so the allowance covers routines and workflows only.

That is the first thing to fix before you compare $29.99 with $50. A plan that publishes a run cost lets you count the runs you need. A plan that publishes only a monthly allowance leaves the arithmetic to you.

What the plan does not cover

The pricing pages answer a narrower question than a chief of staff answers. alfred_ covers email, calendar and CRM. Vybe covers Slack. WorkBoard covers the OKR cycle. Each is a slice, and the slice decides which part of the job you still do yourself.

The work outside all three is the recurring reporting a chief of staff owns: the weekly report, the standup summary, the meeting follow-ups, the investor update, the QBR deck. None of it is email triage, and none of it is an OKR. It is a document built from data in three or four systems, on a schedule, in a format your team already reads. What an AI chief of staff actually does is mostly this, and it is the part a plan priced per inbox never reaches.

The investor-update agent is the clearest case. It reads the core metrics from the database, compares them to last month and to plan, drafts the update in the usual format, and stops for a founder to edit and send. No email tool drafts that, because the numbers live in a database, not an inbox. The weekly report, the standup and the QBR prep have the same shape, and each one is a job you can write down and hand to an agent.

That is the second number. A $29.99 plan that covers email and calendar leaves the investor update, the weekly report and the QBR deck on your desk, and those are hours a month you either do yourself or pay someone for.

The open-source option, priced

Kortix is the open-source option, and the cost question changes shape there. The agents, their skills, the company memory and the connectors are files in one git repo you own. A job you write once keeps running: a schedule, a webhook or an event in another app starts a session with nobody at the keyboard.

Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free. Any model provider with your own keys. Or the ChatGPT plan you already pay for.

A hosted plan charges for the runs its product supports. A repo you own runs the jobs you wrote, on infrastructure you choose, and the recurring cost is the credits you spend or the model bill you already pay. The longer version of that argument is in the case for an open-source AI chief of staff.

How to price it for your own week

Start with the jobs, not the plans. Write down the five recurring jobs you do every week: the report, the standup, the follow-ups, the update to your board or investors, the prep for the next review. For each one, note where the data lives and who reads the output.

Then price the plan against that list. A plan that covers two of the five is not cheaper than one that covers four, whatever the sticker says. The number to compare is the monthly plan plus the hours the remaining jobs take, and only you can put a rate on the second part.

The last check is the exit. Ask what happens to the agents, the prompts and the memory if you stop paying. On a hosted plan they live in the vendor's product. In a repo you own they are still there.