FAQ
Questions, answered plainly.
About membership, the loop, the products, and your data. If yours is not here, ask it in the community or on the contact page.
Membership
The stewards, the AI development team, verification and the testing cohort, and the community: the queue, the weekly changelog video, the monthly call and the scorecard. It does not pay for AI usage. The apps run on your own keys, and nothing is metered inside them.
No. Tiers differ in involvement (early access, the testing cohort, office hours, an advisory slot), never in whose requests get built. No tier moves a request up the queue.
Membership covers all four. If a single-product plan would change your decision, post that as a problem in the queue; it is a fair request and it will get a public decision.
A limited tier for the founding cohort of twenty to fifty practitioners who already use at least one product and run the loop in public first. Lifetime recognition, a named advisory slot on one product, first call on new products.
A change built for one organisation alone, priced as a project with its own scope and timeline, and kept visibly separate from the community queue. It never jumps the line, and community requests never wait behind it.
The loop
“I lose an hour every Monday reconciling replies across three inboxes” is a problem. “Add a unified inbox” is a guess at a solution. Problems are evidence; the steward and the AI team work out the solution against the charter.
First response within one business day. A public decision within ten business days, with a reason. Build started within the published sprint. No release without a passing check. A weekly release window. A weekly video and a monthly call. These are the only promises.
A practitioner steward, one per product, who uses it daily and holds its charter. The AI development team does the work between decision and release. A human release gate confirms the checks passed, the change matches the decision, and the help article is ready.
It stays in the queue with the reason and the charter line it would break, so the queue records the product’s direction, not just its backlog.
Because it makes the cost of building small enough that judgment becomes the scarce resource. Safety is evidence, not argument: published test reports, the verification log on every change, regression, layout, accessibility and data-consistency checks, releases behind flags, one-action rollback, and a human release gate.
The build log (what was read, written and run), the check results, the testers’ notes, the flag, the changelog entry, the help article and the week’s video.
The products
LevelUp is live at levelupnow.tools. Publish is live. Producer is version 0.1.0 for Windows. Velocity is in early access until the top items from its August test report ship; the report and the list are on its page.
They hand off by workflow: Producer makes the video, Publish distributes it, Velocity works the leads, LevelUp holds the plan. Built integrations between them arrive only when the changelog says so.
Not yet. Producer is a Windows x64 desktop app in version 0.1.0. A Mac build is the kind of request the queue exists for.
Yes. Several companies on one install, each a sealed workspace with its own keys, voice, accounts, schedule and content. Co-branded with the client’s logo and accent colour, never white-labelled.
Only through a dial you set to Auto, per feature, and only as an admin. Daily caps, warmup, suppressions, unsubscribe handling and LinkedIn limits apply in every mode. No mode removes them.
Data and keys
Yes. LevelUp and Velocity take OpenAI, Anthropic or Gemini keys. Publish takes Anthropic or Gemini for writing and OpenAI or Gemini for images, plus a Tavily or Serper key for live search. Producer uses your Claude subscription through Claude Code and never holds a key.
In your workspace, in every product. LevelUp keys never leave the server and Money is private even from admins. Velocity contact data stays in the workspace it was enriched in. Publish storage is served only to signed-in users. Producer keeps recordings, narration and renders on your machine; Claude sees the transcript and selected frames.
Only with consent. Telemetry is opt-in per workspace, aggregate only, with a plain list of what is collected, and the findings are published to the community in the same form the stewards see them.
[Cancellation and data-export policy, one paragraph, to be written.] The principle is fixed: your data and your keys are yours, before and after.
Still have a question?
Ask it on the contact page, or post it in the community if it is really a problem in disguise.
