We do not present roadmap behavior as live capability.
Before someone touches your digital presence, know what they can touch.
Six questions an owner should ask anyone — an agency, a platform, an AI — before handing over access. Our answers, grounded in how the product actually works today, including where it is not finished.
Who owns my website, domain, accounts and data?
Your domain is registered to you. Your Google Business Profile, analytics and social accounts are yours; when we operate them, we do it under access you granted and can revoke.
The website is built as a standalone codebase for your business — one repository, one deployment — not a page inside a shared platform. That isolation is deliberate: a mistake on another client's site cannot touch yours, and yours can be handed over whole.
One repository and one deployment per client site. Your domain, your accounts, your revocable access.
What can Lehvel access?
Access is granted per system — the site, the profile, the analytics property — and scoped to what the work needs. We do not ask for a master password to everything.
Internally, our own systems follow the same rule: credentials are never written into code, a model never chooses which client environment it writes to, and access to a client's live site resolves from your identity, not from a guess.
Per-system, revocable grants. Trusted-target resolution: customer identity → the one site it may touch.
What can Lehvel change, and what needs my approval?
Updating hours, answering a routine review, publishing an approved offer: inside the permissions you set, an Agent may act. Anything that moves money, changes a domain, deletes content or changes the terms of a relationship is a stop-and-ask, every time.
Automated changes fail closed. If the target is ambiguous or the authority is unclear, nothing happens and you hear about it — that is a feature we test on purpose.
A permission boundary written down per Agent. Fail-closed mutations. Money, domains, deletions and terms always require a person.
How do I know what changed?
Our execution contract says a commit is not completion and an HTTP 200 is not completion: the requested change has to be observed on the live page before the system calls it done, and the report you get names what was changed and where.
That read-back is live for our own operations and is the acceptance gate for customer-facing execution, which is still being finished. Where the product does not yet produce the evidence, we say so rather than showing you a green checkmark.
Verify-before-report: observed on the live site, then a plain-language record of the change.
What happens if I leave?
Because each site is its own codebase and deployment, it can be transferred to you or to whoever you choose. Your domain and your accounts were always yours.
Data we hold about your business — the record, notes, requests — is yours to export on request. We do not publish a formal portability guarantee beyond what the product currently supports, and we will not imply one.
Standalone site, transferable. Accounts never held in our name. Record export on request.
How is AI used?
AI drafts, structures, executes routine steps, observes and reports. It does not set its own permissions, choose its own targets, spend money, or declare something done without verification. A named person is responsible for every Agent.
We build with third-party models and services — OpenAI, Anthropic, Google, Vercel, Supabase, GitHub — under their terms. Using a vendor is not a partnership or an endorsement, and we say which vendor does what when it matters to you.
TALK → STRUCTURE → RECOMMEND → ACT, with a person responsible and consequential actions gated.
Tell us what is true about your business.
A few questions, in your own words — type them or talk. A person reads what you send and follows up. Nothing is published and nothing is sent on your behalf.