FAQ

No. Assistants help you interact with AI. Genie keeps what you tell it and puts that record to work — accepting bounded work into a managed production system and returning a reviewed deliverable.

No. Memory tools store what you file. Genie captures what you say in ordinary use and actively chooses the context for your agents — nothing to file, nothing to organize.

Any. The model is interchangeable; your record is the asset. Same context, whatever you connect.

Where you keep it: local-first, or our beta — separated by organization, never sold, never shared across organizations. Details on the privacy page.

No. Genie is your memory — a knowledge graph that works as the context layer for your agents. Tasks are simply how you assign work with context to our cloud team, without managing an AI yourself.

One task is one assignment from your record: a bounded piece of work our cloud team executes with your context, approved by you first. A broad request may require multiple tasks; Genie shows the count before you approve it.

Not until Genie has enough context, shows you its understanding and approach, and receives approval for the scope and cost.

Edit the alignment receipt or answer the specific clarification. The task returns for alignment; it does not enter production.

The delivery includes review status, acceptance checks, relevant evidence, and a record of what Genie delivered.

Yes. Successful recurring work is saved with its context, rules, deliverable format, and acceptance contract — so the next run starts with everything Genie already learned.

Genie pauses and requests approval before adding tasks or performing materially different work.

Genie is the production layer underneath the client relationship. The agency retains strategy, judgment, sales, client management, and trust.

Individual access is free. Production work is quoted per task on the alignment receipt, and you approve it before execution begins. Retries and internal steps never become surprise line items. There is no public rate card.

Website updates, documents and written deliverables, research and synthesis, spreadsheets and structured files, analysis and reporting, and bounded multi-step workflows with clear outputs.

Unlimited undefined projects for one task, unsupervised high-risk legal, medical, or financial decisions, work requiring missing credentials or unauthorized access, and deployment or publication without explicit authority.