Spend more time on the work people choose you for.

Before you buy

• Which plan should I choose?

  Explorer ($35/mo): learn with one starter agent and 800 monthly credits.

  Builder ($99/mo): assign recurring work with up to 3 hosted agents, broader model choice, personal group topics and 4,000 monthly credits.

  Innovator ($500/mo): support customer inquiries, quoting, operations and online-store work with up to 5 AI Hires and 30,000 monthly credits.

  Coming soon: one-click automation setup for Receptionist / Speed-to-Lead, Internal Operations / Quoting, and Online Store Agents. Innovator and Agency can choose any or all three within their total plan limits. Connections, business rules, testing and approval are still required.

  Every paid plan includes hosted runtime maintenance, the full course library, AI tutor, AI Studio and builder community.

• What happens after I buy?

  All three plans are guided self-service. After checkout, use AI Studio setup for deployment and Telegram pairing, then connect the tools your job needs. You set the rules and approval boundaries. Open guided setup

  Once an agent is active, the optional six-stage interview lets you practice preferences, recall, a bounded job and result review. It does not automatically write your Agent Brain or memory; completion is based on your report, not runtime verification. Interview messages use AI credits; hosted-agent work is metered separately.

• Do I own the agent and my data?

  Yes. You own your agent and data. We do not sell your data or use it to train our own models. Hosting and AI providers and connected services process relevant information to deliver your requests. Provider terms and settings can differ, including retention and training practices; see our privacy policy.

  Export your files, memory, skills and settings, or request deletion. Restore exports on a compatible runtime and reconnect tools and credentials; an export is not a full VM backup. Saved notes provide continuity, not perfect recall—verify important details.

• How do credits and top-ups work?

  Credits power agent replies, research and AI Studio. Usage varies by model, context and task; heavy media generation uses more. AI Studio and Success Agent actions use their own metering.

  Need more usage, not more agents? Stay on Builder and top up. Packs stack on your plan and never expire. Choose Innovator for business workflows with up to five agents and a larger included credit budget. The upcoming one-click automation setup is for Innovator and Agency. See model estimates and top-up prices

• What happens if I run out of credits?

  Agent execution is limited by credits and service availability, even while hosting continues. Add a credit pack or wait for your monthly credits to reset. Credit exhaustion does not delete your saved workspace or memory.

  Message replay and automatic task resume are not guaranteed. After replenishing credits, check agent status and confirm unfinished work before retrying.

• Can I cancel, switch plans, or get a refund?

  Cancel anytime in your Whop portal.

  Switch plans anytime: upgrades take effect immediately when you activate the higher plan; downgrades take effect at your next billing cycle.

  90-day money-back guarantee on subscription fees. Credit packs are non-refundable but never expire. For a refund, email info@cypher.camp.

• What is an AI Work Shift?

  An AI work shift is a planned bundle of work for your AI Hire—such as preparing sales conversations, organizing customer notes, and drafting follow-ups. You define the job and what good work looks like.

  For cost planning, we currently budget 500,000 tokens per AI work shift: 350,000 new input/context-tool tokens, 50,000 output/reasoning tokens, and 100,000 cache-read tokens. This is a starting assumption, not a measured human-workday equivalent. Idle wall-clock time does not consume tokens. Voice and connected tools may incur additional usage.

  Planning estimate. Actual work delivered depends on the task, model, and review.

  These are token-cost comparisons, not measured job costs or human workdays. The illustrative range is 0.5–1.5 times the reference estimate, not an observed range. Actual usage depends on context, retries, tools, and model choice.