Skip to content
Home / FAQ
05 · FAQ

Straight answers.

The questions every operator asks before they sign. Still wondering something? Ask us directly →

01 · The big six

What everyone wants to know.

Tap a question to expand.

Yes. Every build is custom and the IP is yours — code, agents, prompts, evals, the lot. We're not building a platform to lock you into; we're building your system and handing you the keys. Cancel the monthly retainer with 30 days notice and the system keeps running.
About four weeks from kickoff. Week 1 is discovery, weeks 2–3 are build, week 4 is install + tune. Most teams have their first workflow firing on schedule before the month is out.
Your data stays on your hardware. We design for least-privilege access, audit logging and human-in-the-loop on anything high-stakes, and every rollout ships with a kill switch. No cloud subscription chatbot involved — the model runs where your business runs.
No. The whole point is that AI lives inside the systems you already run — CRM, ERP, helpdesk, inbox, spreadsheets. We integrate, we don't migrate. Adoption is zero-friction because nothing new has to be learned.
In hours and dollars. We baseline before we build, commit to a target, and report against it monthly. If the numbers aren't there, the work isn't done. Every Atlas Control client gets a Friday impact report — what fired this week, how many hours back, what's queued.
If your team spends real hours on repetitive work, there's almost certainly something worth automating. The Pilot tier exists precisely so you can prove it on three workflows before committing further — credit-back if you scale within 90 days.

02 · Implementation

The how-it-actually-works questions.

Usually a Mac Mini (M4, 16GB) lives in your office or warehouse. About the size of a paperback. Runs the agent stack, talks to Anthropic Claude (or local Llama for cheap workflows), connects out to your tools via API. Costs ~$1,200 one-time. We size it during discovery.
Scheduled workflows queue and run when connectivity returns. Real-time workflows (e.g. inbox replies) pause until back online. We design every workflow with a fallback path — "if Claude is unreachable, do X" — so nothing leaves customers hanging.
Yes — that's what the monthly retainer is for. Either swap a workflow out of your tier limit, or upgrade tiers. Most clients add 1–2 workflows per month for the first quarter, then stabilise.
You keep everything. Code, prompts, evals, docs — all in a git repo we hand you on day one. The monthly retainer pays for refinement, monitoring, and engineering support. Cancel with 30 days notice and the system keeps running on your hardware indefinitely.
Yes — the Pilot tier is essentially a one-off project with optional retainer attached. We can also quote pure consulting (audit, roadmap, strategy) without a build. Talk to us about what you need.
You bring your own AI provider account (Anthropic, OpenAI, or both). We size it for you during discovery. A typical 7-workflow client runs $50–$200/month in API costs, billed direct to you. Transparent. No markup.
They will if you don't change their tools. Atlas Control integrates AI into the systems your team already uses — email, Slack, your dashboard, WhatsApp. Nobody has to learn a new app. The AI does the work; your team gets the output where they're already looking.
A great VA is $3k–$8k/month plus management overhead, holidays, sick days, training, churn. Atlas does the same recurring work for $200–$800/mo, runs 24/7, never quits, and doesn't need onboarding. Use a VA for judgment calls and relationship work. Use Atlas for the recurring stuff.
ChatGPT is a chat window — you copy and paste. Atlas Control integrates AI directly into your workflows: it fires on schedule, in your tools, against your business rules. Different category entirely. The AI does the work, you check the output.

Still got questions?

Book a 30-minute discovery call. We'll answer them and map your workflows while we're at it.