5-12 days is not a guess. It's not aspirational. It's what we've shipped every build to, and it's possible because we don't write code until we've done the hard work first.
Most agencies treat discovery as a box to tick before they start billing. We treat it as the majority of the intellectual work. The build itself — the n8n workflows, the API connections, the voice agent configuration — is mechanical. It's the understanding of your business that's hard to get right.
Here's exactly how 5-12 days breaks down.
We map everything. Your inbound call flows. Your roster rules. Your brand voice. Your approval preferences. Who the VIP callers are. What constitutes a complaint versus a normal query. What the edge cases look like. We ask a lot of questions that feel tangential — they're not.
Output: a spec document. A full written description of what we're building, how it routes, what it does in every scenario we could imagine. You read it. You correct it. We don't write a line of code until this is signed off.
This is where the actual construction happens. n8n workflows get built, API credentials get connected, the voice layer gets configured. It looks unimpressive from the outside — a lot of JSON, a lot of test calls, a lot of incremental checks.
Internal tests run from day five. We push synthetic inputs through the system: fake calls, fake emails, edge cases from the spec document. Most things work. Some things don't. We fix them before you see them.
No client calls during the build phase. We're heads-down.
This is where the perceived value shows up. Day 10: you test it as a customer. You try to book, cancel, ask an odd question, probe the escalation path. You tell us what felt wrong — tone, routing, a phrase it used that sounded off.
Day 11: we fix everything you flagged. Tune the prompts. Adjust the escalation thresholds. Rerun the edge cases. Day 12: handoff call with your team — how to update the knowledge base, how to read the Telegram briefings, what the monitoring looks like.
Day 12: go live.
Why we don't rush discovery
We've been asked to compress this. Clients who are excited want to go faster. We generally don't — not because we can't build faster, but because the discovery phase can't be skipped.
The build is fast. Understanding your business well enough to build it right takes time.
If we skip or compress discovery, we build something technically correct but behaviourally wrong. The voice agent doesn't know your VIP's name. The approval threshold is set wrong and floods you with tier-2 notifications you don't want. The brand voice is slightly off in a way that's hard to articulate but immediately obvious to your customers.
A proper discovery phase produces a system that works as built. Rushed discovery produces a system that requires weeks of post-launch fixes. The calendar math doesn't actually improve.
What can extend beyond 12 days
Two things legitimately push a build past 12 days. The first is scope complexity — a build that involves more than two or three system integrations, or that needs significant custom logic, needs more build time in phase two. We scope this in the discovery document and tell you before we start.
The second is slow client-side feedback. The test-and-tune phase requires you to test it, give us your feedback, and review the changes. If that takes five days instead of two, the handoff shifts right. That's normal. The system doesn't go live until it's right, not until a specific day arrives on the calendar.
But in the standard case — one or two core integrations, a motivated client who has time to test — 5-12 days is real and repeatable. We've shipped it across multiple clients including Bodyline, JMJ Homes, and MaxSil. Different industries, different systems, same timeline.