A full walkthrough of the actual flow, step by step — from a blank idea to deployable Terraform or Kubernetes config.
One thing worth knowing up front: almost everywhere you see a small circled inext to a component, a connection, or a config field, it's clickable — it explains, in plain language, why Archwise chose or suggested that specific thing for your project. It's the fastest way to actually understand the design instead of just accepting it.
1. Start a project
Your dashboardlists every project you've started, as cards or a compact list (toggle in the top corner), each tagged with its stage — Just Started, Brainstorm In Progress, Requirements Complete, or Architecture Ready. Click + New Project. Give it a name, then describe your idea in the Product Idea & Contextbox — a few sentences is enough ("a marketplace app connecting local tutors with students, expecting a few thousand users in year one"). If you're modernizing an existing system rather than starting from scratch, check I have an existing system to describe it or upload a short .txt/.md brief. Click Launch Brainstorm Workspaceto begin. (A project card's trash icon deletes it permanently, with a confirmation first — there's no undo.)
2. Brainstorm the details
You'll land in the Interactive Discovery Chat. A guided conversation asks the questions that actually shape the architecture — expected scale, budget, team size, compliance needs — often with clickable suggested replies so you don't have to type everything out. Answer at your own pace; when enough detail has been gathered, a banner reading "Discovery complete" appears and your answers are automatically turned into structured requirements.
3. Review your requirements
Once you're inside a project, everything lives under three tabs: 1. Requirements, 2. Architecture Diagram, and Team (covered later on). The Requirements tab shows what was extracted: what the product does, plus how it should perform — expected traffic, read/write pattern, data types, latency, budget range, team maturity, and security/compliance needs. Anything you want to correct or fill in is editable directly — click Edit, adjust a field (AI suggestions are offered per field, each with a "why" you can check), and Save Requirements. When it looks right, move to the 2. Architecture Diagram tab.
4. Generate and explore the architecture
Click Generate Architecture Design and Archwise reasons through a real design — not a generic template. The Multi-Cloud Design Board that follows lets you switch between AWS, Azure, GCP, Kubernetes, and a private/on-prem view of the same design, and between several ways of looking at it:
- Diagram — the interactive topology, with a cost breakdown and an overall Architecture Health Score.
- Compare Clouds — the same design side by side across providers.
- User Journey — a plain-language walkthrough of how a real request flows through your design.
- Migration Roadmap — shown only if you flagged an existing system, a step-by-step path to get there.
Click any component in the diagram to open its detail panel: a plain-language description, what it connects to and why, its estimated cost, and — for the currently chosen cloud service — Alternatives Considered, the other real services that could fill the same role and why this one won. The Flow Steps toggle above the diagram overlays numbered, colored steps on the connections matching the User Journey tab, so you can see exactly which path a request takes.
The Security Findings panel lists a deterministic security/compliance audit of the design. Curious what a design decision would change? The What-If Simulator lets you try alternatives without committing to them.
5. Fix security findings in one click
Many findings in the Security Findings panel show a Fix this button. Click it, confirm with Yes, fix it, and Archwise applies the fix directly to your architecture — no manual editing — and shows you exactly what changed as a before/after diff, scoped to just the affected component.
6. Edit the architecture yourself
Click 🔧 Edit Architecture(only available on the latest version) to open the editor as a full split-view: the diagram stays visible on one side, and your editing controls — Add Component, Add Connection, and the active connection list — sit on the other, so you never have to scroll back and forth between what you're changing and what it looks like. (On a narrower window it stacks top/bottom instead of side by side.) From here you can:
- Add a new component, with an optional note on why you're adding it — Archwise runs a quick relevance check and flags it if the component looks disconnected or redundant, before you save.
- Wire up or remove connections between components. Every connection Archwise generated on its own already carries a reasoning tooltip (the i icon) explaining why that link exists.
- Remove a component entirely with the small × badge on its corner.
- Click a component and use Switch to this under Alternatives Considered to swap it for a different real cloud service, or pick a more specific variant (e.g. which database engine) under its granular sub-choice section if one applies.
- Expand ⚙️ Technical Details (LLD) on a component to edit its actual config values (instance size, replica count, retention, and so on) directly, with a note on why.
Any structural problem — a component with no connections at all, compute with no database, or a circular chain of connections — shows up as a Blocker right in the editor, the same way Security Findings do: an explanation of why it matters, and (where Archwise can resolve it automatically) a Fix this → Yes, fix it button. Saving is blocked until every blocker is resolved.
When you're done, 💾 Save Changes persists it as a new version, or Cancel discards the draft and leaves the last saved version untouched.
7. Regenerate Design — start over from your requirements
Regenerate Design(next to the version selector) is different from manual editing: instead of changing individual components, it reruns the full design process against your currently saved requirements — useful if you've updated a requirement field and want a fresh design built around it. It always previews first: you'll see the full added/modified/removed diff and a cost comparison before anything is created, with ✓ Apply as New Version or Discardto decide. It's disabled while you're mid-edit, since it rebuilds from your saved requirements and would otherwise ignore whatever you're currently editing.
8. Versions — nothing is ever silently overwritten
Every save, regenerate, or applied change creates a new version rather than overwriting the last one — and you'll now see an explicit "Saved — this is now version vX.Y"confirmation whenever that happens, so there's never any doubt about whether something actually persisted. Older versions stay fully intact and browsable:
- The Version dropdown at the top switches between any past version (read-only) or back to the latest.
- Evolution History (further down the page) shows a phase-by-phase account of what changed at each version and why, with a View this version button on each entry.
- Viewing an older version shows a ⏪ Revert to this version button — it creates a brand-new version that's a copy of the old one and becomes the new latest, without touching or deleting anything in between.
9. Work with your team
The Team tab (next to Requirements and Architecture Diagram) is where you invite collaborators — the project owner enters an email and picks a role, Editor or Viewer, and Invite. The other person needs an account on Archwise already; no invite email is sent, so tell them directly. Owners can Remove a member later. Anyone with access can leave and read Comments on the project, and delete their own.
To share a read-only view with someone who shouldn't log in at all — a stakeholder, a client — click Share above the diagram and + Create new link. Anyone with that link can view the diagram, cost, and flow story with no account and no edit/export/generate controls. Revoke it anytime from the same menu.
10. Export what you need
When you're ready to build it for real, export options are on the same screen:
- Export Terraform / Export Kubernetes Config — real, deployable
.tffiles or Kubernetes YAML manifests, zipped, matching whichever provider tab is active. - Export Image — a PNG/SVG of just the diagram.
- Export Docs — a Markdown file with the project summary, this provider's flow story, and the full component list with reasoning — real documentation, not just a picture or code.
- Executive Summary — a non-technical PDF (no diagrams or code) for sharing with stakeholders who just need the cost/security picture.
Every export can also be emailed to yourself instead of downloaded directly.
11. Report a change later
Your architecture isn't frozen once it's generated. Go back to the same chat and describe what changed — "scale increased to 50k users," a new feature, a bigger budget — using the Report a change input. Archwise analyzes the impact and proposes specific updates in a Chat-Proposed Changes panel, where you can accept or reject each one individually (or all at once) before anything is applied. Every change is versioned, so nothing is ever silently overwritten.
Free vs. paid usage
The free plan renews every 7 days with enough brainstorm sessions, architecture generations, and updates to try the whole flow above. The paid plan gets its own daily allowance instead. See pricing for exact numbers, or the Help & FAQ pageif something doesn't match what you're seeing.