Design systems visually,
with Gienie Architect
A system-design studio where architectures are drawn, not guessed. Describe what you need in plain English and Gienie Architect lays it out on the canvas — every component real, every connection validated against documented rules. Then analyze its weak points, walk it at C4 zoom levels, and ship it onward as docs, live embeds, or Terraform scaffolding.
Free to use in your browser — no account required.
An architecture studio that knows the rules
Every component on the Gienie canvas carries real-world placement rules, hard limits, and documented dataflows — and the AI has to obey them too.
Gienie Architect, your AI agent
Describe your system in plain English — or attach a requirements PDF or markdown file — and Gienie Architect designs the complete architecture. Hybrid models: fast answers for step-by-step edits, a deep model for full one-shot designs.
138 real components — plus yours
A curated catalog of databases, queues, CDNs, load balancers, caches, and more — each documented with real-world placement rules, hard limits, and supported dataflows. Define custom components with your own rules and they join the catalog, the palette, and the AI's vocabulary.
Machine-validated designs
Every design — human-drawn or AI-generated — is checked against the catalog rules. No client wired straight to a database, no cache used as a source of truth, no impossible dataflows.
Analyze, version, and record decisions
A resilience analyzer flags single points of failure, unreplicated stores, and queue hygiene issues on the live canvas. Save version snapshots, diff them visually, and have the AI write a MADR-format Architecture Decision Record from any change set.
C4 zoom levels & presentation
One design, three altitudes: a derived Context view collapses groups into systems, the canvas is your Container level, and any service drills down into its own Component screen. Presentation mode walks all three full-screen, with the dataflow animation playing.
Provider mapping & Terraform
Project a generic design onto AWS, Cloudflare, Azure, or GCP with exact managed-service names and honest coverage gaps — then export AWS or Cloudflare Terraform scaffolding, with variables, outputs, and a TODO for every real decision.
Interview practice with an AI scorecard
Twenty classic system-design interview problems — URL shortener to stock exchange — each with requirements, scale targets, and a countdown. Submit your design for an AI scorecard graded across five rubric dimensions, like a real interview. Three problems free.
Sharing, comments & live embeds
Invite collaborators as viewers or editors by GitHub username, discuss with anchored comment threads, or publish a public link. Embed the live design in Confluence, Notion, or a GitHub README — it updates as you save, and revoking the link kills every embed.
Rich exports
Export designs as JSON, PNG, JPG, SVG, animated GIF with dataflow animation, Markdown/Mermaid architecture docs labeled by C4 level, YAML specs, or Terraform — ready for READMEs, design reviews, and repos.
From requirements to a validated architecture
Describe your system
Tell Gienie Architect what you're building in plain English, or attach a requirements document (PDF or markdown). Scale, latency, budget — whatever you know.
Gienie Architect lays it out
The agent designs the full architecture on the canvas using real catalog components, then machine-validates every placement and connection against the documented rules.
Refine, analyze, ship it onward
Edit in conversation or by hand, run the resilience analyzer, present it at C4 zoom levels, and ship it onward — share links and live embeds for your team, ADRs for the record, Terraform scaffolding for the repo.
Rehearse the system-design interview before it happens
Twenty classic problems — URL shortener, rate limiter, news feed, ride-hailing, payment system, all the way to a stock exchange. Each one gives you the requirements, the scale targets, and a countdown clock. Design on the real canvas, then submit.
An AI interviewer grades your design across five rubric dimensions — requirements coverage, scalability, reliability, data model, and bottleneck awareness — crediting only what's actually on the canvas, with concrete strengths and gaps.
Three problems free · all twenty with Pro. In the app: ⋯ menu → Interview practice.
✕ No DLQ on the payment queue — retries will loop forever on poison messages.
✓ Cache-aside on the read path with sensible TTLs.
Your design doesn't stay in the app
Ship it into the places your team already reads — living, not screenshots.
Live iframe embeds
Drop a read-only embed into Confluence, Notion, or any wiki. It re-renders as the design is saved; revoking the share link kills every embed.
[](…)Live SVG in READMEs
One line of Markdown puts the current diagram in a GitHub README or Notion doc — always the latest saved version, never a stale screenshot.
resource "aws_lb" "checkout" {
load_balancer_type = "application"
# TODO: subnets, security_groups
}Terraform scaffolding
Export AWS or Cloudflare Terraform: one labeled resource per component, replicas as count, variables and outputs — a reviewed starting point for the repo.
Simple pricing, generous free tier
Start designing for free in your browser. Upgrade when you want the deep model, cloud saves, and collaboration.
Forever free — no account needed
- Full visual canvas, the 138-component catalog, and your own custom components
- AI on the fast model — 5 messages per day, covering chat, ADRs and interview scoring (free account)
- Designs saved locally, plus 3 cloud-saved designs
- Sharing with viewer/editor roles, comment threads, public SVG links and live embeds
- Machine validation, resilience analyzer, versions with visual diff
- C4 zoom levels, presentation mode, provider mapping
- All exports: JSON, images, animated GIF, Markdown/Mermaid, YAML — and Terraform (AWS/Cloudflare)
- Interview practice — 3 problems with AI scorecards
or $120/year — save 33%
- Everything in Free
- 300 AI credits every month — token-based, a quick edit ≈ 1 credit
- Gienie Architect deep model for one-shot full designs
- Requirement-document uploads (markdown & PDF)
- All twenty interview problems with AI scorecards
- AI-written ADRs from version diffs (billed as credits, ≈ 1 each)
- Unlimited cloud-saved designs, available on any device
- No daily cap on the fast model
Pro is currently invite-based: redeem a code from the dashboard account menu. Self-serve checkout (secure payment and global tax handling by our merchant of record) is coming soon. Prices in USD. Cancel anytime; 14-day money-back guarantee.
Frequently asked questions
Is there a free plan?
Yes. The Free plan includes the full canvas, the complete 138-component catalog, local designs, and every export format — no account needed: open app.gienie.ai and start designing. A free account adds the AI assistant on the fast model (5 messages per day), 3 cloud-saved designs, and sharing.
What does the AI actually know?
Gienie Architect works from the same component catalog you do: 138 components, each with documented placement rules, hard limits, and permitted dataflows. It doesn't design from vague memory — every change it makes is machine-validated against those rules, so you won't get a client wired directly to a database or a cache treated as a source of truth.
Can I use Gienie without an account?
Yes. Without signing in, your designs are stored locally in your browser and you can download them in any export format. Signing in with GitHub adds the AI assistant (5 fast-model messages per day on Free), cloud-saved designs (3 on Free, unlimited on Pro), and sharing with viewer/editor roles.
How does sharing work?
It works like Google Drive. Invite collaborators by their GitHub username and choose a role — viewer or editor — and discuss with comment threads anchored to specific components and connections. You can also create a public share link that renders the design as an SVG in any browser, so reviewers don't need the app or an account to see it.
Can I embed a design in Confluence, Notion, or a README?
Yes. Once link sharing is on, the Share dialog gives you two snippets: an iframe for Confluence, Notion, and wikis, and a one-line Markdown image for GitHub READMEs. Both are live — they show the design as it's currently saved, not a screenshot — and revoking the share link disables every embed at once.
How does interview practice work?
Pick one of twenty classic problems (URL shortener, news feed, ride-hailing, payment system…). You get the requirements, scale targets, and a countdown timer, and you design on the normal canvas. Submitting sends your design to an AI interviewer that scores five rubric dimensions — crediting only what's actually on the canvas — with a verdict, strengths, and concrete gaps. Three problems are free; Pro unlocks all twenty. Scoring is billed like a chat message (a free daily message, or ≈1 credit on Pro).
Is the Terraform export deployable?
It's deliberately a scaffold, not deployable infrastructure. You get one labeled resource per component (AWS or Cloudflare), replicas as count, variables and outputs, your capacity notes as sizing comments, and a TODO everywhere a real decision lives — networking, IAM, credentials. It's the starting point for a reviewed pull request, and every file says so in its header.
What are the AI limits?
Free includes 5 fast-model messages per day — plenty for iterating on a design. Pro gets 300 credits per month with no daily cap, and unlocks the deep model, which Gienie Architect uses to produce complete architectures in a single pass from your requirements. Credits are token-based: a quick edit costs about 1 credit, a one-shot full design typically 10–20.
How do refunds work?
Every first purchase comes with a 14-day money-back guarantee, no questions asked. Renewals are also refundable within 14 days of the charge. See the full Refund Policy for details.
Design your next system with Gienie
Open the canvas, describe what you're building, and watch a rule-correct architecture take shape.
Open the app