Architecture-first · AI proposes, rules verify
Most AI tools generate code and hope the architecture catches up. Solarch flips that — it generates architecture first, grounded in canonical patterns, validated by a strict Rules Engine, refined through a self-correcting loop. The AI proposes; the rules verify; only correct graphs land on the canvas.
Omni-bar · two modes
One command bar, two minds — an autonomous builder and a conversational architect, without ever leaving the canvas.
Type a sentence or sketch a box. Solarch turns the intent in your head into a validated architecture — and that architecture into working code. No blank-canvas paralysis, no boilerplate; just the system you already had in mind.
Inside Solarch
The whole backend on a single surface — grounded by an AI architect, kept honest by the rules, and type-safe end to end.
8 families · 16 edges
Controllers, services, repositories, tables, DTOs, queues — every part of your backend on one canvas, with the 16 semantic edges between them.
8 families
GraphRAG · LangGraph
LangGraph grounds every plan in a vector-indexed library of canonical patterns.
1 home · N references
Each node lives in one tab; others import it as a ghost reference — not a copy.
Grounded context
Schema → contract → UI
The schema becomes an API contract becomes a typed client — every layer checked at compile time. Change a column, and the frontend stops compiling until you fix it.
create table users (id uuid primary key,email text not null);
Rules engine · default-deny
7 anti-patterns · 3 conditional checks
Controllers aren't data-access. A Controller can't reach a Table directly — route through a Service or Repository.
Default-deny — every connection that isn't whitelisted is forbidden.
Try to wire a Controller straight to its Table…
Rejected the moment you draw it — not three commits later.
So the agent inserts a Service and a Repository, and re-applies.
Rejections feed back until the graph is clean — or it never commits.
Hybrid code engine · stage 1 + stage 2
A deterministic compiler turns the graph into a skeleton with zero tokens — DTOs, entities, constructor injection straight from the edges. Then a surgical AI fills only the empty function bodies. Export it as code, a Mermaid diagram, or AI-readable memory.
Source graph
Generated code
export class CreateUserDto {@IsString() @IsNotEmpty()email: string;@MinLength(8)password: string;}
Pricing
From a $5 sketchpad to a self-correcting code engine — pay for as much of the pipeline as you need. Every plan starts with a 7-day trial.
✓ 7-day trial
Start sketching for free — the Rules Engine validates every connection. Upgrade when you need more.
✓ 7-day trial
The drawing tool. Sketch backend architecture as a node/edge graph — the Rules Engine auto-validates every connection.
✓ 7-day trial
Everything in Draw, plus AI that scaffolds your project — ~80% ships as boilerplate; the algorithmic logic stays yours.
✓ 7-day trial
Everything in Build, but Solarch writes the whole codebase — full generation, not just the skeleton. Extend any limit at discounted pricing.
Build together — shared projects with single-home nodes and ghost references across the org.
✓ 7-day trial
Self-host, SSO, audit logs, a custom rules matrix, and dedicated support.
✓ 7-day trial
FAQ
