When to use Atelier
Your operational playbooks weave together Sage skills, PC Studio events, and human decisions — and you want versioned Flows your team can fork, with a single trace timeline that survives across products.
See Atelier pricing →Compare · Atelier vs Make
Make draws diagrams that move data. Atelier runs operational playbooks across the products your team actually uses.
Make (formerly Integromat) gives you a visual canvas to wire together hundreds of SaaS connectors with deep branching and array mapping. It's the right tool when the work itself is "transform this payload, then route it." Atelier is purpose-built for a different shape of work — recurring, multi-step operational playbooks where Sage reasoning, Sentinel verdicts, RFP drafts, and human approvals are first-class citizens. Same identity, same tenancy, same trace. You can still call Make from an Atelier Flow when the long-tail integration is the job; the deep PC Studio steps are native.
When to use Make
You have a transformation-heavy automation problem — incoming webhook payloads that need to be mapped, filtered, iterated, and pushed to a long tail of third-party SaaS apps that don't share an identity layer. Make's visual mapper and iterator modules are best-in-class for that shape of work.
Visual workflow builder for non-engineers
Side by side
Capability comparison
Feature
Atelier
Make
Native to Perpetual Core Studio (Sage, Sentinel, RFP, Janice)
First-class step kinds + shared identity
No native PC Studio integration
Agent-augmented steps (LLM with shared org memory)
sage_skill step kind, server-to-server
OpenAI / Anthropic modules via API key only
Visual scenario builder with drag-and-drop modules
Flows are JSON-shaped + form-driven; visual editor on the v1.1 roadmap
Make's differentiator — best-in-class visual canvas
Advanced array iterators + aggregators
Loop steps over context arrays; aggregator step kind in v1.1
Iterator + Aggregator modules — extremely flexible
Conditional gates on accumulated context
CEL-like expressions on run context
Router module with multi-path filters
Human approval steps with checkpoint timeline
human_task step kind with assignment + checkpoint
No first-class human-in-the-loop primitive — workarounds via email/Slack
Multi-tenant by org (RLS-enforced isolation)
Every row carries org_id; pc_runtime role enforced
Organization plan adds team folders, not per-tenant isolation
Append-only trace timeline across products
atelier_traces queryable via SQL
Scenario execution history with retention per plan
Versioned Flow definitions with global library + per-org forks
atelier_flow_versions + 5 starter Flows on day one
Scenario templates + version history (Pro+)
Per-run cost predictability
Fixed run cap per tier; $0.05/run overage
Operations-based pricing — meters every module invocation
Public API for external automation
Server-to-server today; first-party API in v1.1
REST API + webhook endpoints
Cross-Supabase / cross-product tenancy bridge
pc_external_org_map handles Sentinel (separate DB) + external sources
N/A — Make is its own data plane
On pricing
Make tiers (2026): Free 1,000 ops/mo, Core $10.59/mo (10K ops), Pro $18.82/mo (10K ops + features), Teams $34.12/mo, Enterprise contact. Make's per-operation pricing is much cheaper than Atelier per-run, but the operations meter is fine-grained — a single Atelier run can represent dozens of Make operations stitched together with an LLM call and a human approval. They're different units; the right question is whether your work is "move data" or "run a playbook."
Honest take
Different shapes of work. Often used together.
If your job is mapping incoming payloads onto outgoing APIs across a long tail of SaaS, Make's visual builder is hard to beat. If your job is running operational playbooks on the products your team already lives in, Atelier is built for that shape. A common pattern: Atelier Flows handle the PC Studio + human decisions; Make scenarios handle the long tail of third-party glue downstream.
Open a workspace
Stand up your team's first Flow this afternoon.
Atelier opens in fifteen minutes with five starter Flows and every Perpetual Core product already connected. Evaluate for fourteen days; bring your team when you're ready.