Back to catalog
Pearl
Apache-2.0Pearl is an open-source Slack AI operator replacing black-box tools like Viktor with self-hostable TypeScript workflows and full execution observability.
Open-source alternative to — compare it against:
Overview
An infrastructure team managing incident triage watches alerts pile up across monitoring tools while on-call engineers manually query databases and update tickets. Bubble Lab tackles this operational sprawl by embedding Pearl directly into Slack, backed by an open-core TypeScript execution engine that translates conversational requests into deterministic backend workflows. Developers write, test, and host typed workflow pipelines in their own repositories rather than locking logic inside a closed vendor cloud.
Why Pearl instead of Viktor
Viktor operates as a hosted AI coworker inside Slack, running automated tasks across thousands of SaaS connectors through a proprietary cloud environment. That setup suits non-technical teams that want instant cross-app automations without maintaining code. However, it treats operational logic as a black box: workflows cannot be committed to Git, verified in CI pipelines, or audited when an agent runs unexpected queries against production datastores.
Pearl prioritizes workflow fit for engineers by defining automations as typed TypeScript classes. Developers specify deterministic execution graphs with structured inputs and step-level logging that run locally or on private infrastructure. While Viktor has an edge in out-of-the-box business integrations, Pearl gives engineering teams an auditable development lifecycle where operational runbooks live alongside application code.
What it does well
- Type-safe workflow authoring: Automations compile from TypeScript classes extending
BubbleFlow, giving developers compile-time type validation, autocompletion, and schema enforcement across steps. - Step-level execution observability: Every run logs execution duration, memory consumption, token metrics, and node sequences, making performance bottlenecks obvious during development.
- Local studio testing: Developers run a local web studio on localhost:3000 to construct, inspect, and run workflows before committing changes to Git.
- Modular node architecture: Integration steps isolate into reusable bubble classes with typed inputs and standard
.action()methods for straightforward maintenance. - Git-driven deployment pipelines: Because workflows exist as code files, teams manage operational changes through regular pull requests, branch reviews, and continuous integration runners.