BergFlow and BergAuto documentation.

BergFlow

BergFlow is an invite-only annotation and review workspace for operational data programs. It organizes work into Projects, Batches, Tasks, and Delivery Batches so managers can ingest data, assign review work, measure quality, and export approved results.

Core objects

• Projects define the operating scope, client, task template, and access boundaries.
• Batches ingest CSV or Excel rows, media references, and task metadata.
• Tasks carry the unit of work assigned to annotators and reviewers.
• Delivery Batches package reviewed work for export and client handoff.

Roles

• Managers and admins configure projects, upload data, assign teams, and monitor delivery.
• Reviewers accept, reject, or route tasks for correction.
• Annotators complete assigned tasks against project templates and guidelines.

Operational notes

The source docs describe row limits up to 10,000 rows, chunking around 1,000 tasks, review accept and reject flows, skipped-task handling, task expiry, GT reports, dashboards, and delivery exports.

BergAuto

BergAuto is a visual workflow automation studio for repeatable AI-powered operational pipelines. It is built for workflows that need inputs, AI steps, transforms, conditions, database lookups, API calls, wait states, and structured outputs.

Workflow blocks

• Input captures text, structured fields, images, PDFs, Word files, or spreadsheet rows.
• AI and Transform steps generate, classify, enrich, summarize, or reshape data.
• Condition, Database, and API steps route work or connect external systems.
• Wait steps create human-in-the-loop approval pauses.
• Output steps produce summaries, classifications, reports, images, or exportable rows.

Run surfaces

• Demo Showcase for testing repeatable automation demos.
• Workflows for reusable production pipelines.
• Agents for environment presets such as brand, region, channel, or policy context.
• Eval for run logs, status, cost, and step-level inspection.

How they work together

BergFlow owns structured human work: task templates, assignment, review, QA, and delivery. BergAuto owns repeatable automation: AI steps, transformations, approvals, and exports. Charterr and Quinn provide the operating layer around both, so queue ownership, exception handling, and reviewer accountability are not left to the tool alone.

Controls to preserve

• Keep role-scoped access on every project and workflow.
• Use confidence thresholds to decide when automation acts, waits, or escalates.
• Keep reviewer identity, timestamp, status, and outcome attached to every decision.
• Export delivery data only after review and QA rules pass.