Scenario Packs — business chains ready to run

Each pack is a complete chain: build, run, deliver. Changing it takes one sentence.

Cross-industry (finance / retail / manufacturing)

Nightly Batch Analytics

Business data piles up all day; at night, staff manually run numbers, compute metrics, generate reports, and send alerts—one mistake corrupts the next day.

  1. 1Data Ingestion

    Multi-source extraction and cleansing into the warehouse (ETL orchestration)

  2. 2Metric Computation

    Batch metric computation via the full metric engine

  3. 3Report Generation

    End-of-day grouping, aggregation, and report generation (cron driven)

  4. 4Dashboard Materialization

    Materialize management-cockpit homepage metrics

  5. 5AI Interpretation

    Anomaly interpretation—AI turns metric swings into plain-language summaries

  6. 6Ask-Your-Data

    Ask about reports in plain language—no waiting on the data-warehouse queue

  7. 7Archive & Replay

    Historical archiving and replay (idempotent, dedupe-safe)

Want a change? Just say it

“Start the nightly batch earlier, at 11pm”

Set the first step's cron of the nightly chain to 0 23 * * *

“If upstream data isn't complete, run whatever is ready and log the missing parts”

Switch the ETL step's missing-source policy from fail to partial + log

“Only include metric swings above 20% in the morning report”

Set the Cortex anomaly-interpretation filter threshold to 0.2

“Re-run last month's data for the 15th”

Replay the archive step over the given date window (idempotent by design)

Try it live: say it, see how the system changes

No signup. Type one sentence; the AI shows how it would change the chain above.

E-commerce

E-commerce Full Chain

Product records, content production, listing, payments, reconciliation, analytics, and re-marketing are scattered across disconnected tools, stitched together by manual handoffs.

  1. 1OMS Modeling

    Product / inventory / supplier / order objects—67 business tables

  2. 2AIGC Production

    Product photo → marketing video → asset upload → creative → ad publish

  3. 3Site & Listing

    AI site builder—products listed automatically

  4. 4AI Support

    AI answers customer inquiries (multi-model routing gateway)

  5. 5Payment Loop

    WeChat / Alipay ordering, payment collection, and reconciliation

  6. 6Ask-Your-Data

    Ask about sales / inventory / conversion in plain language

  7. 7Auto Marketing

    Scheduled jobs—automated re-marketing and slow-mover alerts

Want a change? Just say it

“Change the weekly report to every morning at 8am”

Change the "ask-your-data" step's Loom schedule from weekly to daily, cron 0 8 * * *

“Drop the payment step—we only need the website display for now”

Remove the payment step from the chain config; keep website display only

“After 10pm, stop auto-replies and route to a human”

Add a 22:00-08:00 time-window condition to the Cortex answering step; route to tickets

“Switch video generation to a cheaper provider”

Adjust provider routing priority for the video step; keep fallback candidates

Try it live: say it, see how the system changes

No signup. Type one sentence; the AI shows how it would change the chain above.

Media & Content

Media Matrix Operations

Topic selection, content production, multi-platform publishing, and performance feedback each run in silos; throughput and quality depend on manual oversight.

  1. 1Topic Library

    Topic library and asset ledger objects (modeled on the e-commerce template object system)

  2. 2Batch Production

    One-shot auto publish—asset upload + creative + campaign + launch on Google Ads / Meta Ads

  3. 3Multi-platform Publish

    22 AIGC pipeline scenarios—per-product video, provider routing, async video polling

  4. 4AI Review

    Copy / visual prompt generation with safety-review model routing

  5. 5Ask-Your-Data

    Ask about publishing performance—by platform / content / time slot

  6. 6Topic Optimization Loop

    Scheduled batch production and publish dispatch

Want a change? Just say it

“Publish only on Douyin and Xiaohongshu; pause the others”

Narrow the target_platform whitelist of the aigc_to_publish step; skip the rest

“Cap all videos at 30 seconds”

Set the video step's duration cap to 30 seconds

“Publish one post at 6pm every Tue/Thu/Fri”

Set the publish schedule cron to 0 18 * * 2,4,5

“Auto-spin three derivative videos from any hit topic”

After the data-feedback step flags a hit, branch into the video step ×3

Try it live: say it, see how the system changes

No signup. Type one sentence; the AI shows how it would change the chain above.