Data Analysis

Anomaly DiagnosisPlatform Operations

AE Analysis

Operate every analysis asset on the AE analytics platform, reports, dashboards, Ad-hoc analysis, cohorts, tags, metrics, and alerts, through ae-cli, with 79 commands covering the complete analysis-side workflow.

IndustriesAll industries

Agent · Skills Live

skill · loadedindustry+100

Run this Skill on my data

Query console

generated

query.sql · generated

SELECT channel, count(*)

FROM v_payment

GROUP BY channel — 0.4s

12 rows returned

Analysis readyevidence attached

The AE Analysis platform-operations Skill is a full-workflow platform-operations capability in ThinkingAI's industry Skill library. It helps data and operations teams operate every analysis asset on the AE analytics platform through the ae-cli command-line tool, report queries, dashboard creation, Ad-hoc model analysis, cohort and tag management, metric definition and governance, alert configuration, and metadata inspection, with 79 commands covering the complete analysis-side workflow, moving from clicking through the web UI one item at a time to precise batch execution from the command line. This is ThinkingAI's decade of industry experience, packaged into ready-to-use Skills.

Live walkthrough

AgentSave as AgentExport session

ae-analysis

Pull our project's new-user volume and retention for the last 14 days for the weekly review. If that report we built last time still exists, just use its metric definitions instead of rebuilding it.

Project gate: verify a sample MMO game DEMO (project_id=2)

Project check passed, query access confirmed

+list_reports fuzzy search for reports related to "new users, retention"

Matched "New User Volume and Retention," reused directly instead of rebuilding

+query_report_data pulls last 14 days of daily new users and next-day retention

41,286 cumulative new users over 14 days

7-day retention is a metric the report does not cover, so Builder generates a retention analysis QP, then an ad-hoc follow-up query fills it in

Aggregate period averages, +get_resource_url returns the report access link

The last 14 days are pulled, and overall it's healthy: 41,286 cumulative new users (2,949/day), average next-day retention of 32.8%, average 7-day retention of 13.6%, a steady trend with no anomalies. The whole run reused your existing report's metric definitions, with no duplicate modeling.

1. New users and retention trend, last 14 days

Next-day / 7-day retention of new users, last 14 days　Unit: %

Next-day retention7-day retention

Two notes on the gaps: next-day retention for the 07-14 cohort won't be complete until today ends; 7-day retention is only complete for the 07-01 through 07-07 cohorts, which have fully cleared the window. Data points that haven't matured are left blank per convention, with no extrapolation.

2. Retention averages by period

Period	Avg. new users/day	Avg. next-day retention	Avg. 7-day retention
First 7 days (07-01 to 07-07)	2,876	32.9%	13.6%
Last 7 days (07-08 to 07-14)	3,022	32.6%	N/A (window not matured)
Full period	2,949	32.8%	13.6%

← Scroll to see more

The last 7 days averaged 3,022 new users/day, +5.1% vs. the first 7 days; average next-day retention went 32.9% → 32.6%, a 0.3pp swing within normal range: the volume increase did not dilute retention quality.

3. Analysis assets used in this run

Report "New User Volume and Retention," matched via +list_reports fuzzy search, reused its metric definitions directly for the numbers

7-day retention is a metric the report does not cover, so it ran through a Builder-generated retention analysis QP plus an ad-hoc follow-up, with no hand-written parameters

The report access link is already returned via +get_resource_url, ready to paste into the weekly review deck

Recommendation

If you want retention broken out by channel next week, just change the grouping parameter on this same report and reuse it, with no need to rebuild. If "new users + retention" is something you'll track long term, I'd suggest pinning it as a weekly-review dashboard component to save the manual pull each week.

The report "New User Volume and Retention" is pinned to the project dashboard and refreshes automatically every day.

On your data

That was a simulated run

Leave your work email and we will run a live walkthrough on your real business data.

The problem

The AE analytics platform is the core tool for data analysis, yet more than 60% of teams hit efficiency bottlenecks: creating a dashboard means manually configuring 10+ report widgets and filters, averaging 15 to 20 minutes; running an Ad-hoc analysis means hand-writing QP JSON, with field, event, and property names all from memory or the docs and an error rate above 40%; and cohort, tag, and metric definitions require frequent trial and error because of complex parameter formats. A more subtle problem is difficult resource discovery: not knowing which reports, dashboards, and metrics already exist in the project, teams search from scratch or recreate duplicates every time.

What it does

Full 79-command coverage: analysis (40) plus audience (14) plus meta (23) plus common (2), covering all operations across reports, dashboards, Ad-hoc, cohorts, tags, metrics, alerts, and metadata

Project-gating mechanism: every command that needs a project_id validates the project first, with no guessing and no mix-ups

Closed-loop links on writes: after creating a resource, it automatically fetches an accessible URL, completing creation to sharing in one step

Fuzzy-search fallback: when a resource is not found, it does a fuzzy search first and a full fallback next, never operating on something that does not exist based on a guess

Builder mandatory first: the four model types (event/retention/funnel/prop_analysis) must go through Builder to generate QP first, with no hand-writing and no guessing

When to use it

You need to create or query reports, dashboards, or alert rules

You need to run Ad-hoc event analysis, retention analysis, funnel analysis, or property analysis

You need to manage the creation, update, and refresh of cohorts and tags

You need to define or govern metrics, virtual events, or virtual properties

You need to view project metadata (event list, property list, project config)

You need to find existing reports or dashboards in the project to avoid duplicates

In the field

Case

A game operations team · batch-creating monthly dashboards

The team needed to create 6 dashboards each month, averaging 20 minutes each by hand. With the AE Analysis Skill, the system first ran +list_dashboards to search existing dashboards and avoid duplicates, then +create_dashboard to batch-create the missing ones and configure report widgets, and finally +get_resource_url to return accessible links. The 6 dashboards went from 120 minutes to 15 minutes, 8x faster.

FAQ

Can this Skill replace the web UI?

Yes. ae-cli commands cover the vast majority of analysis-side web operations, and the command-line mode supports batch execution, parameter validation, and result reuse.

What is the difference between Ad-hoc analysis and report queries?

Report queries use precomputed data from existing reports, which is faster and more stable. Ad-hoc analysis uses Builder to generate QP and computes in real time, which is more flexible but slower.

How does it relate to ae-analysis-global?

ae-analysis is the base Skill, and ae-analysis-global is a multi-cluster overlay. Single-cluster projects use only ae-analysis, while multi-cluster projects add the global routing rules.