はじめに

Task Mining, AI-based desktop activity analysis that surfaces evidence of automation opportunities and captures task variations, with automation skeleton export as PDD and XAML files.

Task Mining helps you build a deeper understanding of existing processes happening on your team's desktops. It records variations of a known task, uses AI to cluster and merge the captured traces, and surfaces task-level evidence, such as task graphs, variants, and statistics, that analysts and subject matter experts (SMEs) use to identify automation opportunities and process improvement areas.

Task Mining は以下のような特長を備えており、自動化のリーダーからビジネス アナリスト、自動化の開発者、そして従業員に至るあらゆる人々にメリットをもたらします。

• 従来のコンサルティングを主としたアプローチと比較して、コストと時間を節約できます。
• 自動化の機会を客観的に評価できるビジネス チームに参加してもらうことができます。
• 業務の発見と分析の部分を加速して、プロセスの文書化にかかる時間を短縮できます。
• オートメーション スケルトンを使用して自動化を迅速に開始できます。
• プライバシーと General Data Protection Regulation (GDPR) に同意し、準拠しています。

Task Mining (formerly known as Assisted Task Mining) empowers you to capture variations of a known task and review and merge the results in a collaborative manner. With the help of Task Mining, you can:

• collect variations of a specific task by capturing each action you perform (mouse clicks, keystrokes, and hotkeys)
• merge variations into a comprehensive picture of your task for end-to-end understanding and further analysis
• edit the task graph and annotate actions
• accelerate actionability by generating essential assets for automation implementation: a PDD and an XAML file

What Task Mining does and does not do

Task Mining is built around a known task: one that a Business Analyst or Process Subject Matter Expert (SME) has already identified as worth investigating.

Task Mining does:

• Recording and comparing variations (traces) of a known task
• Using AI to cluster screenshots and merge traces into a single task graph with decision points
• Surfacing task-level evidence, such as variants, step statistics, and time-per-action data
• Exporting artifacts, including a Process Definition Document (PDD) and a Studio XAML skeleton, for further use

Task Mining does not:

• Automatically discovering unknown tasks across your organization
• Producing root-cause findings or a ranked list of automation recommendations
• Replacing the review of a Business Analyst or SME
• Analyzing end-to-end system event logs across business systems: see Process Mining for that capability

How Task Mining uses AI

Task Mining's AI clusters screenshots of the same application and screen across traces, then merges matching steps and creates decision points where actions differ. This clustering and merging produces the consolidated task graph, but the AI does not rank findings or recommend which automation candidate to pursue. Determining automation potential and submitting a candidate to Automation Hub remains a manual step performed by a Business Analyst or SME.

Expected workflow

1. Defining the known task to analyze.
2. Collecting comparable traces of that task from one or more recording users.
3. Merging or comparing the traces to consolidate the variations into one task graph. See Merge traces.
4. Reviewing the merged trace as a Business Analyst or SME.
5. Quantifying the finding, for example by estimating time savings, frequency, or complexity.
6. Submitting the candidate to Automation Hub for further assessment. See Automation Hub.