| To begin discovery | Observational discovery can start without first obtaining source-system log access or table-level integration as the entry point. | Data Core is integration-centric, with 100+ prebuilt extractors. Task Mining provides a separate desktop source. |
|---|
| Employee endpoint | No-install, application-agnostic recording. The validated architecture uses remote and browser-connected recording rather than a recorder installed on each observed employee endpoint. | Desktop capture uses the Celonis Task Mining Client; current releases continue to ship Task Mining Client software. |
|---|
| Work discovery | Observation across applications and screens, including alternate paths, interruptions, and legacy and remote scenarios. | Task Mining captures desktop work outside major systems, including email and spreadsheets. AI-driven Task Discovery clusters desktop clicks into tasks. |
|---|
| Business data in context | Compound AI recognises and generates a business-data dictionary. A process engineer reviews, corrects and approves it before downstream use. | Business data from enterprise sources is combined with desktop actions through Data Core and the Context Model. This is a major platform strength. |
|---|
| Case and process reconstruction | Approved outputs are associated to the client’s cases, then combined deterministically into end-to-end paths and variants. | AI-driven Task Discovery connects desktop activity to business objects and end-to-end processes; the Context Model combines systems, applications, devices and business knowledge. |
|---|
| Time and variation | Pauses, alternate paths, frequency, different user methods, step time, and touch time against overall elapsed time. | Task Mining captures timestamped desktop activity and application behaviour; broader process analytics add throughput and variant context. |
|---|
| Deployment and data control | On-premises, customer cloud, or a dedicated hosted environment with dedicated resources, regional governance controls and encryption. | Task mining data is integrated into the Celonis Process Data Model and processed through the Celonis platform. |
|---|
| Operating model | A business-led discovery model, with process-engineer review, correction and approval at defined AI-recognition checkpoints. | Extensive low-code and enterprise UX. The combined data, task and process stack commonly involves multiple enterprise stakeholders and data setup. |
|---|
| Beyond discovery | Current validated focus: discovery, measurement and operational analysis. | The Context Model supports root-cause analysis, predictions, recommendations and what-if analysis, and the platform extends into orchestration and AI context. |
|---|