Coming soon
We are building Ghost Observatory to redefine process intelligence: detailed knowledge of how work happens, available to everyone. Inspect bottlenecks, workload and controls in your own records. Follow the calculations and keep control of the result. The complete core will be free of charge.
Object-centric process map from the development demo.
01 / A finding
The sample result holds 300 synthetic claims and 7,402 events. One rule runs across all of them: the person who assesses a claim must not approve it. Here is what the Engine found.
Follow the evidence109 of 224
That is 49% of the claims the rule could check; 76 could not be checked.
Explore by decision
Find recoverable capacity in rework, repeated checks and idle time. Build the case for stronger local production.
Explore manufacturing 02 / ControlCheck who assessed and approved each claim. See which approvals breach the rule and which cannot be checked from the available records.
Inspect the sample 03 / OperationsFind slow steps and delayed handovers. Explore the actual Reader interface to see how your team can investigate the flow of work.
See the process view02 / Business value
A target can be met while the operation gets worse. Fast completions may conceal repeated work. A team’s apparent delay may begin with a late handover elsewhere. Totals alone leave those questions unanswered.
Ghost Observatory connects each result to the records behind it. Inspect the event sequence, the people involved and the time between steps. Compare similar work across people, teams and periods. Missing records and rejected rows stay visible.
Make the decision answer to the evidence. Check whether an employee faced more difficult assignments, inherited a delay or carried a disproportionate workload. Use the findings to challenge unsupported judgments and decide which process needs to change.
The records still need scrutiny. A missing event cannot explain what happened off the system, and a recorded delay needs operational context. The same source trail lets employees and managers question the conclusion.
03 / Fairness through evidence
Make assignments, workload and delays visible to the people making and questioning a decision.
Process intelligence can show assignments, volumes, handovers, rework and delays beyond an employee's control. Compare people and groups on genuinely comparable work, expose inconsistent application of rules and give decisions a record that can be checked.
That evidence helps challenge unfair treatment and investigate potential discrimination. But a calculation is only as sound as the records, definitions and context behind it. Missing events, different roles and unequal workloads must be visible; not turned into a misleading score.
Explore people and process analysis04 / Method
Start with exports from your existing systems. A YAML contract maps the work object, activity, time and person in each export; 114 contracts cover 23 system families. The Engine reconstructs the process on your machine and publishes one SQLite result file for the Reader and your own analysis tools. Core product free of charge. See Technology for access controls, pseudonymisation and the full architecture.
05 / The Reader
The Reader’s 5 workspaces cover process flows, individual work objects, people and timing. Every view reads the same result file.
Compare the recorded activity sequences, repeated steps and completion times of work objects.
Explore the Reader06 / Your data. Your control.
Ghost Observatory runs on your infrastructure. We do not receive or collect your process data, analyses or results, and the core runs without calling a model or sending telemetry to Ghost Citadel.
Choose to connect an external AI service? You decide what information to send and to whom. Or keep inference local. No AI model required: the Engine remains deterministic.
Start with the method, check what the measures establish, then inspect the input requirements for your own records.
