Deterministic workflow
A deterministic workflow is an ordered, parameterised sequence of steps that produces the same result when run again with the same inputs, making it suitable for recurring reporting and review processes.
Conversational analysis is exploratory by nature: the same question asked twice may take a different route and arrive at a slightly different answer. That variability is acceptable while investigating and unacceptable once a process becomes recurring. A monthly regulatory submission cannot depend on how a question happened to be phrased.
A deterministic workflow resolves this by fixing the steps. Parameters are declared explicitly, the sequence is stored rather than improvised, and each execution produces a run record. Because the definition is stable, the output of two runs can be compared meaningfully — a difference indicates the data changed, not the method.
The practical value is that exploratory work can graduate. An analysis worked out interactively becomes a workflow that runs on a schedule or on demand, without being rewritten in a different system by a different team.
What distinguishes it
- Ordered steps with explicitly declared parameters, stored rather than improvised
- Same inputs produce the same output, so differences between runs are meaningful
- Each execution produces a run record that can be reviewed and compared
- Runs manually, on a schedule, or when called by another process
In DataQube
In DataQube a workflow is a first-class artefact, distinct from the thread it came from. Analysis worked out in conversation can be extracted into a workflow with ordered steps, parameters, and permissions; runs append their results back to a thread, so scheduled output stays connected to the same auditable record as interactive work.
Common questions
Why not just re-ask the AI the same question each month?
Because the route it takes may differ, and so may the answer. For a recurring process you need the difference between two runs to reflect a change in the data, not a change in how the question was interpreted. Fixing the steps is what makes the comparison meaningful.
How is a deterministic workflow different from a notebook?
A notebook view is a projection of work that already happened — a record for review. A workflow is a definition intended to be run again, with declared parameters and its own run history. One documents, the other executes.
Can a deterministic workflow use a language model?
Yes, though the determinism then applies to the structure rather than to every token generated. The steps, sources, parameters, and permissions stay fixed, which is what makes runs comparable, even where a model contributes narrative text.
Related terms
AI data agent
An AI data agent is software that answers analytical questions by planning and executing its own steps — querying approved databases, running code, and citing what it used — rather than only generating text.
Data provenance
Data provenance is the recorded lineage of a result: the sources, queries, transformations, and decisions that produced it, retained in enough detail that someone else can verify the figure independently.
Concurrent-session licensing
Concurrent-session licensing charges according to how many users are active at the same time, rather than how many named accounts exist, so occasional users do not each consume a permanent seat.