Axis Agents
← All work

Desktop GIS to cloud · The first version of axis-migrate

Migration Engine: modernise the workflow, then prove the new result

We reconstruct a legacy geospatial workflow, rebuild it for a modern environment and compare the new result with the old, so your team can decide whether it is ready to accept. The Migration Engine was the first version of this work. It ran the six-stage journey that the axis-migrate hosted service runs today.

The Migration Engine drawing the workflow graph of an ArcPy script, with steps such as Load Flood Hazard Zones, Buffer Properties, Enrich with Wind Risk and Export Analysis Results. On the right, an analysis summary: 8 tasks, 7 data flows.

A migration your expert can accept

The job is not finished when new code appears. It is finished when the workflow is understood, the implementation is owned and the differences can be reviewed.

Source workflow

  • ArcGIS
  • QGIS
  • FME
  • Python
  • Jupyter

Migration Engine: understand, rebuild, run and compare

Chosen environment

  • Databricks
  • AWS
  • GCP
  • Snowflake
  • PostGIS

What your team gets

  1. 1

    An understood workflow

    The recovered steps, dependencies, decisions and open questions are explicit enough for a domain expert to correct.

  2. 2

    An owned implementation

    The result is delivered for the environment your team chooses, with code and artefacts your team can inspect.

  3. 3

    A parity record

    Comparisons, provenance and unresolved differences travel with the implementation instead of disappearing into a success message.

Six stages, one acceptance record

The lifecycle comes from the first shipped Migration Engine.

  1. 1

    Discover

    Inspect the files, scripts, tools, dependencies and sample outputs that make up the real workflow.

  2. 2

    Interview

    Surface the choices and edge cases that live with the analyst rather than in the code.

  3. 3

    Refine

    Turn the recovered process into a workflow the team can inspect, correct and accept.

  4. 4

    Generate

    Build an owned implementation for the chosen target, with the assumptions kept visible.

  5. 5

    Deploy

    Run the rebuilt workflow in a controlled target environment and keep a record of what happened.

  6. 6

    Verify

    Compare outputs, investigate differences and package the comparison for a person to accept.

The Discover step after a scan: one file parsed, complexity 9.0, and the detected ArcPy operations, including Buffer_analysis, SpatialJoin_analysis and CalculateField_management.
Discover, after scanning one ArcPy script. Every geoprocessing call is listed before anyone reads the code.

Built once, then rebuilt to go further

The first Migration Engine ran the six-stage journey, generated implementations and compared raster, vector and tabular outputs.

In July 2026 we started rebuilding it to bring project understanding, long-running execution, comparison and review onto one foundation for early pilots. That rebuild is axis-migrate: a hosted service, and an open-source plugin for Claude Code and Codex.

Read the axis-migrate case study

Tell us which legacy workflow matters

We work with teams that have a defined source workflow, a target environment and a real acceptance question.

See What an Agent Could Do in Your Business

Describe how the job is done today and who signs off. A short note is enough.