Read the bundled business documents, classify them, extract structured fields, reconcile invoice lines, and publish a checked report. Every stage performs local data processing and passes its files to the next activity.
Requires Go and Python 3.12+. Blender Render Farm also requires Docker. Run from a fresh checkout:
git clone https://github.com/UFFeScience/akoflow-examples.git
cd akoflow-examples/document-ai/document-intelligence
./prepare.sh && ./run.shPreparation downloads the Go YAML module. Results are in outputs/manifest.json, outputs/dag-validation.json, and outputs/stage-<activity>.json. Independent branches run concurrently as soon as their own dependencies and a local execution slot are available.
Start the local AkôFlow daemon with Docker available and configure AKOFLOW_API_TOKEN in your shell. From this example directory:
go run ../../tools/daemon-local -root ../.. -example document-ai/document-intelligenceThe command builds the declared image, validates Docker connectivity, registers local-only infrastructure, and executes the workflow through the daemon. Current verified screenshots are in screenshots/current/; their manifest identifies the exact run and image. The full daemon projection is in evidence/daemon-run.json.