Finance ops · Production

ChatGPT for Accounting: How Digits is using Generative Machine Learning to transform finance

The problem

Accountants on the Digits platform needed to ask clients clarifying questions about individual transactions, but typing these questions manually for every interaction cost time and introduced tedium into each client communication.

Workflow diagram · grounded in source
1
Accountant flags transaction
trigger
“assist accountants when they want to raise questions about a transaction with a client”
2
T5 model generates question
ai_action
“this model generates questions based on the description and features of the transaction”
3
Toxicity and obscenity screening
validation
“we automatically screen every generated message for toxicity, insults, and obscenity”
4
Hallucination pattern check
validation
“we validate generated suggestions against common patterns of hallucinations and automatically discard matches”
5
Accountant reviews suggestion
human_review
“a human accountant will always review and confirm the model's suggested question before it is sent to a client”
6
Question sent to client
output
“The accountants can either confirm the question and send it to their client with just a click, or they can edit the question, but they won't have to type it all from scratch every time”
Reported outcome

Digits' generative ML workflow is live in production, saving accountants time on every client interaction by automatically drafting transaction questions they can confirm with a single click or quickly edit.

Reported metrics
Time saved per client interactionvaluable seconds on each and every client interaction
Accountant time and tediumsave our users both time and tedium
Reported stack
T5TensorFlow TransformGoogle Cloud DataflowTensorFlow Extended (TFX)Vertex AITensorFlow TextTensorFlow ServingHuggingFaceRouge
Source
https://digits.com/developer/posts/assisting-accountants-with-generative-machine-learning/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Digits' generative ML workflow is live in production, saving accountants time on every client interaction by automatically drafting transaction questions they can confirm with a single click or quickly edit.

What tools did this team use?

T5, TensorFlow Transform, Google Cloud Dataflow, TensorFlow Extended (TFX), Vertex AI, TensorFlow Text, TensorFlow Serving, HuggingFace, Rouge.

What results were reported?

Time saved per client interaction: valuable seconds on each and every client interaction; Accountant time and tedium: save our users both time and tedium (source-reported, not independently verified).

How is this finance ops AI workflow structured?

Accountant flags transaction → T5 model generates question → Toxicity and obscenity screening → Hallucination pattern check → Accountant reviews suggestion → Question sent to client.

Related finance ops cases