Skip to content
Julius BongCPA

AI + Finance

AI should not replace accounting discipline. It should amplify it.

I use AI every week — to analyze, to draft, and to build the tools my team uses. I also spent years as an Assistant Controller, which is why I think about it the way I think about any new process: what could go wrong, who reviews it, and how would we know.

AWhat does not change

An AI-enabled finance function still requires…

  1. 01

    Accounting knowledge

    AI can draft a journal entry. Someone still has to know whether it is the right one.

  2. 02

    Controls

    Automated preparation, human approval. Segregation of duties does not disappear because the preparer is software.

  3. 03

    Data quality

    A model is only as reliable as the ledger and source data beneath it. Reconcile first.

  4. 04

    Governance

    Know which data goes where, what is retained, and who is accountable for the output.

  5. 05

    Financial judgment

    AI is good at finding patterns. Deciding which ones matter to the business is still a finance job.

  6. 06

    Human review

    Every AI-assisted number that leaves finance has a named person who reviewed it.

BWhere it helps

Where I apply AI — and the guardrail on each.

AI-assisted FP&A

Analyzing trends, testing assumptions, and working through scenarios and drivers faster — so more of the cycle goes to the conversation about what the numbers mean.

GuardrailEvery figure traces to the ledger or a stated assumption.

Automated management reporting

Reducing the manual assembly of monthly reporting, and drafting first-pass variance commentary for analysts to refine.

GuardrailCommentary is drafted by AI and owned by a person.

AI in accounting

Support for reconciliations, account analysis, policy research, transaction review, and explaining financial reports to non-finance readers.

GuardrailAssists the preparer; never approves.

AI agents in finance

Agents that gather information across systems, run defined analyses, and handle repetitive steps of a finance workflow.

GuardrailNarrow permissions, logged actions, clear hand-off points.

Finance workflow automation

Combining AI with workflow tools such as n8n and system APIs to route documents, extract information, and move work between steps.

GuardrailExceptions surface to a person instead of failing silently.

AI-assisted development

Building internal finance applications — planning, reporting, and analysis tools — far faster than a traditional development cycle would allow.

GuardrailValidated against known results before anyone relies on them.

CCurrently exploring

What I am working on next.

  • Agentic finance workflows

    Where multi-step agents can safely take on close and reporting tasks, and where they should stop and ask.

  • LLM workflow design

    Structuring prompts, context, and review steps so that output is repeatable enough to rely on.

  • AI governance in finance

    Practical control frameworks for AI-assisted preparation, review, and audit evidence.

  • AI-assisted forecasting

    Combining statistical models with language models to explain forecast movements, not only produce them.

  • n8n automation

    Connecting finance systems through workflows that are visible and maintainable by the finance team.

  • Modern finance analytics

    Cloud data platforms and semantic models that let finance own its metric definitions.

Contact

Modernizing finance without loosening what makes it right?

I welcome conversations with finance leaders, executive recruiters, and organizations working on accounting, FP&A, and finance transformation.