The brief
Eagl’s pitch is simple to say and hard to do: catch the financial errors that humans miss before they make it into the month-end close. Mis-categorised expenses, duplicated transactions, accruals that should have reversed, variances that should have raised flags but did not.
The team had a working prototype, real customer interest, and a clear product vision. They needed senior engineering bandwidth that could plug in fast, ship safely against a moving target, and not need to be told twice how a general ledger reconciles.
Where we fit
Helsh-js works as part of Eagl’s engineering team on a long-running freelance contract. Day to day this looks like:
- Building and maintaining integrations with ERP systems (currently Exact Online and NetSuite, with more in the roadmap).
- Designing data pipelines that pull ledger entries, classify them, run them through detection rules and LLM-assisted reasoning, and surface anomalies in the dashboard.
- Building the actual product surface: the UI a CFO opens at 7am to see “yesterday’s close caught 14 things you’ll want to look at.”
- Pairing closely with Eagl’s in-house team. We are deliberately not a black box.
Why this matters
Accounting software has been chasing the same dream for thirty years: no more errors, no more month-end. What has changed is that modern LLMs can finally reason about why a transaction looks wrong, not just whether it matches a pattern. That changes the product. We get to build something the previous generation of accounting software could not, and that is genuinely worth showing up for.
What we like about the engagement
Eagl trusts us to make real engineering decisions, not just execute tickets. They are tough on quality and kind on delivery, which is the only combination that makes long-form contract work fun.