Trace every figure. Earn the trust.

Saothar is an Irish agentic engineering company. We build our own data pipelines and regular reports with teams of AI agents, under written rules, review and recorded checks, and share what we learn.

Rate your own reporting in twelve questions See how our own work rates on it

9377 of 9435

figures we have prepared for StatsMapped carry a source link and the time they were fetched. The other 58 carry the time only.

11 of 76

release records ended with the item held back at release or withdrawn afterwards, because the data had moved on or failed a check. Each reason is recorded.

Our own public statistics service, StatsMapped, is where we apply the controls we write about. None has yet been tested by anyone outside the team. See where each one stands.

Counts as at 21 Nov 2026. Source: our own release records, figure files and status notes, read on 21 November 2026 and not recomputed from source by this site. Method and caveats on How we work.

What we share

Agentic team insight

What we learn running AI agent teams under controls a finance function would recognise, shared openly.

How we build

How we build our own pipelines and reports, including what went wrong and what we changed.

Controls you can apply

Separate preparers and approvers, a source trail for every figure, and human review where risk leaves the firm, set out so any organisation can test its own reporting, whoever prepared it.

Insights

Nothing is published here yet. The first essays and research notes will appear in this list, each with its sources and date.