Source-linked or it doesn't ship
Every accepted answer has to trace back to evidence you can check.
The firm
Amidflow is built for organisations that want AI to work on real business knowledge — not just demo data. It combines practical experience across data strategy, BI, analytics engineering, governance, semantic modelling and decision intelligence.
It's also early and founder-led — and the site says so plainly. There are no client logos to parade, no manufactured case studies and no invented numbers here. What there is: senior, hands-on experience with the unglamorous layer that has to exist before AI can be trusted, and a contained way to prove it on one workflow before anything scales.
A semantic layer over scattered knowledge — governed, owned and traceable. That discipline is the whole point: it's what turns a confident-sounding AI answer into one a leader can actually defend.
What founder-led means here
Hands-on experience building reporting, analytics and data capabilities in complex, high-volume operating environments — not slideware.
A working understanding of definitions, ownership, lineage, quality controls and the executive accountability behind a number.
Focus on the infrastructure and governance that has to exist before AI agents can be trusted — not the demo, the foundation under it.
The goal isn't more dashboards or another chat interface. It's better, faster, more traceable decisions leaders can stand behind.
Founder-led discovery, design and delivery for early clients. No pitch by a partner and a handoff to juniors — the same person does the work.
The founder
Pasi Abeyweera · Founder
Pairing hands-on data, BI and governance delivery with a behavioural lens on how organisations actually make decisions.
Built machine learning and analytics across 250+ retail sites at Maxol on Azure Databricks, advised on commercial strategy at Stax, and holds an MSc in Behavioural Economics from University College Dublin. The work connects strategy to architecture, data engineering, semantic modelling, analytics, AI workflows and governance — and to the behavioural reality of how decisions get made.
How we work
Every accepted answer has to trace back to evidence you can check.
Definitions, ownership and controls come before AI features.
One corpus, one workflow, one success test — before anything scales.
Clients keep the model, the findings and the documentation. No black box.
Experience areas
Entity
| Company | Amidflow Limited |
| Registered | Ireland |
| Contact | hello@amidflow.com |
Company registration number and registered address on the imprint.
Start
Bring one workflow, corpus or reporting problem where scattered knowledge is blocking AI or a decision. In 30 minutes we'll know whether it's worth a pilot — senior, from the first call.
No deck, no obligation.