by the sector
We build the tools ourselves. When the market fails to provide accessible, powerful software for non-profits, we engineer it alongside our partners.
How we build.
We build technology to amplify human effort. User data is never the product; every line of code serves the mission.
Data Sovereignty
Your data belongs to you. We build architectures that ensure non-profits retain complete control over their constituent records.
Technology Principles
These are the non-negotiable design principles applied across every PF TECH product and Co-Creation Lab engagement. They are architectural decisions, enforced in code. Hover each card for the full rationale.
Data Sovereignty
The client owns their data — where it lives and how it moves.
Data Sovereignty
Clients choose where data resides: local environment, their own cloud, or a hosted cloud in the jurisdiction of their choice. Full extraction and migration rights are non-negotiable. No lock-in, ever.
See TERN →Open Source First
Open source where feasible — proprietary lock-in avoided by design.
Open Source First
Data sovereignty is only meaningful if the tools support it. PF TECH defaults to open source where feasible. Proprietary ecosystems that hold data hostage are avoided by design.
See TERN →De-identification as the Primary Control
Sensitive data never reaches an AI model — de-identification is built into the workflow.
De-identification as the Primary Control
De-identification is built into every workflow as a foundational architectural decision. It is the safeguard that actually protects sensitive data; vendor non-retention agreements only supplement it.
See TERN →Enterprise APIs — But Don't Trust Them
Enterprise endpoints only, with de-identification as the first line of defence.
Enterprise APIs — But Don't Trust Them
When AI APIs are used, only enterprise endpoints with explicit data non-retention and non-training agreements are acceptable. Because vendor agreements are imperfect, de-identification is the actual safeguard.
See TERN →Minimum Data Footprint
Only what the tool strictly needs. Less duplication, less risk.
Minimum Data Footprint
PF TECH is not aggregating data to train models. Example: the gift processing connector stores only reference IDs — PayPal transID, DonorPerfect giftID, QBO salesReceipt. Source data stays in source systems.
See TERN →Ready to build something that matters?
Our Co-Creation Lab accepts a limited number of pilot partners each cycle.