Skip to main content
Mission Multipliedby PF TECH

The AI-Ready Non-Profit Back Office · Part 01

TERN: Building the Non-Profit Back-Office Infrastructure the Sector Never Had

Twenty years of working inside non-profits left me watching extraordinarily capable organizations spend their best operational energy fighting infrastructure that was never designed for them. TERN is our answer to that — a non-profit back-office integration framework built from lived experience. Here's what we're building, why it doesn't exist yet, and how you can help shape it.

Greg Zatulovsky· Founder & CEO, PF TECH·· 9 min read
A dawn landscape across an under-served valley with new infrastructure being gently laid — bridges, light-strung paths, glowing nodes — under warm morning light. The terrain is not hostile, just waiting.
A dawn landscape across an under-served valley with new infrastructure being gently laid — bridges, light-strung paths, glowing nodes — under warm morning light. The terrain is not hostile, just waiting.

I've spent twenty years watching extraordinarily capable organizations — mission-driven, well-led, deeply committed to their work — spend their best operational energy wrestling with infrastructure that was never designed for them. Donation data that reconciles differently on every side of the ledger. Restricted funds governed by spreadsheets and good intentions. Grant reports rebuilt from scratch each cycle because the data lives in five different places. Budgets that exist as email attachments until the annual process starts, at which point they become email attachments again.

This is not a technology failure. The sector has been excluded from purpose-built infrastructure for so long that the absence has been normalized as the default. TERN is our direct answer to that.

01·Chapter

What TERN Is — and What It Isn't

Define the thing before you build it.

Chapter 01 of 04

Skip chapter intro

TERN is PF TECH's non-profit back-office integration framework. It is not a traditional SaaS platform with a row of modules you click between. It is being built as an intelligent, agent-based system — domain-specific capabilities surfaced through a chat interface, with embedded components and popouts where the workflow requires them. The architecture will follow wherever the technology lands. What will not change is the goal: a coherent, sovereign back-office ecosystem built specifically for the charitable sector.

I want to be direct about something: TERN is in active development. We do not have a finished product. We do not have a public user base. We have twenty years of operational knowledge, a clear engineering direction, a growing portfolio of Strategic Partner organizations co-creating the roadmap with us, and a conviction that the sector deserves purpose-built infrastructure that no general-purpose platform will ever prioritize building.

What we have is the proof of concept — and the proof is in the numbers.

What the interim proved

83%

Processing time reduction

Gift reconciliation for a 50–75 donation/month organization dropped from ~24 hours to under 4 hours monthly via an interim low-code workflow. TERN targets near-zero.

~$0

Marginal cost of AI-assisted engineering

The cost barrier that kept purpose-built non-profit infrastructure out of reach for decades has collapsed. The knowledge barrier — 20 years of sector experience — has not.

The 83% reduction shown above is not a TERN result. It is what happened when I built an interim low-code automation for gift processing at one of our partner organizations — moving from a fully manual process to a Power Automate workflow. Eighty-three percent reduction in monthly processing time, from roughly 24 hours to under 4 hours, for 50 to 75 donations per month. The workflow was brittle. It required manual maintenance. It was not scalable. But the reduction in processing time was real and immediate, and it told us something important: the problem is not complex. It is just unsolved.

The problem is not complex. It is just unsolved.

From the 83% field result, 2025

TERN's gift processing and reconciliation capability will take that interim result toward near-zero. Automated sync from payment processor to CRM to QuickBooks Online, coded as fund-restricted receipts with internal controls baked in at the architecture level — not bolted on afterward.

20 yrs
Non-profit finance practice

The sector-specific domain knowledge layered into every TERN workflow — two decades of audits, grant reconciliations, fund management, and governance practice.

02·Chapter

The Three Capabilities We're Building

Direct translations of problems already solved inside real organizations.

Chapter 02 of 04

Skip chapter intro

TERN is being developed around three core capabilities that address the most persistent operational failures I have seen repeat themselves across every non-profit I've worked with. Each capability is a direct translation of something I have already built manually or with low-code tools inside real organizations.

Claymation scene of gift processing automation — data flowing from a clay payment processor through a CRM to QuickBooks, with an Analytical Squirrel visibly relieved at a clean desk and a Helpful Circuit Robot assisting from the corner.
Capability 01

Gift Processing & Reconciliation

Addresses the most painful data fragmentation in the sector: the same donation described differently by the payment processor, the donor database, and the accounting system. Every non-profit using a fundraising platform and an accounting system has this problem. TERN automates the sync, enforces proper fund restriction coding, and builds the audit trail that funders increasingly require — without any data passing through an AI model without de-identification first.

Claymation scene of a round clay boardroom table with six animal personas collaborating on a budget — Determined Owl as Board Chair, Wise Tortoise as Treasurer, Resourceful Raccoon, Analytical Squirrel, Caring Beaver, Focused Ant — with data streams flowing into a consolidated budget that publishes to QuickBooks.
Capability 02

Collaborative Budgeting

Addresses the annual process that somehow still runs on spreadsheets passed around by email. We are currently building a conversational budget agent that engages the right stakeholders at the right moments: the Executive Director for strategic context, the Finance Committee Chair for governance priorities, the fundraising director for revenue targets, department heads for operating expenses, HR for staffing projections. The agent consolidates inputs, surfaces conflicts, and — once the designated financial authority approves — publishes the final budget directly to QuickBooks Online. This is not a form-filling tool. It is an agent that understands non-profit budget structures, fund accounting logic, and the governance relationships that determine how decisions get made in a charitable organization.

Claymation scene of a transformed clay filing system — restricted funds as colour-coded containers with grant names, Analytical Squirrel presenting a compliance dashboard to a Determined Owl (Board Chair), Wise Tortoise (Treasurer), and Collaborative Goose (institutional funder).
Capability 03

Fund & Grant Management

Addresses the single greatest source of compliance anxiety for non-profit finance teams: restricted funds, grant agreements, and the gap between what development committed and what finance is accountable for. Grant compliance tracking, labour allocation to programs, multi-funder reporting — all governed by a verifiable source of truth rather than a binder someone updates quarterly. The budgeting capability and the fund management capability are designed to be deeply integrated: once a budget is published to QBO, grant agreements are reviewed against it for compliance analysis and reporting.

03·Chapter

Architecture Principles That Cannot Be Compromised

Designed by someone who has lived inside the compliance anxiety these concerns create.

Chapter 03 of 04

Skip chapter intro

The sector's relationship with technology is complicated by an entirely legitimate set of concerns. Donor data. Service user data. Volunteer records. These are sensitive, often legally protected, and increasingly scrutinized by funders, auditors, and regulators operating under PIPEDA and provincial privacy frameworks.

TERN's architecture was designed by someone who has lived inside the compliance anxiety these concerns create. Every principle below is non-negotiable.

Claymation scene of the de-identification architecture: sensitive donor data held secure on the left behind a translucent wall, abstract reference codes flowing to a small helpful robot processing them in the middle, authorized re-identification happening in a sealed room on the right.
Privacy as architecture — a physical barrier built into the system itself.
Principle 01

De-identification first

Only de-identified data ever passes to an AI model. TERN does not know who your donors are. It does not know who your service users are.

How it holds

  1. 1.Reference IDs are all that flows to AI processing
  2. 2.Re-identification only in secure, controlled environments you authorize
  3. 3.Not a feature — it is the architecture itself
Principle 02

Data sovereignty by design

You choose where your data resides — local infrastructure, your cloud, hosted cloud, Canadian data centres. Migration is always possible.

How it holds

  1. 1.Open-source infrastructure — sovereignty is technically enforced
  2. 2.You can extract and migrate at any time
  3. 3.Lock-in is structurally prevented, not contractually promised
Principle 03

Enterprise API only

We work exclusively with AI providers that offer explicit non-retention and non-training agreements on API-accessed data.

How it holds

  1. 1.Only enterprise endpoints — never consumer products
  2. 2.Non-retention and non-training contractually guaranteed
  3. 3.The vendor layer must be one you can trust — de-identification remains the real safeguard
Principle 04

Minimum data footprint

TERN stores only reference IDs. The actual data lives in the source systems that already hold it.

How it holds

  1. 1.PayPal transaction ID = DonorPerfect gift ID = QBO receipt number
  2. 2.All operations happen via API calls
  3. 3.We do not duplicate what does not need to be duplicated
Principle 05

Built by a CPA

Internal controls, audit trails, and fund accounting logic — baked into every workflow.

How it holds

  1. 1.Not researched. Not approximated.
  2. 2.Drawn from twenty years of solving these problems inside real organizations
  3. 3.Designed by someone who has lived inside the compliance requirements
04·Chapter

Why We're Building This in Public

Built with the sector, not for it.

Chapter 04 of 04

Skip chapter intro

There is a deliberate choice embedded in how we're approaching TERN's development: we are building it with the sector, not for it.

Strategic Partners

Our Strategic Partners — a boutique portfolio of Canadian non-profits working with us in an embedded operational capacity — are co-authors of the system. Their real operational challenges surface what is worth building. Their workflows validate what we have built. Their results become the proof points that demonstrate the approach.

Mission Multiplier Program

The Mission Multiplier Program extends this to practitioners. MMP participants are the first outside audience to access TERN capabilities as they become available — part testing community, part learning cohort, part product development partner. If you are a finance professional or executive leader in the sector who wants to understand where this technology is going and have a voice in shaping it, the MMP is the fastest path in.

Co-Creation Lab

And if your organization has a specific operational problem that you believe is worth building a solution for — something that doesn't fit the standard product roadmap but would transform how you operate — the Co-Creation Lab is how we work together on bespoke development.

Follow the build

Monthly notes from inside the work

What we're building next, what the field is telling us, what's working across the cohort. Unsubscribe anytime.

By subscribing you join the PF TECH mailing list to receive Mission Multiplied posts and occasional related PF TECH updates. Monthly cadence. Unsubscribe anytime from any email. See our privacy policy at purposeforwardtech.com/privacy for how we handle your data.

Privacy policy

Friction Point Submission

The lowest-barrier option of all: tell us what is breaking. We have a friction point submission form for exactly this purpose. No cost, no commitment. Just tell us where your operations are failing, and we will use that signal to inform what we build next.

We are building this with the sector, not for it. And we would genuinely like your help making it better.

Help build what the sector needs

Join the Mission Multiplier Program to be first in line as capabilities launch. Bring a specific problem to the Co-Creation Lab for bespoke development. Or simply tell us where your operations are breaking — friction point submissions are free, no commitment required.

How did this land?

Greg Zatulovsky

About the author

Greg Zatulovsky

Founder & CEO, PF TECH

Greg founded PF TECH to multiply the operational capacity of purpose-driven organizations. CPA with fifteen-plus years in non-profit finance, operations, and technology. Writes from inside the work — practitioner voice, not pitch deck.

More reading

A weathered steel guardrail running along a mountain pass cliff edge at golden-hour dawn, with the words CONTROLS BEFORE CAPABILITY rendered in bold caps across the darkened lower band.
AI automationguardrailsnon-profit finance

Your Accounting System Needs a Bouncer, Not a Butler

Why the automation story needs a second chapter

There's a version of the AI-in-non-profits story being told right now that goes like this: AI will automate the tedious back-office work, free up your staff, and let you focus on your mission. That version is true — I believe it deeply and I'm building the tools to make it real. But there's a part of the story that isn't being told, and the gap between those two versions is where a lot of organizations are going to get hurt.

· 7 min read
A line of wind-bent pines on a rocky shoreline at dawn, shaped by years of storms — with "Resilience is a verb." rendered in serif display over the soft left third.
Non-Profit LeadershipAI GovernanceFinancial Resilience

Resilience is a Verb: What the CPA Ontario Not-for-Profit Conference Revealed

Field Note · CPA Ontario NFP Conference 2025

In November 2025 I hosted the CPA Ontario Not-for-Profit Conference. The theme was Resilient Futures. Seven speakers, eight hours, topics ranging from AI adoption to board governance to front-line service delivery. What I took away was more urgent than the theme: the sector's biggest obstacle to building resilience is not funding or technical skills. It is the calculus that makes staying stuck feel safer than taking the next step.

· 6 min read
A traditional Canadian stone-and-timber watermill at dawn beside a clear running river, mid-restoration — new cedar fitted beside weathered beams, scaffolding still in place, hand tools on a workbench, deep forest rising behind.
PF TECHNon-Profit TechOrigin Story

Building a Better Engine: The PF TECH Origin Story

Origin Story · PF TECH

PF TECH didn't start at a whiteboard. It started in a filing cabinet — in 2013, watching a colleague print every email he received and file the paper copies in rows of cabinets along the wall. Nearly two decades later, that pattern is still the sector's central problem. Here's what I built because of it.

· 12 min read