Provenance Manifesto
arrow_back Back to Blog

Article artifact

The Provenance Manifesto

Source: artifacts/provenance-slides.md

Table of contents

expand_more

The Provenance Manifesto

10-Minute Meetup Version

Topic: Decision provenance in the age of AI
Format: Compressed Markdown deck
Goal: Explain the idea, value, and principles in a short live talk


Slide 1

The Provenance Manifesto

Decision provenance in the age of AI

Addressing organizational context amnesia through the art of provenance.

"The history of every decision is the architecture of our future."


Slide 2

About Me

My name is Yauheni Kurbayeu.

  • software engineering leader with decades of experience building and operating complex systems
  • experienced in managing large engineering teams
  • worked on enterprise platforms, large-scale SaaS products, and distributed engineering organizations
  • increasingly focused not only on how systems are built, but on how the decisions behind them are made, preserved, and evolved

Provenance is an exploration of that idea.


Slide 3

The Problem

Software teams preserve:

  • code
  • tickets
  • deployments
  • infrastructure
  • documentation

But they rarely preserve:

  • why a decision was made
  • which trade-offs were accepted
  • which assumptions shaped it
  • who owned it

We keep the artifact. We lose the reasoning.


Slide 4

Why This Matters Now

For a long time, this was a delivery problem.

Now AI is turning it into a governance problem.

AI is making the "what" cheaper:

  • code
  • plans
  • architectures
  • outputs

So the "why" becomes more valuable:

  • intent
  • constraints
  • accountability
  • decision ownership

AI accelerates execution. Provenance preserves accountability.


Slide 5

The Core Idea

Decisions should become first-class artifacts.

A meaningful decision should preserve:

  • the problem
  • the context
  • the assumptions
  • the alternatives
  • the reasoning
  • the owner
  • the later evolution

Decision provenance explains why a system exists in its current form.


Slide 6

What Provenance Changes

It moves us:

  • from undocumented intuition to traceable decisions
  • from private memory to institutional memory
  • from hidden assumptions to transparent reasoning
  • from static documentation to evolving decision history
  • from uncontrolled automation to governed human-AI collaboration

This is not just better documentation. It is a memory layer for reasoning.


Slide 7

The Principles

  1. Decisions are first-class artifacts
  2. Decisions must carry context
  3. Decisions evolve but are never erased
  4. Decisions must be queryable
  5. Decisions must be attributable
  6. AI must operate within decision governance
  7. Institutional memory is a strategic asset

Memory compounds.


Slide 8

The Value

Decision provenance helps organizations:

  • reduce rework and repeated rediscovery
  • onboard people faster
  • make architectural change safer
  • expose hidden assumptions
  • improve incident, audit, and compliance reasoning
  • give AI access to the reasoning layer, not only the artifact layer

The more AI generates, the more provenance matters.


Slide 9

What This Looks Like In Practice

Start by preserving major decisions around:

  • architecture trade-offs
  • scope and product choices
  • incident responses
  • policy and risk boundaries
  • AI-assisted implementation plans

For each important decision, capture:

  • why it was made
  • what alternatives were considered
  • what assumptions and risks existed
  • who owns it
  • how it changed over time

Slide 10

The Real Shift

The hardest part is not technical.

It is cultural.

Provenance is a mindset shift:

  • decisions should not disappear into conversations
  • reasoning should not remain private
  • AI should not operate outside visible governance

Provenance is not primarily about tools. It is about how organizations remember.


Slide 11

Closing

The real divide is not:

  • human-written vs AI-generated

The real divide is:

  • opaque outputs vs traceable reasoning

If we cannot reconstruct how a conclusion was formed, then even a polished result is still just a well-presented guess.

Discussion

  • Where does your organization lose decision context today?
  • Which decisions are most worth preserving?
  • How should AI participate in governed organizational memory?