Shiftby.pro

Independent fieldwork publication

Work thatstays connected.

Practical fieldwork on how AI-assisted research, decisions, evidence and systems carry understanding forward as the work changes.

Explore Fieldwork →Explore ProjectsAbout Shiftby.pro

What is Shiftby.pro?

Shiftby.pro is an independent fieldwork publication by Ananda Krishna Marri about AI-assisted knowledge continuity, semantic content engineering and evidence-bounded systems. Its current project, Inspiral, explores how sources, decisions, context, limitations and human authority can survive between continuing AI-assisted tasks.

AI-assisted knowledge continuitySemantic content engineeringEvidence-bounded systems

Founding fieldwork

View all fieldwork →
Isolated generation
Prompt↓Output
Continuing work
SourcesDecisionsEvidenceCurrent stateLimitations↓Current task↓Output↓Retained learning
/ new state

analysis • inspiral

Why Content Generation Was Not the Hard Problem

Why faster AI content generation did not solve ShiftBy's harder problem: preserving evidence, decisions, current state and context across continuing AI-assisted work.

Knowledge & Context Engineering

Read founding analysis

Continuing publication

Latest fieldwork

Explore all fieldwork →
  1. technical note · inspiral · 3 Sept 2026

    What ContextCore Currently Contains

    A bounded fieldwork note on what ContextCore currently makes inspectable, and what its evidence does not establish.

    Read What ContextCore Currently Contains →
  2. case study · inspiral · 2 Sept 2026

    How OpenAI Supports Inspiral Fieldwork

    The concrete roles ChatGPT Work, connected context and Codex play in Inspiral fieldwork—and the human authority they do not replace.

    Read How OpenAI Supports Inspiral Fieldwork →
  3. method · inspiral · 1 Sept 2026

    What Must Survive Between AI-Assisted Tasks

    A lightweight handoff for recovering the evidence, decisions, current state, limitations and authority a later AI-assisted task needs to reuse work safely.

    Read What Must Survive Between AI-Assisted Tasks →

Work in progress

01 / Exploration • Active

Inspiral

Exploring continuity of governed knowledge, context and meaning across AI-assisted work.

Current state: Active
02 / Practice • In progress

AI Security Assurance

Repository-grounded experiments in assurance, testing and accountable AI engineering.

Current state: In progress
How this publication works

Evidence before
certainty.

Current state is distinguished from historical rationale. Evidence strength is made visible. Limitations are retained. AI assists; humans retain publication authority.

01

Evidence

From sources, code, runs and decisions.

02

Bounded claims

What the evidence supports—and what it doesn’t.

03

Fieldwork

Context, decisions, trade-offs and intent in the work.

04

Public learning

Published for others to build on, challenge and extend.