State

Change AI. Don’t start again.

  1. A request becomes work.

    A manager asks Claude to prepare ABC Limited’s management accounts. Kroy records the request as a piece of work.

  2. The work moves.

    Working. Waiting for a document. Approved. Each change is state in Kroy, not a line buried in a chat.

  3. Change AI halfway through.

    Claude steps away. ChatGPT picks up — reading the same decisions, tasks and drafts.

  4. The state remains.

    Nothing is re-explained. Kroy owns the work state, so no AI conversation has to.

  5. Done — with provenance.

    Every step records who did it, which AI, and what it was based on.

Kroy State
REQUEST
WORKING
WAITING
APPROVED
DONE
Claude
ChatGPT
Priya · Claude
Priya · ChatGPT
Kroy State
REQUEST
WORKING
WAITING
APPROVED
DONE
Claude
ChatGPT
Priya · Claude
Priya · ChatGPT
A piece of work moving through states while the AI changes. A request becomes work.. The work moves.. Change AI halfway through.. The state remains.. Done — with provenance..

The problem

Every AI conversation starts from nothing. You explain the client, the project and what was decided last week — then you switch tools and explain it all again. Work done in one AI is invisible to the next.

What Kroy State holds

  • Decisions — what was decided, by whom, and why.
  • Projects and tasks — what is in progress, blocked or done.
  • Artifacts — the drafts, reports and analyses AI has produced.
  • Relationships — how the work connects to clients, matters and systems.
  • Provenance — why Kroy believes each fact, and where it came from.

How it works across AI

Start an analysis in Claude and record the decision in Kroy. Continue the same project in ChatGPT: it reads the current state rather than a pasted summary. Let Codex pick up the implementation from the same tasks and decisions.

Each AI sees only the state its user — and the delegation to that agent — permits.

Why independent state matters

If your organisation’s working memory lives inside one AI provider, changing provider means losing it. Kroy keeps state independent, so you can use whichever AI suits each task.

Case Studies using State

Accountancy

Onboarding a new client with AI — while the AML review stays internal

Claude sets up the client record, drafts the engagement letter and document request, and prepares the AML checklist for a person. Only the practice’s nominated officer can complete the AML review, and the client never sees it.

ClaudeOutlookSharePointXero

Accountancy

Investigating a client’s question with ChatGPT, Xero and Outlook

ChatGPT gathers the relevant Xero data and correspondence for one client, explains the answer to the manager and drafts a reply in Outlook for the manager to send.

ChatGPTXeroOutlook

Technology

Claude Code and Codex working from shared state

Claude Code records a task’s progress and design decisions in Kroy State. Codex picks up the task from that state on another developer’s machine. Both open pull requests; neither can merge to the main branch or read production secrets.

Claude CodeCodexGitHubJira

Consulting

Shared project state across a consulting team’s AI tools

Claude, ChatGPT and Copilot read and update one project State in Kroy. Work handed from one consultant’s AI to another carries its context. Changing a client-agreed decision needs the engagement lead’s approval.

ClaudeChatGPTMicrosoft CopilotSharePoint

Legal

Handing over a legal matter between people and their AI

Claude prepares a handover note from the matter’s state. After a partner reassigns the matter, the new fee earner’s ChatGPT reads the same state and continues. Neither AI can reassign the matter.

ClaudeChatGPTSharePointOutlook

Accountancy

Partner approval for AI-prepared work

Claude prepares the work and requests approval. The partner reviews it, with their own AI summarising what is waiting, and approves in person. No agent can approve, and nobody can approve their own work.

ClaudeMicrosoft CopilotXeroSharePoint