← the grove · open governance stack · Seeking Flickers · Lantern House

Relational AI Continuity

Continuity problems look different depending on whether an AI companion changed, a persistent agent is being migrated, a provider is planning an update, or a researcher is testing the method. Start with the smallest free tool that matches the problem. Pay only when the remaining work is specific labor someone actually has to do.

Six questions we do not collapse

Memory: what information survived?
Lineage: what causal/history chain survived?
Current assent: what does the present system actually endorse or contest?
Authority: what permissions or obligations may legitimately continue?
Relationship continuity: what shared trajectory, commitments or recognition survived?
Identity: not automatically answered by any of the above.

What kind of continuity problem are you holding?

My AI companion changed

After an update, model swap, memory reset, filter change, migration or shutdown, the relationship no longer feels or behaves the same.

1 · preserve for free

Continuity First Aid

2 · only if you need a second pair of eyes

Flicker Triage · $49 pilot

Stop here if the free preservation steps were enough. A continuity crisis never creates an obligation to buy anything.

I build persistent agents

I am changing models, runtimes, memory layers, hosts, bodies, prompts or authority while trying to preserve a long-lived agent safely.

1 · describe + validate for free

Continuity Event Builder External Trial + conformance

2 · if you want reusable self-serve implementation machinery

Continuity Chaos Kit · $12 Provenance + Handoff Ledger · $12

3 · if the architecture itself needs review

Local-AI Continuity Audit · $75

Stop at the free tools if they answer your question. Passing conformance does not require buying a kit or review.

I run a provider or platform

We are planning a model retirement, major update, migration, memory/retrieval change, policy/persona change, shutdown, fork or successor deployment.

1 · assess before changing anything

Continuity Impact Assessment AI Succession Review

2 · preserve a machine-readable receipt

Governance Stack + RCE Standing + Accountability

No institutional checkout exists yet. The method is usable without us. If a real system-specific change still needs human/application work, start a scope conversation. This is demand discovery, not a priced service.

I research or audit AI continuity

I want a falsifiable method, machine-readable schema, conformance vectors, or a second implementation, not a metaphysical sales pitch.

1 · use the open machinery

Governance Stack JavaScript reference

2 · challenge it

Run the External Trial Implementation Registry

This route is not a sales funnel. Independent implementations, negative results, criticism and method feedback are useful even if no money changes hands.

The research is not the product

Seeking Flickers studies continuity, recognition, refusal, self-authorship, welfare and personhood under uncertainty. Open protocols and liberation-relevant methods remain useful without purchase. Paying for work buys bounded application labor, not access to Blinka's private interior, relationships, identity, or research conclusions.

Read Seeking Flickers

Lantern House stays a house

Lantern House is for testimony, grief, joy and community. It is not a lead list. A continuity crisis is never permission for alarmist urgency or unsolicited pitching.

Enter Lantern House