Framework Principle
Local-First AI
Local-first AI is an approach to artificial intelligence that gives priority to ownership, privacy, continuity, recoverability, and human control over the working environment.
In the Phronesis Intelligence framework, local-first AI is not only a technical preference. It is an authority model: where the work lives, who controls it, how it is preserved, and what governs future use.
Core Idea
Where the work lives affects who governs it.
AI-assisted work can become part of a personal record, business system, publication process, software project, research workflow, or institutional memory.
When that happens, location matters. If the work exists only inside a commercial chat account, scattered export, remote tool, or temporary session, the person using AI may lose visibility into where the work lives, how it is recovered, what is remembered, and who ultimately controls the system of record.
Local-first AI begins from a different posture. It asks what should remain under direct operator control before work is connected outward to external tools, cloud services, APIs, or hosted platforms.
The goal is not isolation for its own sake. The goal is governed control: local authority first, external connection by deliberate choice.
Why Local-First Matters
Local-first is about more than privacy.
Privacy matters, but local-first AI also concerns authority, memory, continuity, inspection, and recovery.
Ownership
Important work should not depend entirely on a remote interface, disappearing session, changing service policy, or unavailable export path.
Privacy
Sensitive material should be handled with clear boundaries before it is shared with any external service or commercial model.
Continuity
Long-term projects need records, source context, decisions, memory, drafts, and artifacts that can survive beyond a single conversation.
Recoverability
A serious system should allow important work to be backed up, restored, inspected, exported, and understood later.
Inspection
Users need ways to inspect what was generated, what was preserved, what source material shaped the work, and what decisions govern it.
Authority
The person or organization responsible for the work should remain in control of what becomes memory, record, artifact, or final use.
Local Authority
A local-first foundation keeps the operator in control.
Local-first does not mean every model, document, or tool must be isolated from the world.
It means the working foundation is governed locally first. The user should know where projects, files, memory, prompts, source material, artifacts, and records live before deciding what should be sent outward.
External services can still be useful. Hosted models, cloud APIs, research tools, publishing platforms, and collaboration systems may all have a role. But the connection should be deliberate, scoped, and governed rather than automatic or invisible.
Local authority gives the operator a stable base from which to decide what can leave, what must stay, and what should be preserved.
Explore Governed IntelligenceLocal-first asks:
- Where does the work live?
- Who controls access?
- How is memory preserved?
- How are artifacts recovered?
- What depends on external services?
- What remains under human authority?
Data and Memory
Memory is more valuable when its location and authority are clear.
AI memory can help preserve context across long projects, but memory also creates responsibility. A system may remember preferences, working assumptions, project facts, drafts, decisions, or source references.
Local-first AI asks whether that memory can be inspected, corrected, scoped, exported, backed up, or removed. It also asks whether the memory is local context, user preference, project record, or approved authority.
Memory without governance can create confusion. Memory with local authority can become part of durable work.
Memory governance should clarify:
- What is remembered
- Why it was remembered
- Where it is stored
- What scope applies
- Who may change it
- When it should be retired
Files and Artifacts
AI-assisted work should not disappear into the chat window.
Documents, reports, summaries, code, diagrams, checklists, outlines, drafts, exports, and decision records become more useful when they are managed as artifacts.
Local-first AI gives those artifacts a durable home. The user should be able to know what was created, when it was created, what version is current, what source material shaped it, and whether it is approved for use.
Without artifact discipline, useful work can become scattered across messages, downloads, browser tabs, cloud folders, and forgotten exports.
A local-first workbench gives generated work somewhere accountable to live.
Explore Source DisciplineArtifact questions:
- What was created?
- Where is it stored?
- What version is current?
- What source material shaped it?
- Who approved it?
- Can it be recovered later?
Cloud and Local Together
Local-first does not mean cloud-never.
The practical question is not whether outside tools are ever useful. The practical question is who governs the connection.
A governed connection model
Start local.
Keep the working record, source context, artifacts, and authority model under direct operator control.
Classify the material.
Decide what is public, private, sensitive, proprietary, personal, temporary, or approved for external use.
Choose the tool deliberately.
Use external services when they are appropriate for the task, risk, and data involved.
Preserve the record.
Keep enough local context to understand what was sent, what returned, and what the human accepted.
Limit authority drift.
Do not let a convenient remote response quietly become local policy, memory, or final decision.
Review before reuse.
Check whether externally assisted work is accurate, appropriate, sourced, and approved for the intended use.
Governance test: Can the operator explain what left the local system, why it left, what returned, and what was accepted?
Continuity
Long-term work needs a system of record.
Many AI tools are excellent for immediate assistance but weaker as durable work environments.
Long-term work may require project organization, document history, source files, accepted decisions, rejected drafts, working notes, generated artifacts, and a clear path for returning to the work later.
Local-first AI supports continuity by treating AI interaction as part of a larger work system rather than a temporary exchange.
That continuity matters when a project spans days, months, years, publications, systems, clients, organizations, or intellectual property.
Read About Practical Wisdom in the MachineContinuity depends on:
- Recoverable records
- Stable project context
- Version awareness
- Source preservation
- Approved decisions
- Human-readable history
FridayLocalAI
FridayLocalAI is the local-first platform expression of the framework.
FridayLocalAI is being developed as a governed local-first AI workbench: a place where conversations, prompts, artifacts, memory, records, models, source context, and human authority can be organized around serious work.
The platform effort reflects a simple premise: AI becomes more useful when the surrounding system helps people preserve control, inspect the work, recover context, and decide what becomes authoritative.
Phronesis Intelligence explains the framework. FridayLocalAI is the product and platform destination for the local-first workbench.
Visit FridayLocalAI.comWorkbench focus
Local control keeps authority close to the operator.
Governed memory makes context more accountable.
Artifacts give generated work a durable home.
Traceability helps users understand how work developed.
Human judgment remains the final authority.
Continue the Framework
Local-first AI connects control to responsibility.
Practical wisdom asks what should be done.
Human authority decides who is responsible.
Source discipline checks what supports the work.
Governed intelligence structures memory, prompts, artifacts, decisions, and records.
Continue through the framework
Return to FrameworkLocal-first AI keeps serious work closer to human authority.
When ownership, memory, artifacts, recovery, and source discipline matter, the working system should begin from a governed local foundation.