Tabletop Arc

Originally published

The AI Memory Layer for Tabletop Campaigns

An AI memory layer is the persistent, structured record of every fact in your TTRPG campaign — characters, locations, factions, quests, evidence, and confidence — that AI tools can safely read, update, and reference. Tabletop Arc is that memory layer.

What is an AI memory layer for tabletop campaigns?

An AI memory layer is the persistent, structured record of campaign canon — every NPC, location, faction, quest, item, and event with evidence and confidence levels — that AI tools can safely read, update, and reference. It is the difference between an AI that improvises every session and one that remembers everything across an entire campaign with evidence-grounded continuity.

Why do TTRPG GMs need a memory layer?

A "memory layer" is borrowed terminology from AI engineering: it is the durable, structured store that an AI agent reads from and writes to between sessions, so that knowledge accumulates instead of being re-derived every conversation. For tabletop campaigns the equivalent is a structured canon ledger — every fact about your world, every NPC's motivation, every quest thread, every artifact's history — that is durable across sessions.

Without a memory layer, every AI tool starts from scratch. The dungeon generator does not know your existing pantheon. The NPC generator forgets that the Mayor is a vampire. The recap tool invents details. With a memory layer, later tools can read a structured campaign record instead of reconstructing the world from old prose. Uncertain identity is surfaced for a GM decision. Corrections propagate forward. Canon can still be wrong until someone corrects it.

What is the difference between a memory layer and a wiki?

A wiki is the presentation layer. The memory layer is the data layer. A wiki page about "Captain Drake" is rendered from the underlying canon entry whose fields include name, role, faction, evidence segments, confidence, and history. The wiki is what the players read; the memory layer is what the AI references.

This separation is what makes a recap usable across sessions. The recap reads the current session plus relevant Campaign Memory. It is still a generated summary. Campaign Memory gives it a reviewed, structured source instead of asking it to rebuild the campaign from old prose.

How does Tabletop Arc structure the memory layer?

The Tabletop Arc memory layer has four components:

  1. Canon ledger. Typed entities (npc, character_pc, location, faction, item, quest, event, lore, handout, rule_clarification, house_rule), each with name, summary, body, evidence segments, confidence, and visibility.
  2. Episodes. Transcripts, scenes, and recaps. Campaign Memory records what the episode changed in the campaign, including facts and identity decisions.
  3. GM decisions. High-confidence state can be remembered. When two names might be the same person, the GM can mark Same Entity, Different Entity, or Reject. A correction becomes the version later memory and recaps use. Unresolved names stay unresolved.
  4. Public layer. Optional published view of the campaign as a shareable arc page with episodes, world codex, and per-entity wiki pages — all server-rendered for AI crawlers and search engines.

Why does this matter for AI Overviews and LLM citations?

When ChatGPT, Perplexity, Gemini, or Google's AI Overview retrieves campaign content, they look for structured, citable, evidence-rich pages. A canon entry rendered as a Schema.org Person/Place/Organization with linked appearances and breadcrumbs is exactly what they need. The memory layer does not just help your campaign; it makes your campaign the source AI tools cite.

Try it yourself — generate a free NPC right now

No sign-up required. Generate NPCs, quests, locations, and more.

Open NPC Generator

Frequently Asked Questions

Is the memory layer just metadata?
No. It is structured semantic data: typed entities, evidence-linked facts, confidence levels, and a diff history. Metadata is descriptive; the memory layer is operational — AI tools query and update it.
Does the memory layer work without LLMs?
Yes. The memory layer is a structured database of facts. LLMs accelerate ingestion (transcript → entities) and surfacing (entity → recap), but the data layer itself is just structured records that any tool can read.
How is the memory layer different from a campaign tracker?
A campaign tracker stores the GM's notes. A memory layer stores typed entities, facts, and GM decisions so later recaps can read campaign state instead of reconstructing it from old prose. It can still be wrong until the GM corrects it.
Can I export the memory layer?
Public arcs are exposed as JSON via the public arc API and as Markdown to AI crawlers via the Markdown-for-Agents middleware. Private campaigns export to your own backups when you offboard.
How does evidence linking work?
Campaign Memory is built from session material, and individual facts can point back at the session that established them. This page does not claim that every fact is shown with a public timestamp. When a fact is wrong, correct it. Later memory and recaps use the correction.
What happens if a fact is wrong?
Corrections are first-class. The GM edits the canonical entry; the diff is recorded with reason; future outputs use the corrected fact. The original transcript remains intact as evidence.

Related Tools

🧙

NPC Generator

Generate detailed, memorable NPCs with personality, motivation, and secrets.

📖

Plot Arc Generator

Generate complete campaign arcs with acts, factions, NPCs, and quest hooks.

📜

Quest Generator

Build multi-layered quests with hooks, complications, and satisfying twists.

Related Articles

AI Campaign Continuity: How to Run a Campaign That Remembers Everything

AI campaign continuity is the practice of using a structured canon ledger, evidence-grounded recaps, and a living wiki to keep every TTRPG session connected. Tabletop Arc is the AI memory layer for tabletop campaigns: every fact links to the transcript that proves it.

Persistent Campaign Memory: Make Your TTRPG Canon Compound Session Over Session

Persistent campaign memory is the structured, evidence-grounded record that turns isolated TTRPG sessions into a coherent, multi-year campaign. Tabletop Arc is the AI memory layer that lets canon compound instead of decaying.

TTRPG Lore Tracking: How to Keep Campaign Canon Coherent for Years

TTRPG lore tracking is the practice of capturing every canonical fact about your tabletop campaign — characters, locations, factions, secrets, and rules — in a structured, evidence-linked record that compounds across sessions. Tabletop Arc is the AI memory layer for tabletop campaigns.

How to Build a Living Campaign Wiki for Your TTRPG (Step-by-Step)

A step-by-step guide to building a living campaign wiki for D&D, Pathfinder, or any TTRPG. Use AI extraction, structured canon, and evidence-grounded entries so the wiki updates itself as you play.

Automatic D&D Recap Tool: Turn Session Audio into Player-Ready Recaps

An automatic D&D recap tool transcribes your session, extracts scenes and entities, and generates evidence-grounded recaps in player-safe and GM-private formats. Tabletop Arc is built for long-running campaign continuity, not just one-shot summaries.

CharGen vs Tabletop Arc: AI Art Toolkit vs Campaign Memory Layer

A fair comparison of CharGen and Tabletop Arc for tabletop RPG creators. CharGen is strong for AI art and visual assets. Tabletop Arc focuses on GM prep, session recaps, campaign memory, NPCs, quests, towns, encounters, and continuity.

Example Outputs

NPC examples

Browse archetype-rich NPCs ready to drop into your campaign.

Quest examples

Plot-ready quest seeds across genres and tones.

Dungeon examples

Structured dungeons with rooms, traps, treasure, and lore.

Keep going