ChatGPT
Primary full workflow. ChatGPT remains the smoothest, fullest support target for capture, search, highlights, and controlled new-chat updates.
Continuity Engine
Great ideas shouldn't get lost between the chats.
Save your AI chats permanently on your own computer as searchable, near-identical HTML transcripts, mark important highlights, and carry controlled context forward across ChatGPT, Claude, and Gemini browser workflows when a chat gets too long. Safe and searchable. Local and identical.
Product tour
Click any screenshot to enlarge.
The problem
Long AI work breaks in predictable ways.
Chats get too long. New chats lose the thread. Important ideas are buried in old sessions. Summaries miss detail and original phrasing. The user becomes the memory system.
Search the transcript. Capture the turning points.
Search finds the moment. Highlights preserve it. Continuity carries it forward.
- Searchable transcripts keep the full local conversation available when you need to find an exact decision, requirement, phrase, link, or turning point.
- Projects and sessions let you combine multiple chats into one project and carry the context from the entire project forward into a new chat.
- Manual highlights let you choose the text, images, files, formulas, and decisions that should not get buried.
- Images and file targets are saved with the transcript when available, including active download targets where the engine can extract them.*
- Local project memory keeps the archive, highlights, assets, searches, and update packages on your own computer.
*Download links can have limited lifespans. Continuity Engine can only extract active download links that are still available in the chat interface.
Build a better update package.
Continuity Engine merges three kinds of context so a new chat gets a compact handoff without flattening the work into a vague summary.
- Deterministic compression preserves controlled, repeatable context from the original transcript, including much of the original language and structure.
- Summary abstraction captures the deeper meaning behind the conversation: goals, decisions, assumptions, open questions, and why the work matters.
- User-selected text and image highlights protect the exact moments, evidence, visuals, formulas, files, and decisions you choose not to lose.
Together, these layers create one comprehensive, accurate, highly compressed update package that can represent a single chat or merge multiple project sessions into one new-chat handoff.
Privacy and control
Saved locally. Controlled by you.
The most important Continuity Engine design choice is simple: your AI project memory lives on your machine. The Engine recreates and stores the chat record locally as a Verbatim HTML transcript that looks almost identical to the original chat, including layout, formatting, images, and file downloads or active download targets when capturable. It is not a PDF and not just a summary. Safe and searchable. Local and identical.
Purchases and license activation use normal checkout/account systems. They do not require Betweenware to store your private project transcripts. Optional PRO API features are customer-enabled and customer-keyed.
Export
Capture the old chat into the local project record verbatim, including HTML formatting.
Review & Highlight
Search the transcript and select images or text blocks that should persist one update or on-going.
Build Package
Three context layers are created, then merged.
Deterministic Capsule
Controlled transcript compression with original language preserved.
AI Summary
Goals, decisions, assumptions, and deeper meaning.
Highlights Library
User-selected text, image, file, formula, and decision highlights.
Merged update package
Update
Start a new chat with one chat or the full project carried forward.
Platform support
One engine. Major AI workspaces and Chromium browsers.
Continuity Engine is built around a local archive and update workflow, not a single AI company's memory system. The supported path is intentionally practical: preserve the work locally, then use the best handoff method each AI interface allows.
Claude
Supported browser workflow. Core continuity works, with some handoff steps depending on Claude's current browser interface.
Gemini
Supported browser workflow. Gemini support uses the same local archive and update package, with guided manual upload or copy-paste steps where Gemini requires them.
Chrome / Chromium
Extension layer. The CE extension runs in Chrome and Chromium-family browsers, connecting browser AI sessions to the local Engine for capture, search, highlights, and update handoffs.
AI companies change their browser interfaces. Continuity Engine keeps the important part under your control: the local record, searchable transcript, selected highlights, and update package. Platform-specific handoff steps can be found in the CE Manual.
Product ladder
Start free. Rescue a bad chat. Upgrade when your work needs continuity.
Chat Health Meter
A free tool that helps users see when a chat is aging, slowing, or becoming unstable.
- Free Chrome Web Store extension
- Chat Health Meter in browser chat
- Chat Health Meter in ChatGPT/OpenAI app
- No Developer Mode loading
- Useful before buying CE
Schrodinger's Chat
$19 / year
Annual unlock for the same Chrome Web Store extension as Chat Health Meter.
- Unlocks paid features in the free extension
- Platform-aware AI Browser Reset
- Manual summary handoff
- No local Engine required
- No Developer Mode loading
Activating Schrodinger's Chat...
Continuity Engine Basic
$10 / month
or $79/year
The core local continuity engine for preserving and carrying forward serious AI work.
- Verbatim HTML transcript archive
- Independent HTML Archive backup
- Incremental chat imports
- Transcript search
- AI Browser Reset
- ChatGPT, Claude, and Gemini browser workflows
- User-selected Highlights
- Three-part new-chat context update
- Multi-session updates
- Selectable update token density
- Project backups and archiving
Continuity Engine PRO
$15 / month
Monthly only at launch
Advanced continuity and retrieval workflow for power users.
- Everything in Basic, including HTML Archive
- AI-operated context search
- Selectable context search depth
- Prompt-time context injection
- AI-assisted multi-session merge
- Optional GPU assist and customer AI API key
- Chat update package export
First-month guarantee: First-time monthly customers can request a refund during their first 30 days. Annual Basic is discounted and non-refundable after activation. Start monthly if you want to evaluate the workflow first.
Compare
Four simple choices.
| Feature | Health Meter | Schrodinger's Chat | CE Basic | CE PRO |
|---|---|---|---|---|
| Chat Health Meter in browser chat | Yes | Yes | Yes | Yes |
| Chat Health Meter in ChatGPT/OpenAI app | Yes | Yes | Yes | Yes |
| New-chat manual update instructions | Yes | Yes | Yes | Yes |
| Chat Health & Health Log | No | Yes | Yes | Yes |
| AI Browser Reset | No | Yes | Yes | Yes |
| Manual summary handoff | No | Yes | No | No |
| Verbatim HTML transcript archive | No | No | Yes | Yes |
| Independent HTML Archive backup | No | No | Yes | Yes |
| Transcript search | No | No | Yes | Yes |
| ChatGPT / Claude / Gemini browser workflows | Limited guidance | Yes | Yes | Yes |
| User-selected Highlights | No | No | Yes | Yes |
| Three-part new-chat context update | No | No | Yes | Yes |
| Multi-session updates | No | No | Yes | Yes |
| Selectable update token density | No | No | Yes | Yes |
| Project backups & archiving | No | No | Yes | Yes |
| AI-operated context search | No | No | No | Yes |
| Selectable context search depth | No | No | No | Yes |
| Prompt-time context injection | No | No | No | Yes |
| AI-assisted multi-session merge | No | No | No | Yes |
| Optional GPU-assisted merge/search | No | No | No | Yes |
| Optional customer AI API key | No | No | No | Yes |
| Customer-key AI access to local transcript archive | No | No | No | Yes |
| Chat update package export | No | No | No | Yes |
The API key is customer-supplied and optional. Continuity Engine licensing and core operation do not require an OpenAI API key.
Feature details
What each feature actually does.
Open the rows that matter to you. The page stays simple, but the details are here when a buyer wants to understand the machinery.
MONITOR
Chat Health Meter
Chat Health Meter in browser chat
Shows visible chat-health status while you work in a browser chat, helping you notice when a long session is aging, slowing, or approaching the point where an update to a new chat should be planned.
Chat Health Meter in ChatGPT/OpenAI app
Provides the same basic health-warning concept for users working through the ChatGPT/OpenAI app environment where supported.
New-chat manual update instructions
Gives users a simple manual path for moving important context into a fresh chat when they are not using the full Continuity Engine update system.
Summary Handoff
Schrodinger's Chat
What is Schrodinger's Chat?
Schrodinger's Chat is the paid feature unlock inside the same Chrome Web Store extension as Chat Health Meter. It adds platform-aware AI Browser Reset and Manual summary context handoff for supported browser AI workspaces.
Does it require Continuity Engine?
No. Schrodinger's Chat does not require the local Engine. It runs in the Chrome Web Store extension layer for users who need occasional long-chat rescue tools without a monthly continuity system or Developer Mode loading.
Summary Handoff
The extension can place a summary-generation instruction into the current chat. The user presses Send, downloads or copies the AI-generated summary, then drags it into the prompt of a new chat. It is useful, but it depends on the current AI chat's own remaining context quality.
When should I upgrade to Continuity Engine?
If you need searchable local HTML transcripts, highlights, projects, deterministic capsules, multi-session updates, or safe and searchable long-term project memory, use Continuity Engine Basic instead of treating every long chat as an emergency.
BASIC Continuity
Continuity Engine Basic
Local verbatim archive
Each entire chat is recreated locally on your hard drive as a Verbatim HTML transcript that looks almost identical to the original chat. It preserves layout, formatting, images, and file downloads or active download targets when they can be captured. The result is a permanent, fully searchable project record, not a PDF and not a flattened summary. Safe and searchable. Local and identical.
HTML Archive
Each time a chat is imported, Continuity Engine also creates an independent HTML archive copy of the captured chat. The archive is saved with linked image and file assets when available, so the transcript remains usable outside the Engine as a permanent local record.
This backup is independent of the Engine database. Even if the user stops using Continuity Engine later, they still have the local HTML transcript and captured assets on their hard drive: safe, searchable, local, and nearly identical to the original chat.
Privacy-first local storage
Continuity Engine does not require Betweenware to hold your private project transcripts in the cloud. The working record is stored locally, and update packages are generated from that local record when you choose to carry context into a new chat.
Transcript search
Search the entire project transcript for keywords, phrases, and remembered fragments instead of digging through long browser scrollback.
AI Browser Reset
AI Browser Reset is platform-aware. The Engine detects the active supported AI workspace, such as ChatGPT, Claude, or Gemini, then adjusts the reset behavior and button label for that platform. It can clear and refresh browser-side AI chat page state that contributes to visible slowdown, hesitation, or interface drag during long sessions.
In some cases, Browser Reset can improve responsiveness and delay the moment when a full new-chat refresh is needed. It is not a context repair tool: it does not fix damage to the chat's underlying context or logic state caused by repeated AI-backend compression rounds. When the conversation itself has degraded, Continuity Engine's update workflow is still the proper handoff path.
ChatGPT, Claude, and Gemini browser workflows
Continuity Engine uses one local engine and extension layer to support the major browser AI workspaces. ChatGPT is the primary full workflow target. Claude and Gemini support use documented platform-specific handoff steps where their interfaces require them.
User-selected Highlights
Highlights can include user-selected text blocks, images, formulas, decisions, and downloadable files when active download links can be captured. They can be set to persist as one-shot context for the next chat only, or as pinned context that continues across ongoing chats. The Library also carries agent-generated Summary Highlights, so the update package can include both user-selected evidence and the most important summary-level concepts.
Three-part new-chat context update
The new-chat update package combines three parts: 1) a deterministic capsule, 2) a General Summary, and 3) a Library document. The capsule preserves structured transcript context, the summary adds meaning, and the Library carries selected evidence and highlights.
Deterministic capsule
A roughly 10-50:1 deterministic compression of transcript data preserves accuracy of language and context without relying only on loose summary memory.
General Summary and Summary Highlights
The browser AI agent is automatically instructed to generate a detailed summary of the current chat for inclusion in the new-chat update package. It also generates eight Summary Highlights describing the most important concepts in the chat session.
Library document
The Library contains the agent-generated Summary Highlights plus user-selected text, image, and file highlights. It is included in the new-chat update package so important evidence can travel with the capsule.
Multi-session updates
For projects with multiple sessions, the Engine merges deterministic capsules for all sessions and includes either the last summary or all summaries in the update package.
Selectable update token density
The size of the update package is user-selectable at 5K, 10K, 20K, or a user-designated maximum from 20K to 50K tokens.
Project backups and archiving
Projects and highlights can be archived in user-selectable drive locations. The Assets folder for images, Summary Highlights, and files can also use a user-designated hard-drive location separate from the Engine directory.
ADVANCED Continuity
Continuity Engine PRO
Everything in Basic
PRO includes the full Basic continuity system: local archive, independent HTML Archive backup, transcript search, highlights, deterministic capsule, Agent Summary, Summary Highlights, Library document, multi-session updates, token-density controls, and archiving.
AI-operated context search
The PRO context search lets the user enter a prompt into the CE PRO prompt window, then the local system can search the transcript for relevant context using CPU and, where available, optional GPU acceleration.
Selectable context search depth
Controls the breadth and depth of a context search. When API-backed search is enabled, this also helps the user limit the amount of API usage and related charges.
Prompt-time context injection
Relevant context found by prompt-based search can be injected into the browser chat along with the user's original prompt. The user can bring context from the previous project into each prompt, not just during a one-time update.
AI-assisted multi-session merge
If a GPU is available, or if the customer enables an API key, the system can use AI agents to more intelligently merge capsules and summaries for multi-session projects.
Optional GPU-assisted merge/search
Where supported, local GPU power can assist merge and search functions so more work can happen locally instead of through an external API.
Optional customer AI API key
When a customer enters their AI API key into the Engine, it creates a customer-controlled back channel between an AI agent and the transcript stored in the Engine. API usage is optional and customer-paid.
Customer-key AI access to local transcript archive
When the API is enabled, the connected AI agent can access transcript data to perform selective summaries, merges, and concept searches. Results can be stored in the Assets folder as a document or injected into the browser chat.
Chat update package export
If the user works with other AIs besides the one the Engine is connected to, the update package can be generated as a ZIP document and uploaded into other AI models outside the Chrome browser chat environment, including Claude, Gemini, xAI, and others.
Get COMPLETE control of your chats, permanently, regardless of what the AI companies may do in the future!
Continuity Engine gives long-running AI work a local memory layer you can search, preserve, and carry forward.
Compatibility status and updates:
Continuity Engine depends on browser pages controlled by the respective AI companies. Betweenware tests supported browser-based AI interfaces daily and posts customer update notes when a compatibility fix is needed.
View Continuity Engine updates and current compatibility status