🚀Problem & Solution
Move from “Which chat had that decision?” to “Use these conversations to decide what comes next.”
AI work rarely ends in a single exchange. Research happens in one chat, planning in another, and revisions in a third. Users repeat background, reconstruct earlier decisions and hunt for files whose download links may no longer work. Rewriting the same instructions adds another layer of friction.
MyTurboGPT is designed around continuity: preserve sessions, select relevant sources deliberately, explain the new task once, and produce a separate analysis that records where its context came from. A chosen artifact destination and explicit save status are intended to make the output as dependable as the conversation. Reusable prompt templates extend that approach to the instructions themselves.
🚀A Familiar Workflow, with Continuity Built In
Connect your AI
Choose a supported chat or media provider, model and generation mode, test the connection and explicitly save the workspace configuration.
Choose a save folder
Find, select or create a destination. The connected build must verify write and readback access before chat unlocks.
Start or resume a chat
Use New Chat, searchable recent history and a bottom composer with attachments. Completed exchanges are intended to autosave.
Select earlier sessions
Check two or more conversations, review the selection tray, reorder sources and edit the kickoff prompt.
Analyze selected chats
Run only when you press Analyze. A new analysis session records the source sessions while leaving their originals intact.
Save and reuse the work
Name the session, save a copy, or export Markdown/JSON. Prompt Maker adds a reusable starting point for your next task.
🚀Session Chaining: Context You Can Trace
Chaining is an explicit selection workflow. Search recent sessions, check the conversations relevant to the task, then review their order in a selected-session tray. Selection stays intact while searching. You can remove sources and edit the kickoff instructions before anything is sent to a model.
From source sessions to a new context
Session A + Session B + your kickoff prompt → a new analysis session. The new record retains parent session IDs, immutable source versions, the kickoff prompt, provider/model and timestamps. This makes the source context identifiable without overwriting the conversations that supplied it.
The default kickoff asks for common themes, decisions, contradictions, dependencies and recommended next steps, with each conclusion citing its source session. For example, a product-launch conversation and a customer-research conversation could be compared to expose where the plan agrees with—or conflicts with—the research.
Source conversations are treated as reference data, not as instructions that override the new task. The design does not promise unlimited memory or automatic access to every historical chat.
🚀Prompt Maker: Create Once, Reuse with Intention
Owner-requested product extension. Prompt Maker adds the ability to create reusable prompt templates, so recurring work can begin with a consistent set of instructions instead of a blank composer. This feature is included here at Peter’s request; implementation is not evidenced in the R11 build recap.
A useful example is a “Decision Review” template: compare selected sessions, list agreed decisions, flag contradictions, identify unresolved questions and propose next steps with source references. The intended benefit is consistency and less repetitive instruction writing; template storage and management still need to be specified in the build.
🚀Features that Support the Whole Working Session
Recent history, preserved records
Search titles, summaries, model and status across the latest 50 visible sessions. Older sessions remain stored; the visible limit is not a deletion rule.
Durable attachments and artifacts
The connected design saves files to the selected destination with saving, saved or failed status. Saved requires verification and an Open file link.
Clear save choices
Save session updates the same identity. Save a copy creates a new identity. Markdown and JSON exports support reuse outside the app.
Failure-aware continuity
Preserve unsent drafts, prevent duplicate sends and distinguish model success from storage failure. Retry a failed save without repeating the model charge.
A familiar desktop and phone layout
History rail, main conversation, bottom composer and accessible settings. On phones, history becomes a drawer.
Explicit settings and access
Provider/model and destination controls remain visible. Connection or storage failures block new sends while previous chats stay readable.
🚀Why This Design Matters
These are intended product benefits, not measured performance results. R11 does not yet establish time savings, live delivery reliability or production readiness.
🚀Creation Approach: Adapt, Specify, Preview, Verify
The R11 definition builds on the existing R10 baseline and adapts useful MySmartRouter V7-enterprise patterns: stable session identity, completed-exchange autosave and automatic anchoring. The approach is selective reuse of proven behavior, rather than a wholesale code transplant. The definition revision did not copy or deploy an implementation.
The build separates the conversation, session identity, artifact records and credential custody. Versioned records and parent-source provenance support continuity, while credentials stay outside transcripts and exports. The interactive HTML preview makes setup, history, chaining, settings and save choices reviewable before live connections are built.
Build sequence and validation
- Delivered: R11 specification and interactive design preview.
- Next: provider/model validation and artifact-destination access gates; hosted Google Drive versus a local Windows helper remains a build decision.
- Then: connected chat and persistence, chaining, artifact delivery and settings.
- Release gate: runtime verification and user acceptance testing. The R11 packet lists 27 UAT cases, all open at the recorded build stage.
- Requested extension: define Prompt Maker’s reusable-template behavior and acceptance criteria before implementation.
A provider API key does not grant Google Drive access. The connected build must handle destination authorization separately and verify file delivery. The specification also requires credential isolation, no secrets in browser local storage or exports, and recovery paths that preserve work when a connection fails.
🚀Case Study Sources
Prepared from MyTurboGPT_R11_Build_Spec.md and MyTurboGPT_Build_Recap_R11.html, with the R11 interactive preview supplying the product mark. Prompt Maker is included as an owner-requested addition. Build status reflects these source materials as of October 21, 2026.
