Private betaRequest access

Cloud-hosted project memory and decision intelligence

Every decision
has a trail.

Connect conversations, sources, decisions, tasks, and AI work into one reviewable project record with provenance.

Request private beta See how ReasonTrail works
···

Conversation

Imported project discussion

Project: Atlas · Captured from shared conversation

MAY 12 · 9:14 AM

Decision

Choose phased rollout

Project: Atlas · Approved · 4 supporting sources

MAY 12 · 11:03 AM

Source

Research notes

Project: Atlas · URL capture · Reviewed

MAY 9 · 4:42 PM

AI synthesis

Key takeaways and next steps

Project: Atlas · Provider output · Needs review

MAY 12 · 1:18 PM

Bounded project context

CaptureConnectRetrieveReviewContinue

Project memory

One project memory,
every source.

01

One memory, every source

Capture pasted ChatGPT conversations, supported shared ChatGPT and Claude links, notes, URLs, email, repository documentation, and Mac Companion activity.

02

Decisions with evidence

Keep sources, assumptions, owners, approval status, related tasks, superseded choices, and project-health effects together.

03

Bounded context for AI

Discover intent, select a project, rank evidence, respect token budgets, and send only relevant context to the configured provider.

Context Engine

Relevant context, assembled before the model answers.

ReasonTrail does not indiscriminately send an entire database to a model. It builds a bounded, versioned context package for the selected provider.

01Intent discovery
02Project selection
03Evidence retrieval
04Relevance ranking
05Token budgeting
06Bounded context
07Provider request
08Reviewable output

Human review boundary

AI can propose. People control the record.

Important project memory is not silently accepted. Proposed changes enter review with their provenance attached.

ApproveEditRejectArchiveDelete

Verified Operator actions follow one persisted lifecycle. Sensitive writes require scoped approval, and memory is updated only after verification.

Model Context Protocol

A controlled boundary between AI clients and project memory.

Cloud MCP is coming during private beta. The planned endpoint is not operational. Local stdio MCP remains available for developers and self-hosted installations.

https://app.reasontrail.ai/mcp PLANNEDView MCP connection options →
CONTROLLED SCOPEAI client
COMING IN BETAAuthenticated ReasonTrail MCP server
CONTROLLED SCOPEActive organization
CONTROLLED SCOPEApproved projects + tools
CONTROLLED SCOPEBounded context
CONTROLLED SCOPEReviewable changes

Deployment and availability

Choose the mode that fits your operating model.

ModeAudienceStatus
ReasonTrail CloudCustomers and teamsPrivate beta
Mac CompanionApple-silicon Mac usersDeveloper preview
Local / self-hostedDevelopers and advanced operatorsManual setup
Remote Cloud MCPSupported MCP clientsComing in beta

Private beta

Bring the trail to your next decision.

Join the private beta for project memory that people—and AI clients—can inspect and trust.

Request private beta