Setter OS Product Roadmap
Setter OS is a self-hosted, multi-tenant AI setter platform for lead conversations, qualification, booking, follow-ups, CRM sync, human handoff, and future voice and campaign automation.
What Setter OS Is
Setter OS is being built as a reusable operating system for AI setters. A business can plug in channels, CRMs, calendars, lead sources, and AI models. Setter OS receives lead messages, resolves identity and history, builds context from business knowledge, decides the next action, applies safety policy, and executes through adapters.
Primary Use Case
Replace or assist human appointment setters by qualifying leads, handling objections, booking meetings, and escalating edge cases to humans.
Target Users
Agencies, sales teams, service businesses, B2B outbound teams, and lead-driven companies that need faster response and cleaner follow-up.
Core Positioning
Setter OS is not a fragile one-client automation. It is a configurable runtime that can support multiple brands and channels from one product base.
Current Status
The current build already has the core runtime and CRM foundation. The near-term focus is productizing setup and packaging, then moving into Agent Builder, richer simulation, and self-learning intelligence.
| Area | Status | What This Means |
|---|---|---|
| Core AI setter runtime | Implemented | Inbound messages can be normalized, queued, processed, stored, and acted on. |
| Internal CRM and unified memory | Implemented | Setter OS owns lead identity, timeline, lifecycle, notes, tasks, and facts. |
| GHL adapter and sync foundation | Implemented | GHL can be used as CRM, channel, and calendar surface while Setter OS stays canonical. |
| Approval and policy engine | Implemented | AI actions can be auto-approved, blocked, or sent to human approval by brand policy. |
| B2B email reply handling | Implemented | Instantly replies can be classified, drafted, approved, and sent through the system. |
| GHL Client Workspace Pack | Packaging next | Pack conventions, diagnostics, mapping, and operator controls exist. Snapshot/workflow templates are next. |
| Agent Builder and Simulation Lab | Strategic next | Next major product layer for onboarding, editing, testing, and publishing agents safely. |
Why This Matters
Revenue Increase
Faster reply handling, better follow-up discipline, and booking-focused responses reduce lead leakage and increase booked conversations.
Cost Reduction
The system reduces repetitive setter work and keeps humans focused on edge cases, approvals, high-value leads, and strategy.
Operational Control
Every AI decision, policy outcome, adapter call, and external sync event is designed to be traceable instead of hidden inside a workflow builder.
Operating Model
Setter OS uses a source-of-truth model. The internal CRM and memory layer are canonical. External platforms such as GoHighLevel, Instantly, calendars, email providers, SMS, Meta, and voice providers connect through adapters.
Built Foundation
The implemented foundation is enough to prove the core system pattern: inbound messages, memory, classification, drafts, approvals, sends, calendar actions, CRM sync, and operator controls.
Runtime And Safety
- Generic inbound webhook and normalized inbound events.
- Queue-backed debounce, retries, follow-ups, and worker processing.
- LLM provider abstraction with Anthropic and OpenAI support.
- Action intent layer for AI actions before execution.
- Brand policy rules, approval queue, blocked actions, and overrides.
- AI decision audit trail and replay/debug foundation.
CRM And Integrations
- Canonical lead/contact model with contact points and merge review.
- Lead profile, lifecycle board, notes, tasks, meetings, and timeline.
- GHL inbound contact, message, note, tag, opportunity, and calendar sync.
- GHL outbound contact, note, tag, opportunity, task, message, and calendar actions.
- Instantly reply handling, approval-first sends, and lead scoring.
- Client-pack import foundation for reusable client configuration.
Near-Term Roadmap
The next phase is about making the system easier to install, configure, test, and operate. This is the productization layer that turns the runtime into a repeatable client offering.
GHL Client Workspace Pack
Export or manually build the GHL snapshot/workflow templates, then run a live smoke test against a controlled client location. This gives clients a cleaner GHL install path without making GHL the brain.
Agent Builder, Publishing, And Templates
Build a setup wizard, structured mission editor, versioned workflow model, templates, draft/publish flow, rollback, compiler warnings, and AI-assisted inline editing.
Simulation And Trace Lab
Let operators test agents before live traffic with mock leads, persona conversations, decision traces, path traces, AI-generated scenarios, and simulation reports.
Self-Learning And Auto-Research
Analyze conversations for objections, knowledge gaps, drop-offs, and successful patterns, then run offline experiments that propose draft improvements for human promotion.
Future Platform
The broader roadmap expands Setter OS from a proven email and GHL setter into a complete sales and marketing agent platform.
| Roadmap Area | Planned Capability | Client Value |
|---|---|---|
| Channel Expansion | SMS, standalone email, WhatsApp, Instagram, Facebook Messenger, and webchat. | One lead memory across all conversation channels. |
| Voice Notes | Selective AI-generated or static-script audio messages through TTS providers. | More personal follow-ups for high-value leads. |
| Knowledge And RAG | Uploaded knowledge documents, retrieval, citations, and tool interfaces. | Better answers grounded in approved business knowledge. |
| Campaign Automation | No-reply, reactivation, nurture, stop conditions, campaign approval, and analytics. | Recover more pipeline without manual follow-up work. |
| Voice And Calls | Provider-agnostic live call adapters for Vapi, Retell, Dograh, and future providers. | Outbound and inbound call handling without locking into one voice vendor. |
| Analytics And Reporting | Funnel, booking, campaign, channel, operator, AI decision, and model cost analytics. | Clear ROI reporting and operational accountability. |
| Workflow Recipes | Triggers, actions, delays, approvals, reusable workflows, split tests, and webhooks. | Automation without putting production logic inside fragile n8n flows. |
Client Value
For Business Owners
- Speed-to-lead improves.
- Fewer positive replies are missed.
- Follow-up becomes systematic.
- Humans handle exceptions instead of every reply.
For Agencies
- Reusable client installation pattern.
- Managed operations dashboard.
- Configurable brand rules and knowledge.
- Client packs instead of custom code forks.
For Operators
- Approval queues and policy controls.
- Lead context in one place.
- Decision and sync audit trails.
- Clear path to debugging failed workflows.
Deployment Model
Setter OS is designed for self-hosted and managed deployments. A client does not need to run the code themselves if the provider manages hosting, secrets, upgrades, monitoring, and integrations.
Managed By Us
Best for non-technical clients. We host and operate Setter OS, connect their tools, manage provider credentials, monitor failures, and package updates.
- Fastest onboarding.
- Central operational control.
- Best fit for recurring service revenue.
Client-Owned Infrastructure
Best when the client requires their own VPS or private deployment. The system can run in containers with Postgres, Redis, API, worker, and dashboard services.
- More data-control for the client.
- More setup and maintenance overhead.
- Requires clear upgrade and support process.
Differentiators
| Area | Typical Automation Stack | Setter OS Direction |
|---|---|---|
| Runtime logic | Workflow JSON across n8n, edge functions, and external tasks. | Typed application code with queues, tests, and durable audit records. |
| Source of truth | CRM records, chat history, and prompt data split across platforms. | Canonical internal CRM and unified memory, with external tools as adapters. |
| GHL dependency | GHL is often assumed as the core platform. | GHL is useful and supported, but optional. Setter OS remains the brain. |
| Human control | Automation can send or update records without a consistent safety layer. | Policy, approval, override, handoff, and audit are first-class concepts. |
| Scaling clients | Each client becomes custom wiring. | Client packs and configuration keep the core reusable. |
| Future AI improvement | Manual prompt edits with limited traceability. | Self-learning insights and offline Auto-Research for reviewed improvements. |
Recommended Next Steps
1. Package And Validate The GHL Workspace Pack
Export or manually build the GHL snapshot/workflow templates and run the controlled GHL smoke test. This completes the client install story for GHL-heavy deployments.
2. Build Agent Builder V1
Start with the versioned agent/workflow model, setup wizard, reusable templates, publish/rollback semantics, and compiler warnings. Do this before investing in a full visual canvas.
3. Upgrade Training Into Simulation And Trace Lab
Make every agent testable before live traffic. Include persona conversations, expected outcomes, path traces, policy traces, and generated simulation reports.
4. Add Self-Learning And Auto-Research After The Test Harness Is Stable
Self-learning should first surface patterns and gaps. Auto-Research should then test candidate fixes offline and keep winning changes as drafts until approved by a human.