Operations
Social
Anthropic Product Catalog
Every Anthropic product mapped to its role in Databayt's operations — who uses what and why.
Claude Code
The primary development interface. Kun's configuration loads automatically on all surfaces.
Surfaces
| Surface | Description |
|---|---|
| Terminal CLI | Primary development interface |
| Desktop App | Mac/Windows standalone app |
| claude.ai/code | Browser-based access |
| iOS App | Mobile access |
| VS Code Extension | IDE integration |
Key Capabilities Kun Uses
| Feature | Kun Usage |
|---|---|
| Hooks (21 lifecycle events) | Auto-format, port management, session logging |
| Skills | 55 skills across two plugins |
| Subagents | 46 specialized agents |
| MCP (3,000+ integrations) | 20 servers live |
| Plugins | kun-company + kun-stack distribute the fleets |
| Agent Teams (Experimental) | Parallel feature development |
| Git Worktrees | Transient agent isolation only — the repo workflow stays main-only |
| Remote Control / Dispatch | Cross-device handoff |
| Scheduled Tasks | Cloud, Desktop, or /loop |
Claude Desktop App
Standalone macOS/Windows application.
| Feature | Databayt Usage |
|---|---|
| Chat | General queries, planning |
| Cowork mode | Business and content workflows |
| Code tab | Visual diff review |
| Multiple sessions | Parallel workstreams |
| Scheduled tasks | Daily builds |
| Voice mode | Accessibility and hands-free input |
Cowork Use Cases (Research Preview)
| Role | Use Cases |
|---|---|
| Business | Client outreach, proposals, market research, financial analysis |
| R&D | Arabic content, documentation, research, UX writing |
| Tech | Project planning, architecture documents, specifications |
Agent SDK
Build production AI agents. Python and TypeScript.
Databayt Roadmap
| Agent | Purpose | Phase |
|---|---|---|
| PR Review | Auto-review every pull request | 2 |
| Deploy Verify | Post-deploy health check | 2 |
| School Onboarding | Automated tenant setup | 3 |
| Report Generator | Client reports | 3 |
Models
| Model | Use Case |
|---|---|
| Fable 5 (Mythos class) | The main loop — interactive sessions, captain-grade judgment |
| Opus 4.8 (1M context) | Architecture, orchestration, review, strategy agents |
| Sonnet 5 (1M context) | Standard build agents, fast iteration |
| Haiku 4.5 | Mechanical agents, search, lookups, exploration |
Kun runs Fable 5 in the main loop (canonical in .claude/engine.json), with Opus 4.8 → Sonnet 5 as fallbacks. Agents carry a tier alias (opus / sonnet / haiku), never a pinned version, so the fleet upgrades in one line. API prices live at platform.claude.com pricing — we don't hardcode them; billing is subscription-only anyway.
Databayt's Current Plan
| Item | Plan |
|---|---|
| Primary | Max 5x ($100/mo) — subscription-only, no API spend |
| Shared access | Desktop/Web for team |
Growth Path
| Stage | Plan | When |
|---|---|---|
| Now | Max 5x (1 seat) | Current |
| Team grows | + Pro seats | When team needs own accounts |
| Revenue stable | Team Premium | When revenue target achieved |
| Scale | Enterprise | Multiple products, clients |
Cost Optimization
| Technique | Savings | When |
|---|---|---|
| Prompt Caching | 90% | Repeated CLAUDE.md context |
| Batch API | 50% | Async CI/CD reviews |
| Caching + Batch | 95% | High-volume operations |
| Haiku for exploration | 80% vs Opus | Search, lookups |
All Plans Reference
| Plan | Price | Code | Cowork | Models |
|---|---|---|---|---|
| Free | $0 | Limited | - | Sonnet |
| Pro | $20/mo | Yes | - | All |
| Max 5x | $100/mo | Yes | Yes | All |
| Max 20x | $200/mo | Yes | Yes | All + priority Opus |
| Team Premium | $125/seat/mo | Yes | - | All |
| Enterprise | Custom | Yes | Yes | SSO/SCIM/Audit |
Feature Availability
| Feature | Pro | Max | Team Premium | Enterprise |
|---|---|---|---|---|
| Claude Code | Yes | Yes | Yes | Yes |
| Cowork | - | Yes | - | Yes |
| Agent Teams | Yes | Yes | Yes | Yes |
| Scheduled Cloud Tasks | Yes | Yes | Yes | Yes |
| SSO/SCIM | - | - | - | Yes |
| 1M Context | Beta | Yes | Yes | Yes |