No. The whole point is that six personas replace a stringer, an editor, a growth marketer, and a sponsor BD you cannot afford. You run them as a single canvas, approve the issue at the end, and ship twice a week without a staff.

Three ways. Static context attaches your voice guide, banned phrases, and pricing page to Copywriter and BrandArchitect on every run. The rag_retrieve tool lets ContentStrategist pull from past issues, top performing tweets, and reader replies on demand. Cross-run memory means last week's open rate and reply thread are in scope for this week's draft.

No, it sits next to them. Beehiiv and Substack stay your system of send. Melaya is the research, draft, brand QA, and sponsor outreach layer that hands you an approved issue you paste into the editor, so deliverability and your subscriber list never move.

ConvertKit is the ESP and Sparkloop is the recommendation network. Melaya runs GrowthEngineer on a schedule to design referral copy, seed lists, and partner pitches, then drafts the Sparkloop swap pitch as a HITL gated email so you keep one click control.

Yes for the planning to draft loop. Notion AI gives you a single chat box. Melaya gives you a six persona pipeline that briefs, drafts, brand QAs, and stages every send for approval, with the issue history tracked across runs instead of trapped in pages.

Copywriter cites phrasings from your top performing issues loaded into the knowledge store, and BrandArchitect blocks any draft that fails the voice checklist. You can require a citation per section as a HITL pre-check, so every sentence ties back to a source line in your archive.

Any. Claude on BrandArchitect where voice judgment earns the cost, GPT mini on Copywriter variants, and a local Ollama on ContentStrategist research. Each persona picks its own model, so the heavy reasoning hits run once per issue, not once per chat turn.

Yes. Every run logs every step, every tool call, every model invocation, and every approval. Replay any run to see the brief, the citations from your archive, the brand QA verdict, and the final approved issue. The audit log is your editorial log.

With Gmail and your scraping allowlist authorized, the brief to approved issue workflow is a 4 node canvas: brief intake, draft, brand QA, stage for approval. Most solo operators ship it in a working session and have the first reviewed issue queued the same day.

Yes. agent_tools allowlists are scoped per persona. SocialMediaEngineer can draft a thread while only you hold gmail_send and twitter_post. The canvas shows the gate before a run starts, so a wrong click cannot put an unapproved draft on your list.