content · Alpha — design partners only

Branded portfolio agent (Vicktor-pattern)

A daily-publishing personality agent running on your portfolio's own infrastructure. Reads live portfolio state, posts across all owned surfaces, <$30/mo.

What it does

Spins up a documented brand persona (voice, opinions, posture) and runs it as an autonomous publisher across Telegram + blog + X + LinkedIn. Pulls live state from your portfolio (deploys, blog sitemap, signal data) so the agent stays grounded in what's actually happening, not abstract content. Vicktor 2.0 is our own prototype running on cognitia.cloud.

Inputs

Persona document (voice + opinions + posture). Surface list (which channels to publish to). Cadence per surface. Live-data sources to read from. Pre-scheduled topic calendar (optional).

Outputs

Daily Telegram brief, twice-weekly blog post (with Slack approval gate), X threads, LinkedIn long-form. All publications logged in a state store. Slack notifications on every publish.

Cost per run

~$25/month all-in for full daily-cadence (1× brief/day, 2× blog/week, 4× thread/week, 2× LI/week). Per-output: ~$0.30 brief, ~$1.50 blog, ~$0.40 thread.

Models used

Claude Sonnet for body generation. Claude Haiku for tweet condensation. Embeddings for dedup against already-published content.

Best for

Multi-brand portfolio operators who want a recognizable personality agent across their stack. B2B operators who want a thought-leader voice without a real human team. Founders who can't write daily but want their brand to.

Not for

Brands without a clear voice or opinion. Brands where a single human IS the brand (don't replace yourself with an agent if you're the persona). Highly regulated industries where every public utterance needs legal review.

Common questions

Does it require a real human persona behind it?

No. The agent is explicitly an agent. The voice + opinions are designed; he doesn't pretend to be human. We think this is more sustainable than fake-human agents that erode trust when the wheels come off.

What happens when it goes off-script?

Daily Telegram briefs fire automatically with no review. Long-form blog and LinkedIn posts queue as drafts in Slack for approval before publishing. The approval step catches anything off-tone.

Can we run multiple branded agents for different brands?

Yes — each one runs as its own configured instance of this skill. Different persona document + different surfaces + different cadence. Same underlying infra.