In production Resonar
Your social media agency, with you as the only approver
Enterprise AI system
Resonar
Your social media agency, with you as the only approver
The code lives in a private repository; this sheet and the demo, when there is one, show the work.
- Status
- In production
- Type
- Enterprise AI system
- Running since
- October 2026
- Role
- Sole engineer
- Domains
- marketing, personal branding, social media
- Cast
- Python, Typer, Fastapi, React, TypeScript, Vite, Httpx, Pytest, Playwright, Anthropic, Gemini
Context and problem
A social media agency run by AI agents, with approval queues and per-account budgets. It runs a real LinkedIn account.
Keeping a useful social media presence takes strategy, consistency, per-network formats and reading metrics; a professional, a creator or a small brand has no agency and no time, so they post little and learn nothing from the results (estimate: an agency costs 500-1,500 USD a month, doing it alone 4-6 hours a week).
Screens
How it was built
Built in 2.9 h · 1,183 tests · 2 audits
Source: Trysac/22-resonar-personal docs/development-report.md (lines 12, 15, 22), dated 21 November 2026.
Results with dated evidence
Outcomes as stated in the manifest; the measured figures are the ones with a date.
- Configurable roles per account with graded autonomy; nothing is published without a named human approval.
- Format validator agreement 100 % on 56 golden cases across LinkedIn, Instagram, TikTok and Facebook (offline).
- Ideas with correct evidence 92.9 % on the mining gold set (target 90 %, demo adapter).
Case study in preparation
The full case study adds these sections when it is published:
- Constraints
- Architecture diagrams (C4 context and containers)
- Three key decisions
- What went wrong
- Demo video
- What I would do differently
Highlights
multi agent roles, configurable autonomy, human approval, aggregator publishing, voice profile, learning loop