Home / Engine / Agents and Council Architecture

Affirmology Agent Registry

Updated Jun 16, 2026 · Affirmology_AgentRegistry_v1.md

Summary. Living list of every agent and engine in the system, what it does, and whether it is live. Keep it current as new agents (especially Studio agents) come online. Visual map: AffirmologyAgentPipelineMapv1.svg.

Affirmology Agent Registry

Living list of every agent and engine in the system, what it does, and whether it is live. Keep it current as new agents (especially Studio agents) come online. Visual map: Affirmology_AgentPipelineMap_v1.svg.

Status key: LIVE (working in the pipeline) · BUILT (exists, not yet wired) · DESIGNED (specced, partially scaffolded).

Chart engines (deterministic, no AI)

Generation

Audio

Corpus pipeline (the proprietary knowledge crew, runs on the laptop SSD)

Orchestration

Knowledge layer

Surfaces

Planned agents (the fleet to grow)

Capture every new agent idea here so it gets built, not lost. All DESIGNED unless noted, and all live on the STUDIO side (experimentation), never the locked Demo, until promoted.

Quality control: - Script QC - checks a generated script against its structure template, the house style, length, and the brand rules before it is shown. - Chart-accuracy QC - runs our engines against known public charts (people with published HD/astrology data) and scores the matches, so engine bugs like the Manifesting-Generator one are caught automatically. - Brand/safety QC - flags em dashes, "Soul" for Sol, Dispenza-as-method, and any Tier-C leakage. - (Audio QC is BUILT and being wired in; the placement Verifier is LIVE.)

Ideation: - Idea agent - proposes new structures, categories, and audio types to test, with a rationale, dropped into the Studio for Sol and Jeff to try. - Variant agent - given a render we like, proposes A/B directions to explore.

Analysis and insight (the layer that decides WHAT to say and WHEN; full design in Affirmology_InsightAgents_Design_v1.md): - Blueprint Insight Agent - studies one person's full chart across all four systems and surfaces the convergences, productive tensions, and latent potentials that make them uniquely them. The "what makes me uniquely me" agent. BUILD FIRST (uses data we already have). - Transit & Timing Agent - computes current transits plus personal-year/month timing against the natal chart and names what is activated now, grounded in the transits corpus. Makes audios meet the present moment and gives a reason to return. - Potentials & Recommendation Agent - combines insight + timing into concrete potentials (a launch window, a relational opening, a rest need) and recommends which audio best serves it now. - Internal trend agent - mines our own customer-base audios, comfort votes, and feedback (the leaderboard data) to surface what is landing and what is drifting, and which chart signatures respond to which structures. - External trend agent - scans collective transits (outer-planet moves, eclipse and retrograde seasons) and wellness culture for timely audio themes, tied to real astrology, not hype.

Relationships cluster (synergy between people; full design in Affirmology_AgenticCouncil_Design_v1.md): Eros (romantic), Concordia (business/partnerships/teams, name open), Hestia (family), Gaia (interconnectedness, exploratory). All on a shared two-chart synastry engine fluent in Vedic + Western synastry, Human Design connection (connected/electromagnetic channels, the Penta), and Gene Keys (Venus sequence). The Human Design expert (Prometheus) is expanded to read connection mechanics and the Penta. Build after the Studio is live, per Jeff.

Personalization and growth: - PersonalizationAgent plus specialists (the SubconsciousOS design) - assemble a per-person context bundle and route to specialist writers. - Feedback-synthesis agent - turns Sol, Colin, and team feedback into concrete change proposals. - Corpus-growth agent - keeps adding fresh data (transits, moon omens, new sources) on a schedule. - Techniques-curation agent - maintains the techniques layer and tags which techniques each render used, so we learn what works.

The proprietary backend (the moat, and investor-video material)

This is hard-won, proprietary IP, not an off-the-shelf wrapper: - Verified, accurate chart math across four systems (most consumer apps get Human Design and Gene Keys wrong; ours is fixture-tested). - A multi-agent corpus pipeline that scraped and structured ~12,905 tradition-grounded interpretation records, tier-walled for licensing safety. - A generation system that grounds every word in a real person's chart plus those sources, with a verifier guarding against hallucination.

Honest moat framing (consistent with Affirmology_InvestorVideoScriptBrief_v2.md): the database alone is not the whole moat (a competitor could re-scrape over months). The deeper moat is taste, Sol's Miami community, and velocity. But this backend is the visible PROOF of substance and systems-level seriousness, exactly what the investor video should show: the map, the record counts, the four verified engines, the QC discipline. Use it as proof of depth, not as a claim that the data is uncopyable. Flagged for the investor video.

Status update 2026-06-16