Two visual systems.
One operating model.
A design direction for turning enrichment, evidence, review, and CRM action into work people can understand at a glance.
Dense systems become invisible when every screen looks the same.
Distinct styles, shared rules.
Make the evidence legible.
Reports should tell readers what happened, how certain it is, and what happens next—without asking them to decode a dashboard.
Where did it come from?
How sure are we?
What is the next action?
Archival intelligence for decisions.
Editorial serif headlines, maritime slate, hairline borders, and a signal-mint locator. TGA feels calm because the hierarchy does the work.
Metadata rail + dossier body + evidence table
Use as a locator, not wallpaper.
Industrial clarity for systems work.
Technical grotesk, carbon panels, violet execution paths, crimson failure paths, and rectangular state tags.
Chassis panels make boundaries visible.
LABS is always included.
Research enters. Review happens. The CRM stays clean.
Company/domain
Approved sources
Confidence
Twenty API
Draft outreach
Automatic discovery is acceptable. Automatic irreversible outreach is not the starting point.
Do not build the mobile app first.
The mobile app is a window. The useful product is the enrichment engine, evidence trail, and review loop behind it.
Secure Twenty deployment and backups.
One enrichment provider and one API client.
Confidence, evidence, duplicate control.
n8n is the traffic controller. The worker is the researcher.
New company
n8n workflow
Python/Node worker
Twenty API only — never direct PostgreSQL writes.
Every enriched field needs a receipt.
Automate research. Review changes. Draft outreach.
Automatic discovery → evidence collection → duplicate check → human approval → CRM update
Bulk unsolicited messaging
That boundary protects data quality, reputation, and the people represented in the records.