We pointed Signal
at ourselves.
Every software company has a demo. Every demo is curated. So we did something different: we ran Signal on the production system that builds Signal. Not a sandbox. Not sample data. The tool examined the toolmaker.
Structural Disorder
~800 files
25% of the codebase. 45% of total lines of code. Stale mirrors, legacy artifacts, duplicated content, accumulating for 9 months, invisible to daily work.
Files Perceived
3,119
Every file classified deterministically. 238,722 lines of code. Velocity vectors computed at 6 time horizons. Dependencies mapped across projects.
First Client
Dalton Chemitac
Multinational manufacturer (BR/ES/US). NDA signed. Upload portal live. Stakeholder notifications automated. Same infrastructure, same provenance.
What Signal found
25% of the codebase shouldn't be there
Signal classified every file by structural health: whether it belongs in the living system or not. Four categories of disorder emerged:
- Stale mirrors: files duplicated outside their canonical location. Two copies of the same module. Edit the wrong one and nothing changes.
- Legacy artifacts: directories from prior architectures. Nine months of accumulated technical debt, invisible to daily development.
- Site duplication: the same HTML living in three different places. Which one is real?
- Archive copies: files already archived but still in the working tree, inflating every search and every metric.
No one asked Signal to find this. No one knew the number was 800. The system perceived disorder that the humans building it could not see.
What this means for you: if a 1-person company building the product accumulates 800 files of invisible disorder in 9 months, what does a 200-person company look like after 10 years?
The dependency map shows where risk lives
Signal mapped every import and reference relationship between files. Which files are load-bearing, many things depend on them. Which are isolated, candidates for removal. Where cross-project dependencies create coupling risk.
What this means for you: you can see which parts of your operation are load-bearing: the processes, documents, and people everything depends on. When one leaves or fails, you know the blast radius before it happens.
Velocity reveals what's real
Every file carries a 6-dimensional velocity vector: edit frequency measured at 1, 7, 14, 21, 30, and 90 days. Not a guess. A count. The system classifies each file's pattern: hot, spike, active, cooling, revived, dead, or unknown.
The honest result: with one week of data, most files showed as active. The system reported this accurately rather than presenting a more impressive but misleading breakdown. Over 30-90 days, the patterns differentiate. The system gets sharper every week.
What this means for you: in a consulting engagement, you get a snapshot. It expires the day after delivery. Signal builds a living velocity model that distinguishes real activity from noise, and it gets more accurate over time.
In parallel: first external client
While Signal analyzed echology internally, we deployed the same infrastructure for Dalton Chemitac Group, a multinational manufacturer operating across Brazil, Spain, and the US. Same session. Same system.
Secure Upload
Private portal. No login required. Every file witnessed in an immutable hash-chained ledger with SHA-256 verification.
Stakeholder Notifications
File uploaded, ledger witnessed, all stakeholders emailed automatically. No manual follow-up. No "did you get it?" emails.
NDA Signed
Mutual NDA generated, sent, and countersigned through the system. Witnessed in the Aletheia ledger. Every state change traceable.
Single View
Contacts, agreements, uploads, findings, review gates, pipeline events, and full audit trail. Everything about the engagement in one place.
The numbers
All numbers from ops.db. Not curated for this page.
| What Signal Tracks | Value |
|---|---|
| Files perceived | 3,119 |
| Lines of code | 238,722 |
| Structural disorder found | ~800 files (25%) |
| Initiatives tracked | 138 |
| Steps completed | 531 of 896 |
| Findings across deployments | 87 |
| Architectural decisions logged | 68 |
| Pipeline value | $255K |
| Active deployments | 5 |
| Provenance | Hash-chained, every state change |
The guarantee
If Signal doesn't surface at least 3 structural findings your team didn't already know about, you don't pay.
Today, Signal found approximately 800 files of disorder in a company that builds Signal. The founder, who wrote most of the code, did not know the number. If it finds this in a 1-person operation with 9 months of history, the question isn't whether it will find something in yours. The question is how much.
Data source: ops.db on Mac Mini. File metadata table: 3,119 entries with content hashes. Health classification: deterministic path analysis. Deployment date: 2026-03-24.
See what Signal finds in your business.
A Signal Audit takes one week. Your data, your hardware. Nothing leaves your building.