agentificial
Selected work

Selected work.
Scope stated.

Seven pieces of work, described without client names. Each says what we measured or built, the result as recorded, and where that result stops. Six of the seven are for one client group, a multi-clinic medical-tourism group (UK/TR). The seventh is for an agency affiliated with Agentificial, and is labelled as such.

showing 7 of 7
inbound · flow
  1. 01enquiry arrives
  2. 02qualify
  3. 03draft reply
  4. 04human approves
  5. 05reply sent, logged
Measurement· Jul 2026

Measuring speed-to-lead to the second, and finding the CRM's own field unreliable

A multi-clinic medical-tourism group had no trustworthy measure of how long a new lead waits for its first call. We built one from the phone system's own timestamps.

A multi-clinic medical-tourism group

11 h 01 mmedian time to first call
open →
ops · flow
  1. 01request or event
  2. 02agent works
  3. 03approval gate
  4. 04action taken
  5. 05audit log written
Monitoring· Jul – Oct 2026

A routing failure that stayed silent for 17 days, and the watchdog that watches outcomes

Lead-assignment workflows at a medical-tourism group stopped assigning owners without raising any error. We repaired them and built monitoring that checks the result, not the mechanism.

A multi-clinic medical-tourism group

17 dayssilent failure, 56 contacts
open →
intelligence · flow
  1. 01question asked
  2. 02sources queried
  3. 03answer drafted
  4. 04source cited per figure
  5. 05answer delivered
Diagnosis· Jul – Sep 2026

Diagnosing an 80% organic-traffic collapse from 17 months of Search Console data

A medical-tourism group's website lost four-fifths of its monthly search clicks. We found the cause, retracted two of our own repair recommendations, and set up weekly monitoring. Traffic has not recovered yet.

A multi-clinic medical-tourism group

−80.8%monthly search clicks
open →
workflow · flow
triggerenrichclassifydispatch
Infrastructure· Mar – Oct 2026

Self-hosted outbound email at fleet scale, and the six ways it failed silently

We built and operate cold-email infrastructure of about 160 domains and 790 mailboxes on self-hosted mail servers. Most of the work since has been making silent failures visible.

A multi-clinic medical-tourism group and a second brand under the same ownership

~160domains on self-hosted mail
open →
workflow · flow
triggerenrichclassifydispatch
Fix· Sep 2026

US phone numbers silently rewritten with a UK country code: 33 affected, 30 corrected

The CRM was prepending +44 to numbers submitted without a country code, so US leads from a Google Ads form landed with the wrong number. We found the root cause and corrected the records.

A multi-clinic medical-tourism group

30 of 33numbers corrected
open →
inbound · flow
  1. 01enquiry arrives
  2. 02qualify
  3. 03draft reply
  4. 04human approves
  5. 05reply sent, logged
Diagnosis· Sep – Oct 2026

Meta delivered 140 leads. The CRM captured 28.

A marketing agency affiliated with Agentificial was paying for Meta leads that never reached its CRM. We diagnosed the loss, found a second problem behind it, and monitor the gap. It is not closed yet.

A Canadian marketing agency affiliated with Agentificial (Meta lead generation for insurance brokers)

140 vs 28Meta leads vs CRM captures
open →
outbound · flow
  1. 01pick accounts
  2. 02research
  3. 03draft
  4. 04sender review
  5. 05send, classify replies
Diagnosis· 2026

64% of apparent email opens were Apple Mail prefetches, and link scanners inflated clicks

Cold-email dashboards at a medical-tourism group showed opens and clicks that looked like engagement. Splitting them by cohort showed that most were machines.

A multi-clinic medical-tourism group

455 of 711apparent opens were Apple prefetches
open →
· Could be you next

One wedge.
Measured first.

If anything here looks like a problem you have, let's talk. We pick the smallest thing that moves a number, measure it, and ship it.

Book intro callSee selected work