I turn messy handoffs into systems teams can operate.

CRM workflows, lead qualification, internal dashboards, and supervised AI systems built around the business outcome and the people responsible for it.

Justus Loosli · Owner and Automation Systems Builder · Silex Digital

10participants recruited in month one at one research siteLive campaign result

17stages in a supervised AI operations workflowImplemented system

18.5s → ~2srecorded development response for one campaign-data endpointTechnical measurement

Selected work

Three systems, three different kinds of proof.

Results are labeled as live, technical, or scope-based so the work stays useful without turning engineering evidence into a business claim.

Live campaign result

Prioritized the leads most likely to qualify.

For a clinical-research site, I helped build a deeper GoHighLevel prescreening flow. Answers fed a lead score so the team could see and review the strongest candidates first instead of treating every inquiry the same.

GoHighLevelLead scoringMeta lead flowStaff handoff

System path

Lead capturedSource and screening answers retained

Answers scoredPriority based on agreed qualification signals

Best next action visibleStaff reviews candidates in priority order

The problem

High lead volume made it harder to see which people were most likely to qualify and deserved attention first.

What I built

Prescreen questions, answer-based scoring, CRM visibility, and a prioritized path for staff review.

Recorded result

10 recruits in the first month
~60% of the site's allocated study spots
$200K+ expected revenue over the course of the study

Scope note: the results apply to the first site and first month. The financial figure is expected revenue over the course of the study, not collected revenue or measured ROI. Scoring prioritized staff review and did not determine final study eligibility.

Implemented system

Built an AI workflow that knows when a human must decide.

I designed and built a supervised operating system for clinical-research marketing. The workflow coordinates specialist AI work while keeping publishing, spend, regulated release, and approval under explicit human control.

Python and FlaskReact and ViteNotion APIHuman approval

Control model

Specialist work preparedResearch, strategy, copy, creative, funnels

Version and checks boundBlockers stay visible until resolved

Human decision requiredConsequential actions cannot self-approve

The problem

Prompts, documents, and handoffs gave operators no reliable view of what was ready, blocked, or approved.

What I built

A 17-stage workflow, React/Flask dashboard, version-bound approvals, blocker logic, lineage, usage controls, and durable recovery.

Verification

10 consecutive clean offline acceptance runs
140 / 140 frontend tests recorded passing
Passing full contract suite

Scope note: these are dated offline acceptance and interface results, not proof of a full live production campaign or a quantified business uplift. External campaign actions remain human controlled.

Delivered workflow

Connected lead capture to consistent sales follow-up.

For a media business, I built landing pages, ad funnels, and email and text follow-up systems that connected incoming leads to the sales process. The work centered on clear next steps and fewer manual handoffs.

Landing pagesAd funnelsEmail follow-upText follow-up

Sales path

Interest capturedLanding pages and ad funnels

Follow-up runsEmail and text support the next step

Sales handoffA clearer route from inquiry to conversation

The problem

Lead capture and sales follow-up needed a more consistent path instead of relying on one-off manual communication.

What I built

Landing experiences, funnel steps, and email and text follow-up connected to the sales process.

What it demonstrates

Practical system design across acquisition and follow-up, with the handoff to revenue work kept visible.

Scope note: no quantified revenue or close-rate improvement is claimed for this project. The evidence here is the delivered workflow and system scope.

Start with the business goal. Finish with a system someone can own.

Map

Define the trigger, handoffs, owner, end state, and the decision that matters.

Build

Implement the smallest useful path before expanding tools, channels, or agents.

Prove

Test normal cases, missing data, duplicates, failure states, and stop conditions.

Hand off

Make ownership, next actions, documentation, and support boundaries clear.

Best used on a real workflow with a clear owner and outcome.

I am strongest when the project connects multiple tools or people and someone needs to make the whole path understandable, testable, and easier to operate.

01
CRM and follow-up automationRouting, assignment, email and text follow-up, booking events, and clean stop conditions.
02
Lead qualification systemsForms, scoring, prioritization, CRM visibility, and clear handoffs to the responsible team.
03
AI agents and internal toolsStructured workflows, operator dashboards, APIs, approvals, recovery, and human control.

Send the workflow, the tools involved, and where it currently breaks.

I will tell you where I can help, what I would verify first, and the smallest useful phase to build. We can keep the full conversation and project inside the marketplace where we connected.