A/IAALQUIM IASELECTED WORK · CASE 03
← Back to portfolio

CASE 03 · LIVE DEMO CAMPAIGN INTELLIGENCE

Alquim Ads

A decision layer on top of Meta Ads data.

Alquim Ads turns campaign data into executive KPIs, deterministic recommendations, LLM-assisted analysis and scheduled reports. The design separates calculation, rules and narrative so every conclusion keeps an inspectable base.

ROLEPRODUCT · ANALYTICS · AI ENGINEERING2026
Open public demo Visit product landing
METAcampaign source
KPIexecutive layer
RULESdeterministic recommendation
LLMassisted synthesis

PROJECT CARD / RECRUITER VIEW

The project's technical dossier.

EXPLICIT STATE LIVE DEMO
TARGET USER

Founders, performance marketers and business owners who need to understand campaigns without reviewing dozens of operational views.

WHAT I BUILT

I defined the product, the analytics architecture, the separation between KPIs, rules and narrative, and carried the experience to a deployed demo.

AI ENGINEERING SIGNAL / RECRUITER VIEW

Backend, APIs, deployment, testing and controls.

DEPLOYMENT
Active container behind HTTPS; public landing and demo respond.
BACKEND / API
Observed /api/health/ready endpoint; layers for ingestion, KPIs, rules and reporting.
DATA CONTRACT
Explicit demonstration dataset; it does not mix demo metrics with client results.
TESTING
Readiness 200 and navigation of dashboard, funnel, tables, recommendations and data quality verified.
OPERATIONS
Main service ready; backup failure documented as visible operational debt.
ARCHITECTURE OVERVIEW
  1. 01Meta Ads data
  2. 02KPI layer
  3. 03Rules engine
  4. 04LLM analysis
  5. 05Scheduled reports
WHAT WORKS TODAY
  • Public landing
  • Demo route
  • Main-service readiness
  • Executive product narrative
HARD TECHNICAL DECISIONS
  • Never delegate critical calculations to the LLM
  • Distinguish deterministic diagnosis from explanation
  • Present early access without overselling maturity
MEDIA AND DEMO
DEMO
Interactive public route
SCREENSHOTS
Available inside the demo
VIDEO
Pending recording
WHAT IS STILL PENDING
  • The public demo proves the product surface, not an active integration with the visitor's ad account.
  • Readiness confirms the main application responds; it does not replace full monitoring of dependencies and backups.
  • Recommendations must be read alongside business context, budget and attribution quality.

01 / PROBLEM

More metrics do not automatically produce better decisions.

Ad platforms deliver an abundance of data, but teams still need to detect deviations, prioritize actions and explain what changed without confusing correlation with recommendation.

02 / DESIGN DECISION

Separate calculation, rules and explanation.

The platform calculates KPIs from the data, applies explicit rules and uses the LLM as a synthesis layer. This reduces the risk of turning persuasive text into an opaque source of truth.

SYSTEM / END TO END

The product as an explicit flow.

Each phase has a distinct responsibility. That separation makes it possible to observe, test and fix the system without relying on intuition.

  1. 01

    Connect

    Ingest campaign data and preserve its operational granularity.

    DATA
  2. 02

    Calculate

    Build executive KPIs with consistent, comparable definitions.

    MEASURE
  3. 03

    Diagnose

    Apply deterministic rules to detect conditions and priorities.

    RULES
  4. 04

    Explain

    Use AI to turn structured findings into an executive read.

    NARRATE
  5. 05

    Distribute

    Package results into an experience and scheduled reports.

    SHIP

PROOF / WHAT IS ACTUALLY SHOWN

Evidence, not decoration.

The labels indicate whether the data was observed, comes from the artifact or belongs to the professional record of the case.

OBSERVED · AUG 11 2026DEMOnavigable public route
OBSERVED · AUG 11 2026READYreadiness endpoint
PRODUCT ARCHITECTURE3 LAYERSKPIs, rules and synthesis
DECLARED SCOPEEARLYavailability stage

SCOPE / HONEST BOUNDARIES

What is demonstrated and what still needs validation.

A technical portfolio must also show where the available evidence ends.

  • The public demo proves the product surface, not an active integration with the visitor's ad account.
  • Readiness confirms the main application responds; it does not replace full monitoring of dependencies and backups.
  • Recommendations must be read alongside business context, budget and attribution quality.

CASE 03 / ALQUIM ADS

A decision layer on top of Meta Ads data.

Open public demo View the full portfolio