F.A.L A.I

Agency

Evidence by stage

Cases with verifiable stage and evidence

The portfolio distinguishes 1 delivered project from 7 initiatives in validation, with available context, scope and evidence.

F.A.L A.I Agency (internal dogfooding)
In validation

Coven — Marketing AI Control Tower

Internal SaaS under validation, with a Next.js frontend and tenant-isolated FastAPI backend for governed marketing operations.

Stage

Internal validation

Authenticated smoke

2026-05-05

API, not a customer outcome

Challenge

Centralize approvals, policies, channels, and costs without presenting implemented surfaces as customer outcomes.

Solution

The code mounts auth, operator/control tower, approvals, social, Meta Ads, content velocity, workspace, analytics, and subscription surfaces.

Available evidence

  • Separate Next.js frontend and FastAPI backend
  • Mounted operations, approvals, workspace, analytics, social, and billing routers
  • An authenticated API smoke validated auth, dashboard, ROI, and billing on 2026-05-05
Source
fal-ai-agency@d0fe4f1 — docs/runtime-matrix.md, docs/deploy-matrix.md, and authenticated API smoke dated 2026-05-05
Verified on

Limitations

The current audit did not log in.

Executive summaryTechnical detail
Coven — Marketing AI Control Tower
LA Suplementos
In validation

Smart Engine Sales — Bia for LA Suplementos

A production single-tenant deployment for LA Suplementos using Bia, FastAPI, LangGraph, PostgreSQL, and Evolution API.

Current deployment

1 tenant

LA Suplementos

Follow-up

Up to 2 touches

bounded and conditional

Challenge

Handle and follow up WhatsApp contacts with explicit limits, inventory lookup, and human intervention for sensitive decisions.

Solution

Bia serves the Evolution API channel, checks operational data, limits follow-ups, and alerts the owner through Telegram when human judgment is needed.

Available evidence

  • Documented single-tenant LA Suplementos operation
  • Bounded follow-up and conditional reactivation
  • Human handoff and operational Telegram alerts
Source
Smart_engine_sales@42b9571 — docs/PROJECT_STATUS.md and https://la.falai.agency/health (verified 2026-07-15)
Verified on

Limitations

Health does not prove conversion, sales, or response quality.

Executive summaryTechnical detail
Smart Engine Sales — Bia for LA Suplementos
F.A.L A.I Agency (internal)
In validation

Agency Lead Ops — Assisted B2B Prospecting

A local pipeline under validation to discover companies, cross-check CNPJ and public sources, retain evidence, score fit, and prepare drafts.

Validated mode

Local / no-send

Delivery gate

Fail-closed

human approval required

Challenge

Turn public research into a reviewable queue while blocking delivery when evidence, operational consent, or approval is missing.

Solution

The flow combines CNPJ records and public sources, factual enrichment, scoring, and drafts with fail-closed contact, opt-out, idempotency, and human-approval rules.

Available evidence

  • Discovery and enrichment with source URLs
  • Scoring and drafts subject to approval
  • Sendability, opt-out, and idempotency rules
Source
Prospeccao_Clientes_Agency@6c2b74e — outbound_policy, contact_enrichment, and no-send tests
Verified on

Limitations

No live provider, real delivery, or commercial pilot was proven.

Executive summaryTechnical detail
Agency Lead Ops — Assisted B2B Prospecting
Industrial Proof of Concept
In validation

MiningOps Copilot — Local Mining POC

A locally installed POC that adapts ASIC, energy, and environment CSV files, detects deterministic anomalies, and asks Qwen3 for explanations.

Input

CSV / JSONL

Polling window

10-30s

when the source is available

Challenge

Adapt heterogeneous operational files and produce understandable incidents without automatically controlling equipment.

Solution

FastAPI polls CSV/JSONL files, applies deterministic rules, records incidents, and uses Qwen3 through Ollama for explanations and suggested actions.

Available evidence

  • Docker/FastAPI stack and UI installed in the client environment
  • Deterministic anomaly detection
  • JSONL incidents and daily summary
Source
Integracao_LLM_Mineracao@27fb5b0 + local installation report dated 2026-04-06 (unversioned)
Verified on

Limitations

The later installation report is unversioned.

Executive summaryTechnical detail
MiningOps Copilot — Local Mining POC
AdTech Proof of Concept
In validation

Mobility Ad Network — Tablet POC

A POC under development for distributing creatives to tablets with a Next.js admin, Android player, and Supabase/Postgres backend.

Surfaces

Admin + player

Stage

Technical POC

Challenge

Coordinate campaigns, devices, scheduled playback, and verifiable telemetry before a proven operational rollout.

Solution

The schema covers tenants, campaigns, creatives, temporal assignments, QR, proof-of-play, scans, and heartbeat; the Android player consumes schedules.

Available evidence

  • Implemented Next.js admin and Android player
  • Temporal assignments and playlists
  • Proof-of-play, QR scans, and heartbeat
Source
Uber_marleting_tablet@2653faa — STATUS.md, Supabase schema, Edge Functions, and Android player
Verified on

Limitations

No deployment or operational fleet use was proven.

Executive summaryTechnical detail
Mobility Ad Network — Tablet POC
PRISMA Project (demonstration)
In validation

PRISMA — Technical Oncology Prototype

A Next.js prototype with two demonstration cases, PostgreSQL/Prisma, Claude analysis with OpenAI fallback, and PubMed references.

Demonstration cases

2

Clinical validation

None

Challenge

Demonstrate an analysis and literature interface without presenting generated output as diagnosis, care recommendation, or clinical efficacy.

Solution

Analysis and literature routes connect demonstration data to Claude, OpenAI, and PubMed; the interface organizes output for technical exploration.

Available evidence

  • Two demonstration cases: liver and lung
  • Prisma/PostgreSQL persistence
  • Claude analysis with OpenAI fallback and PubMed references
Source
PRISMA@5a3232e — medicalAnalysisEngine, analyze/literature API, Prisma schema, and CI
Verified on

Limitations

No clinical validation, diagnosis, or care recommendation.

Executive summaryTechnical detail
PRISMA — Technical Oncology Prototype
Bellatrix
Delivered

Bellatrix System — Integrated Financial Operations

A production Flask financial and operational system using MySQL, Redis/Celery, Nuvemshop integration, dashboards, and automated deployment.

Checkpoint: sales

140

historical migration dated 02/06/2025; not a current KPI

Checkpoint: installments

222

historical migration dated 02/06/2025; not a current KPI

Checkpoint: records

512

historical migration dated 02/06/2025; not a current KPI

Challenge

Centralize financial and e-commerce operations while preserving history and maintaining synchronization and operational checks.

Solution

Financial modules, OAuth/webhooks, and Nuvemshop synchronization run with MySQL, Redis/Celery queues, and a GitHub Actions deployment pipeline.

Available evidence

  • Public health responded healthy during verification
  • Operational sales, expenses, and cash-flow dashboard
  • Documented historical migration checkpoint
Source
sistema_bellatrix@ffde968 — README, tests/integration, and https://app.falai.agency/health (verified 2026-07-15)
Verified on

Limitations

Health does not prove all historical synchronization remains intact or daily use of every module.

Executive summaryTechnical detail
Bellatrix System — Integrated Financial Operations
AutoGrid
In validation

AutoGrid — MIT Core and Private Cloud

A platform under validation consisting of an MIT Grid/DCA and backtesting core plus a private cloud with a Next.js dashboard and Telegram.

Connector exchanges

3

Binance, MEXC, and Bybit

Native strategies

2

Grid and DCA

Core license

MIT

Challenge

Clearly separate the open core from the private cloud without promising a public runtime or financial results.

Solution

The core provides API/CLI, Grid and DCA strategies, CCXT for Binance, MEXC, and Bybit, and backtesting; the cloud adds dashboard and Telegram surfaces.

Available evidence

  • MIT-licensed core
  • CCXT connectors for three exchanges
  • Backtesting with Sharpe and drawdown
  • Private cloud under validation
Source
autogrid-core@859e6b4 + AutoGrid cloud@98d08c5 — MIT core and private cloud
Verified on

Limitations

The documented domain did not serve AutoGrid during verification.

Executive summaryTechnical detail
AutoGrid — MIT Core and Private Cloud
Applied AI case studies | F.A.L A.I Agency