Custom marketing technology Built for small business

Run your marketing.
Own the system.

A marketing stack built around your workflow.

I help small businesses connect their knowledge, tools, production, approvals, publishing, and reporting in one system their team can run.

Use the software you already rely on, add custom capabilities where they matter, and keep day-to-day marketing in your team’s hands.

Learn how the parts fit
Your workflow. Your controls. Your system.

15+ years across marketing, analytics, and systems

Based in the Philippines Experience with US and Australian teams

Your marketing should not have to fit someone else’s template.

Small businesses often outgrow a patchwork of generic tools, manual handoffs, and knowledge that lives in people’s heads.

01

Keep business context with the work

Put brand knowledge, campaign decisions, customer insight, and operating rules where your team needs them.

02

Connect the marketing stack

Link production, approvals, publishing, tracking, and reporting instead of coordinating every step by hand.

03

Build the missing layer

Add focused interfaces, integrations, and AI-assisted capabilities when off-the-shelf software cannot support your workflow.

The result may combine tools you already use with custom software built for the parts that make your operation different. Your team gets the controls, documentation, and training to run it.

01 / Selected systems

Built from real constraints,
not hypothetical prompts.

What needed to change, which decisions mattered, and what was actually tested or delivered.

Delivered and handed off

Production workflow & team handoff

AI-assisted legal content operation

A production workflow the team could run.

ControlGoogle Sheets
ProductionAI + automation
OutputWordPress

Research, drafting, imagery, formatting, issue monitoring, and publishing were connected behind a familiar Google Sheets interface.

SOPs and training enabled nontechnical staff to manage routine production without the builder coordinating every step.

What this demonstrates

Design the operating interface, documentation, and handoff alongside the automation—not after it.

Reported outcome

A fragmented production process became a repeatable operation with documented procedures and team ownership of routine work.

See the operation

The full case distinguishes reported throughput from unverified content-quality, search, and business outcomes.

On hiatusIndependent project

Data reuse & system-wide control

Real-time multi-asset trading architecture

Shared data. Coordinated decisions.

Per instrument1m source
Local resampling5m · 15m · 1h…
AuthorityPortfolio risk

Separate processing paths across assets and timeframes overwhelmed the available server. The redesign reused one-minute source data per instrument and derived larger timeframes locally in Python.

A portfolio-level controller governed allocation and trading permission above the individual strategies.

What this demonstrates

Find the architectural constraint before adding infrastructure, then place controls where the combined consequences appear.

Reported outcome

Multi-coin operation resumed on the available VPS after the data redesign.

Read the architecture case

An architecture case—not a claim of verified investment performance or measured compute savings.

02 / Build your marketing system

Your workflow first.
Then the technology.

I turn the way your business plans and runs marketing into a practical stack your own team can operate.

That may mean connecting existing tools, building a focused internal application, introducing a controlled AI workflow, or developing the system in manageable stages.

  1. 01

    Map the marketing operation

    Identify who does the work, what they need to know, which tools they use, where work slows down, and what a useful result looks like.

  2. 02

    Build and connect the stack

    Keep what already works. Add the interfaces, integrations, shared context, AI components, and controls the workflow actually needs.

  3. 03

    Put it in your team’s hands

    Test the complete workflow, document the operation, train its users, and define what your team runs versus what needs technical support.

Discuss your marketing workflow

03 / Learn to build in-house

See how the system works.
Build with better judgment.

Running marketing in-house starts with understanding what to connect, what context the system needs, and which decisions should remain human.

These free field notes unpack decisions from the projects—not just the tools used to implement them.

Available now: free field notes. Short conceptual notes, not full implementation tutorials. Step-by-step training, implementation assets, and deployable examples will be listed when ready.

Have a workflow you want your team to learn?

Tell me what they need to accomplish, what they already know, and where implementation becomes difficult.

Share a training need

04 / The operating method

Design the complete workflow.
Not another disconnected tool.

Marketing experience shapes the build: begin with the commercial job, understand how the work moves, and choose technology only after the real constraint is clear.

01

Find the constraint

Define what the business needs to make possible and what is actually blocking it.

02

Connect the context

Bring the right knowledge, data, and operating rules into the workflow.

03A

Generate where variation helps

Use AI for creative exploration, drafts, imagery, and other probabilistic work.

03B

Control where certainty matters

Use software for state, approvals, storage, routing, and publishing—and people for consequential decisions.

04

Test the whole workflow

Evaluate whether the complete operation produces a useful result, then refine it.

A working component is not enough.
The whole workflow has to work.

05 / After the handoff

Your team runs the marketing.
Support keeps the stack moving.

Owning the operation does not mean your team must maintain every integration, model, or infrastructure component alone.

Ongoing support can be scoped around

  • Maintenance and compatibility fixes
  • Infrastructure administration
  • Integration and model-provider changes
  • Agreed upgrades as the operation evolves

Support responsibilities, response expectations, third-party costs, and new development are defined separately.

Discuss a build and its support needs

06 / Marketing experience behind the build

Built by someone who understands
the work behind the stack.

I began in SEO, paid acquisition, websites, and analytics. The work expanded into customer journeys, reporting, team coordination, automation, and product development.

That experience helps me connect the technology to the marketing decision, handoff, or business result it needs to support.

Find opportunity before increasing spend

~1,000 → nearly 4,000Monthly bookings/orders

At a service marketplace, analysis of historical acquisition data informed campaign restructuring and additional organic activity. Reported monthly volume reached nearly 4,000 within the first quarter after joining, without additional advertising spend.

Improve the economics of acquisition

$45–$60 → $9Reported cost-per-lead low, in USD

An established solar acquisition campaign reached a reported cost-per-lead low of $9 following restructuring and optimization—not a sustained average.

Make reporting useful and repeatable

Half a day → minutesRecurring reporting preparation

Custom analytics reports and combined dashboards reduced recurring preparation from almost half a day to minutes. Following the registration-to-bidding journey also exposed an existing reporting error.

Historical, self-reported outcomes from previous engagements; not independently audited results or guarantees of future performance.

About Borj

I’m Borj.

Illustrated Borj character working on a laptop

I work across marketing operations, customer journeys, data flows, software implementation, and team handoff.

My focus is helping small businesses turn the way they market into systems their own people can use and improve.

AI expands what I can explore and build. Responsibility for the workflow, its decisions, and its results remains human.

07 / Next step

How should your marketing
work together?

Tell me how your team runs marketing today, which tools are involved, and where the workflow breaks down or slows down.

You do not need a finished technical specification.

After your inquiry

I’ll review the workflow and discuss whether a focused prototype, custom build, or another next step makes sense.

jvcatingub@gmail.com Pampanga, Philippines · Remote collaboration

Who will use it? What tools or processes are involved? What would a useful result look like?

Budget and timing Optional

Project case / Active development

Klaroha / BOS

The operating problem

A simple marketing request can require a strategist, writer, designer, and channel operator to reconstruct the same context. Brand knowledge and campaign decisions become repeated briefing work at each handoff.

The design decision

Separate the operator’s request from the context needed to fulfill it. Structured brand briefs and brand kits supply reusable information to production stages. Models generate concepts and creative assets; software coordinates stages, approval status, storage, and distribution.

What was tested

The same generic instruction was tested across a memorial-garden narrative, a Klaroha product walkthrough, and a fictional motorcycle-brand image. Completed publishing tests involved Meta, YouTube, and TikTok across those scenarios—not necessarily every output on every platform.

Status and limits

These reported tests demonstrate contextual generation and completed publishing workflows—not scaled customer adoption, sustained campaign performance, or months-long historical learning. Some interface artifacts are prototypes rather than evidence of connected backend functionality.

Does your team repeatedly reconstruct the same brand and campaign context?

Discuss your marketing workflow

Delivered and handed off

AI-assisted legal content operation

The operating problem

Research, drafting, image preparation, formatting, issue tracking, and publication were fragmented activities. The requirement was not simply to generate more text, but to create an operation nontechnical staff could manage.

The design decision

Use Google Sheets as the familiar control surface and connect the production stages behind it. The workflow combined OpenAI-supported prompt chains, Zapier and Make, later n8n, and WordPress-ready output.

The handoff and reported result

SOPs, documentation, and training enabled staff to manage routine production without constant involvement from the builder. The historical account reports throughput reaching hundreds of published, indexed articles per day at peak.

Evidence boundary

Throughput does not establish content quality or commercial impact. Ranking movement, traffic, leads, legal accuracy, uptime, and article-quality metrics were not supplied. Workflow constraints were intended to reduce unsupported statements, not guarantee factual or legal accuracy.

Have a production process your internal team needs to run more independently?

Discuss your operating workflow

Independent project / On hiatus

Real-time multi-asset trading architecture

The operating problem

Manual monitoring limited how many opportunities one person could follow. Automation expanded that reach, but processing multiple assets across seven timeframes introduced duplicated work that overwhelmed the available VPS.

The design decision

Collect one-minute source data per instrument and derive larger analytical bars locally in Python, rather than maintaining separate ingestion paths for every timeframe. The redesign enabled multi-coin operation again on the available server, according to the project account.

The second constraint

Independent strategies could make valid local decisions while creating unacceptable combined exposure. A centralized portfolio controller therefore governed allocation, loss tolerance, and permission to continue trading.

Status and limits

Exact compute savings and sustained trading returns were not measured for publication. Development is on hiatus because of capital constraints and testing costs. This is an architecture case, not evidence of verified investment performance.

Working through duplicated data processing or decisions that need shared limits?

Discuss a systems problem

Free field note / AI-assisted development

Plan an AI-assisted build without losing direction.

Start with one coherent description of the intended operating model. Give that same problem to multiple agents or models, then compare where their recommendations conflict. The useful output is not a vote. It is the human-led synthesis of assumptions, risks, and alternatives into a working brief.

The brief remains alive during implementation. Decisions, rejected approaches, bugs, and changes in direction enter the backlog so architectural memory is not lost between coding sessions.

This keeps three activities distinct: AI helping plan a product, AI helping write product code, and AI operating inside the customer’s workflow.

Next: generation and workflow control

Free field note / Workflow design

Separate creative generation from workflow control.

Creative concepts, copy, imagery, and generative reasoning benefit from probabilistic output. Brand selection, approval state, record persistence, file routing, allocation, and publishing permission usually do not.

A resilient AI workflow makes this boundary explicit. Structured context enters each creative stage, people approve consequential transitions, and deterministic orchestration records what happened and controls what may happen next.

The goal is not to remove people from the process. It is to reserve human attention for decisions while software handles repeatable coordination.

Next: find duplicated work

Free field note / Data architecture

Find duplicated work before adding infrastructure.

The first multi-asset trading design treated each coin and timeframe as a separate live-data problem. As the number of instruments and intervals increased, duplicated ingestion and calculation overwhelmed the server.

The redesign used one-minute data per instrument as the common source and derived five-minute, fifteen-minute, hourly, and larger bars locally. That turned a scaling problem into a data-architecture problem.

The broader lesson applies outside trading: find the smallest reliable source of truth, then derive views from it instead of building independent pipelines that repeat the same work.

Back to the learning library