Skip to main content
00 / INIT SCROLL 000%
MARK CARMONA
AI · AGENTS · AUTOMATION
Available for remote AI builds worldwide

Mark Carmona — AI Automation & AI Agent Engineer

I build AI agents that turn work into systems.

I design and build production AI agents, workflow automations, CRM systems, multi-agent orchestration, and self-hosted agent infrastructure for businesses in the Philippines and remote teams worldwide. My work includes ClawBridge, ClawBridge Codex, OpenClaw / Hermes-Agent, APIs, websites, and operational automations with clear ownership and trust boundaries.

Self-hosted agents Container-isolated runtime Client-owned code Websites + backend automation AI outreach & follow-up
clawbridge — agent live
$ npx clawbridge-codex
→ verifying Docker + Node 20
→ codex login --device-auth
→ building agent container
→ starting service (launchd)
✓ agent live on 12 channels
@Andy brief me on the pipeline at 9am
Role AI Automation Engineer
Focus Agents · Automation · Sites
Status Taking projects
SCROLL
SELF-HOSTED AGENTSAI OUTREACHAUTOMATIONWEBSITES SELF-HOSTED AGENTSAI OUTREACHAUTOMATIONWEBSITES
CLAWBRIDGEOPENCLAW / HERMES-AGENTCLAWBRIDGE CODEX CLAWBRIDGEOPENCLAW / HERMES-AGENTCLAWBRIDGE CODEX
ORIGIN
01

About

From manual operations to business AI systems.

I started in the kind of admin work most businesses quietly drown in: repetitive follow-ups, data entry, customer questions, scheduling, reports, and tool-hopping. That background is why I build automation around real bottlenecks instead of chasing demos.

Today I build AI systems that receive a business task, reason through the workflow, call tools, update records, send messages, and keep improving through memory. I work across ClawBridge, ClawBridge Codex, OpenClaw / Hermes-Agent, Claude/Codex-based agents, custom APIs, and production websites.

My sweet spot is practical: self-hosted agents, CRM automations, AI outreach, booking flows, web scraping, reporting, and landing pages that actually help a business get leads, serve customers, and save hours every week.

What I build
  • ClawBridge & ClawBridge Codex deployments
  • OpenClaw / Hermes-Agent systems
  • Multi-channel AI agents
  • Lead follow-up & AI email outreach
  • Websites, landing pages & funnels
  • CRM, calendar, voice & messaging workflows
  • Data scraping, reporting & API integrations
Why hire me
  • I build working systems, not chatbot mockups
  • Client-owned infrastructure and code
  • Container isolation and credential separation
  • Fast deployment with room to customize
  • Strong design plus backend automation
  • Clear outcomes: leads, bookings, replies, reports
Current build surface
ClawBridge ClawBridge Codex OpenClaw / Hermes-Agent Codex CLI Claude Agent SDK Docker Hindsight MCP Resend GoHighLevel TypeScript Node.js
FLAGSHIP
02

Featured Work

A self-hosted AI agent platform I designed, built, and ship — not a demo. It's the foundation under most of the services below.

Flagship · Open Source · MIT

ClawBridge Codex

self-hosted, multi-channel AI agents powered by OpenAI Codex

A single Node host orchestrates per-session agent containers, so AI assistants live in your messaging apps and tools — running in isolated Docker sandboxes with credentials kept host-side. Built as a smaller, auditable alternative to heavyweight agent frameworks: keep the useful parts while staying small enough to actually understand.

1,346
Commits
36
Tags
86
TypeScript
12+
Channels
What it does
Multi-channel agents. WhatsApp, Telegram, Slack, Discord, Teams, email + more from one runtime.
Container isolation. Codex agents run in Docker with explicit mounts; channel secrets stay host-side.
Semantic memory. Optional Hindsight retain / recall / reflect injects context before each turn.
Multi-agent orchestration. Agents delegate tasks, pass files, and run parallel workflows.
Tools + skills. MCP connections, browser automation, web search, and scheduled jobs.
Architecture
messaging apps host router inbound.db container (Bun · Codex) outbound.db delivery apps
TypeScript Node 20 Bun Docker OpenAI Codex MCP Hindsight
03

AI & Automation Use Cases

01 / 15

Representative systems across agents, automation, outreach, data, and websites. Open any workflow to see how its signals, decisions, tools, and handoffs connect.

01Flagship

Multi-Agent Orchestration

Specialist agents delegate, research, draft, schedule, analyze, and consolidate results in one workflow. Built for complex work that needs parallel execution without manual coordination.

specialist agents · parallel workflows · Hindsight memory
02Insight Engine

Intelligent Data Scraping

Automated extraction of competitor pricing, market data, prospect lists, and industry signals from web sources, cleaned into useful reports or CRM-ready records.

market data · lead lists · competitor tracking
03Zero Clicks

Workflow Automation

Connect CRM, Slack, email, calendars, documents, and internal tools into workflows that trigger, update, notify, and report without repetitive manual clicks.

CRM automation · Slack alerts · tool workflows
04Always Scheduling

Calendar & Booking Agent

An AI agent that qualifies prospects, checks availability, books meetings, confirms appointments, and handles reschedules through calendar, CRM, and email integrations.

Google Calendar · GoHighLevel · rescheduling
05Omnichannel

AI Marketing Campaigns

Coordinated campaigns across email, SMS, CRM, and ad platforms with AI-assisted personalization and follow-up logic at scale.

email · CRM nurture · SMS
06Never Forgets

Memory & Context AI

Agents remember client preferences, lead history, conversation context, and workflow rules so every interaction feels less generic over time.

Hindsight · client memory · semantic recall
07Live 24/7

AI Receptionist

A human-sounding receptionist that answers inbound calls, qualifies leads, books appointments, transfers hot leads, and logs conversations to your CRM.

Retell AI · Twilio · CRM logging
085-Star Pipeline

Review Request Agent

After a job or service is completed, the agent sends personalized review requests for Google, Yelp, Trustpilot, or your preferred platform.

Google reviews · Yelp · reputation
09Speed Wins

Lead Follow-Up Agent

Respond to every inbound lead within seconds via SMS, email, or chat, answer questions, qualify intent, and book appointments before competitors react.

instant follow-up · SMS/email · booking
10Zero No-Shows

Appointment Reminder

Automated reminders through SMS and email before appointments, including confirmation, cancellation, and reschedule handling.

SMS · email · calendar sync
11Pipeline First

Lead Generation Agent

A prospecting agent that finds leads, qualifies fit, reaches out, follows up, and books demos so your pipeline keeps moving while you close.

prospecting · lead scoring · demo booking
12Fully Connected

API & Integration Layer

Custom API integrations that connect your proprietary tools with third-party platforms. If it has an API or webhook, I can usually wire it in.

custom APIs · webhooks · MCP
13Always Informed

Automated Reporting

Scheduled reports from revenue, operations, marketing, support, and sales data, delivered to email, Slack, or your dashboard before the day starts.

daily reports · dashboards · business intelligence
14Launch Ready

Websites & Landing Pages

Custom websites and landing pages designed to capture leads, explain your offer, and connect directly to your AI agents and automations.

mobile-first · SEO-ready · agent-connected
15Cold Email

AI Email Outreach

The outreach agent checks each lead website, finds specific personalization hooks, writes tailored cold emails, sends them, logs replies, and follows up automatically.

website research · AI copywriting · Resend
Workflow map

Workflow

Action Decision Feedback / retry
04

Build & Delivery Loop

01
Discover
Find the bottleneck
02
Blueprint
Map tools & data
03
Build
Agent + website
04
Test
Real scenarios
05
Deploy
Monitor & improve
Production setup

I build for the place your customers already are: email, WhatsApp, Telegram, Slack, Discord, SMS, phone, calendar, CRM, and the website.

deployment surface
Runtime: VPS or Mac Mini, Docker isolated
Channels: chat, email, voice, CRM, webhooks
Client ownership

The strongest systems do not trap you. Your agent, memory, website, and automations should live in your repo or infrastructure with clear handoff paths.

You own the code and deployment path
Credentials stay separated from agent sandboxes
New channels and skills can be added as the business grows
RECALL
05

Memory & Context AI

ClawBridge-style memory keeps agents useful across conversations without exposing credentials or client data to the wrong runtime. Hindsight runs host-side — it recalls relevant context before an agent wakes, retains useful facts after delivered turns, and reflects on memory during maintenance sweeps.

L1
Recall Before Action
Host receives a message, queries the right memory bank, then injects only relevant context into the agent turn.
L2
Retain After Delivery
After the response is delivered, useful summaries, preferences, and business facts are stored for future recall.
L3
Reflect & Consolidate
Background reflection consolidates redundant memories, promotes important facts, and prunes stale context.
Credential-safe memory boundary
Hindsight and channel credentials stay host-side; agent containers receive context, not raw secrets.
Client Profile
  • business goals
  • offer details
  • brand voice
Lead Context
  • website signals
  • pain points
  • personalization hooks
Project Knowledge
  • agent behavior
  • workflow rules
  • integration notes
Channel Memory
  • WhatsApp threads
  • Slack discussions
  • email replies
Outreach History
  • sent campaigns
  • reply status
  • follow-up timing
Guardrails
  • do-not-contact rules
  • compliance notes
  • client approvals
Need an agent, automation, or website?

Tell me the task you keep doing by hand. I'll tell you what to automate first.

Let's build it
ETHOS
06

Engineering Philosophy

OWN
Client-Owned
The best business automation is not rented forever. Your agent, code, data, website, and deployment path should be portable and understandable.
01
ISO
Isolated
Agents run in containers with explicit mounts. Credentials stay host-side, and the agent only sees what the business intentionally gives it.
02
CLR
Understandable
Small systems beat mystery platforms. I prefer clear source, simple routing, and architecture a client can audit without a giant team.
03
CUS
Bespoke
Your workflow should shape the software. I use code changes, custom skills, and direct integrations instead of forcing every client into the same template.
04
MEM
AI-Native
A good agent system can help operate, debug, document, and improve itself. Codex, ClawBridge, and memory make that practical.
05
ROI
Outcome-First
Every build has to move a real number: faster response, more booked calls, fewer no-shows, better reports, or less manual work.
06
07

Technical Stack

Agent Platforms
ClawBridge
ClawBridge Codex
OpenClaw / Hermes-Agent
Claude / Codex agents
Runtime & Code
TypeScript / Node 20
Codex CLI
Bun agent runner
Python automation
Memory & Data
Hindsight
Postgres / pgvector
SQLite session DBs
Semantic recall / reflect
Channels
WhatsApp
Telegram
Discord / Slack / Teams
Linear / GitHub / Email
Integrations
MCP
Chat SDK
Google Workspace
HubSpot / GoHighLevel
Outreach
Resend email
Lead website research
AI personalization
Follow-up sequences
Infrastructure
Docker isolation
VPS / Mac Mini
launchd / systemd
Traefik / SSL
Quality
pnpm
Vitest
ESLint / Prettier
Health checks / doctor
Get in touch

Ready to turn
manual work into
a system?

I can help you scope the agent, automation, outreach engine, website, or integration that will move the fastest for your business.

Project-based builds · Fast scoping · Self-hosted and client-owned when needed

Get in touch