All work02 / 05
AI SYSTEMS · LIVE VENTURE

AI Growth Incubator.

Make AI useful in the work that repeats.

AI implementationWorkflow designBusiness systems

Screenshot of the project website

STAGELive venture and service offer
MY WORKService design · AI orchestration · workflow scripting
EXPLOREVisit AI Growth Incubator
01 / WHAT I BUILT

Practical AI systems for the work that repeats.

AI Growth Incubator helps founders and local businesses turn repeated work into practical AI and automation systems. Its business-development workflow connects prospect discovery, qualification, personalized outreach and follow-up tracking. The wider offer can include AI voice roles, vertical AI-powered analysis and hands-on digital marketing when a business needs more leads or implementation support.

02 / MY ROLE

My contribution.

I developed the service model and the AI-assisted research-to-outreach workflow, including record processing, review artifacts and a human approval step before sending.

  • Defined the offer around founders and local teams with repetitive business-development and operational work.
  • Designed prospect discovery, qualification, record normalization, personalized outreach and follow-up tracking as one workflow.
  • Created the service's diagnosis → design → deploy method and its public-facing experience.
03 / HOW IT WORKS

Inside the system.

The business-development system starts with public research, passes through structured qualification and record processing, and ends with approved outreach and tracked follow-up. OpenAI Codex coordinates research, evaluates source material, drafts messages and reconciles records. The service model adapts this approach to each client's workflow.

WORKFLOW STACKTHE BUILDING BLOCKS
AI orchestrationOpenAI Codex
ResearchExa search · webpage extraction · official websites
ProcessingJavaScript · Python · duplicate and contact checks
RecordsLocal JSON · prospect profiles · qualification notes
ReviewMarkdown · CSV · Excel artifacts
EmailHostinger Mail REST API through a connected tool interface
SYSTEM ARCHITECTUREFROM INPUT TO OUTCOME ↘
  1. 01Discover

    Exa and official sites surface possible prospects

  2. 02Qualify

    Codex evaluates source material and prepares structured records

  3. 03Prepare

    Scripts normalize contacts, check duplicates and create outreach artifacts

  4. 04Approve + send

    A person reviews the message before the email API sends it

  5. 05Reconcile

    Sent copies, receipts and follow-up records keep the campaign traceable

04 / IMPLEMENTATION

Under the hood.

The details that make the experience or workflow work in practice.

01

Structured prospect records

Local JSON keeps profiles, source notes, qualification decisions, message payloads and send receipts together across a campaign.

02

Normalization and duplicate checks

JavaScript and Python scripts clean contact details, compare records and produce batch artifacts for review.

03

Human approval before sending

AI prepares outreach; a person reviews the recipient and message before the connected Hostinger Mail interface sends it.

04

Traceable follow-up

The workflow can read incoming mail, verify copies in Sent and reconcile receipts with follow-up records. Markdown, CSV and Excel reports support review.

AI ARCHITECTURE

Where the model fits.

Orchestrator

OpenAI Codex coordinates research, source evaluation, outreach preparation and record reconciliation.

Research context

Exa search and webpage extraction retrieve public business information from official sites.

Control point

AI-generated outreach reaches a human approval step before the email integration sends.

Other use cases

The service can apply AI voice models to business roles and vertical AI-powered data analysis where they fit a client's workflow.

05 / WHY IT WORKS THIS WAY

Decisions with intent.

The thinking behind the architecture and product experience.

01 / DECISION

Preserve the research trail

Source material, qualification notes and structured records let a reviewer understand why a prospect was selected.

02 / DECISION

Put approval at the send boundary

Personalization benefits from AI drafting, while a human checks relevance and tone before a message leaves the system.

03 / DECISION

Use portable records and reports

JSON, CSV and Excel make campaign state easy to inspect, reconcile and hand off across tools.

06 / THE HARD PART

Keeping prospecting useful, personal and traceable.

Research sources are uneven, contacts can repeat and outreach can lose context as it moves between tools. Structured records and scripts address normalization and duplicates; source review and human approval keep the message grounded; receipts and Sent-folder checks close the loop for follow-up.

07 / CURRENT STAGE

Where it stands.

Live venture with an AI-assisted business-development workflow and a public service offer for founder-led and local businesses.

PROOF / EXPLORE

Visit the live site for the service model, workflow examples and engagement process.

Visit AI Growth Incubator
KEEP EXPLORINGNext projectMindframe