Gozzy Nwogbo AI Systems Builder & Transformation Partner

Most AI never
leaves the demo. I build the version that runs

I come at it from business, not computer science, and that is what makes it useful. Based in Toronto.

Every item in the fit demo maps to something I have built. Drawn from 483 automations and the systems on this site.

See the work Let’s talk

Now · Jul '26

Building

Atomos

A governed AI knowledge layer where every output traces back to the rule that allowed it.

Reading

Rewired

McKinsey's AI playbook.

Listening

Icon

Brent Faiyaz.

Last shipped

Tender Radar

A hands-off agent that reads new federal tenders and emits a weekly can-bid / can't-bid digest.

01 About

A business head
with frontier-AI hands.

Business model analysis AI transformation strategy AI systems design Agent orchestration Retrieval & knowledge systems n8n Python LangChain Supabase / pgvector

I came at AI from business, not computer science. Business school taught me how to take a messy organization apart on paper: what it sells, where the margin lives, which problems are worth solving and for whom. That training runs underneath every system I build.

Two years ago these tools stopped being toys and became amplifiers, so I went deep while the frontier was still forming. Since then I have built daily: knowledge systems, agents, automation pipelines, real deliverables for real organizations. The systems I rely on every day were built the same way, which is why I can show them instead of describing them.

That combination is rare in one person, and it is the whole point. I sit between what AI can do and how organizations actually work, and I close that gap myself instead of handing off a requirements document and hoping.

The longer story

Gozzy Nwogbo · AI systems builder
Gozzy Nwogbo rising for a layup on an outdoor court

At a glance

Currently
Building Atomos and independent AI systems
Previously
Loop Tech Systems, Co-founder, Head of Strategy & Expansion (2023 to 2025)
Based in
Toronto, Ontario, Canada
Open to
Founding and greenfield AI roles, forward-deployed and solutions engineering, AI transformation

Worked with

MI-C3 Yale (intern) FSRA (advisory) Loop Tech (co-founded, dissolved) client work via Loop Tech: Bupa GlobalNHSBoston Medical Center

Built with

n8nClaude APISupabasePythonLangChainAstropgvector

Consulting projects at Ivey academic program work

ScotiabankCIBCAutism Society of AmericaBoys & Girls Club of London
02 Experience

Through the years, one practice.

  1. '20

    Research Assistant · Yale University

    Selected for research on incentive design. The root of a lasting interest in why people actually do things, which still runs through the persuasion and content work.

  2. '22

    Software Engineer · MI-C3 International

    Built workflow-optimization features on a cross-functional team. The year business questions turned into schemas and shipping discipline.

  3. '23

    Co-founder, Strategy & Expansion · Loop Tech Systems

    Co-founded a health-tech venture and ran enterprise engagements for Bupa Global, the NHS, and Boston Medical Center. Built the AI orchestration layer behind the platform.

    2.3x Client workflow throughput
  4. '25

    Strategy Consultant, IDIS · Ivey Business School

    AI and digital-transformation strategy for CIBC and national nonprofits. Each strategy deck shipped with a working system behind it, not just slides.

  5. now

    Independent AI systems · Building Atomos

    Building daily: knowledge systems, agents, and automation that run inside real businesses.

Full experience

03 By the work

Charging the frontier.

483 automation workflows 11,000+ knowledge entries 59 knowledge domains 107 session logs 2.3x client workflow throughput 3 days full priced tender
P.01 483 The automation library Working workflows across verticals, not slideware.

483 workflows spanning contract review, freight invoice auditing, finance operations, real estate, logistics, and tax. Each carries evals and a preferred trigger and storage pattern per use case, so a new build starts from a proven shape instead of a blank canvas.

P.02 11,000+ The knowledge corpus Entries I can read, search, and audit.

11,000+ knowledge entries across 59 domains, kept as plain text and cross-linked Karpathy-style so context crosses fields: a persuasion principle can inform an ad pipeline, a maintenance philosophy can shape an agent design. 107 session logs record how the system actually gets used, day after day.

P.03 50+ Experts, distilled Working methods pulled out of video and into reach.

The corpus holds distilled working methods from 50+ MasterClass instructors, broken down and cross-linked so their frameworks surface in the middle of real work instead of staying trapped in a lecture.

P.04 3 days The tender engine A priced package against a 4 to 6 week baseline.

A complete priced tender package for a six-country infrastructure group, delivered in 3 days against a 4 to 6 week team baseline. The speed is the same knowledge and automation base as everything else on this page, pointed at one deadline.

Counts are live file counts from the working system, not estimates.

04 Selected work

Work that runs.

knowledge · Active · the flagship

AI is only as useful as what you can put in front of it.

Second Brain / Open Brain

A working personal knowledge OS: an MCP server with semantic search over pgvector, daily capture, processing, and digest pipelines, holding 11,000+ entries across 59 knowledge domains. It runs my daily practice.

Read the case study
Generated print-style motif of a knowledge graph: linked terracotta nodes across a cream page
Knowledge-graph motif · generated in-house

And the rest of the bench, each ending in an outcome.

content · Active

A beautiful ad that does not land its message is a beautiful failure.

Obscura

An automated ad pipeline: a product blurb becomes a multi-scene storyboard, drift-free character renders, and a scored MP4, gated by an audit that measures whether a cold viewer gets the message.

Blurb to finished ad video, end to end

Pipeline

PythonSeedancefal.aiFFmpeg
consulting · Active

The numbers rebuilt, not accepted.

Bid & Investment Screening for an Infrastructure Group

A full priced tender package produced in 3 days for a six-country infrastructure group, plus an investment screen that caught a falsified growth claim before money moved.

Full priced bid in 3 days, against a 4 to 6 week team baseline

Tender package

ClaudePythonParametric estimationFinancial analysis
automation · Active

Most business AI does not need to be a chatbot.

Tender Radar

A hands-off cron agent that reads new federal tenders, decides can-bid or cannot-bid against a company profile with the single decisive clause, and emits a weekly digest.

918 open federal notices at last count

Live feed measured

PythonClaudeCanadaBuys dataCron
content · Active

Retrieval over an owned corpus beats generation from model memory.

Scribe / Scribe Scout

A content engine that drafts from an atom graph of my own knowledge corpus, plus a discovery front end that scans the last 30 days of online discussion and scores topics by argument intensity.

Every draft built from my own atom graph, not model memory

Grounding

PythonClaudeReddit / HN dataKnowledge graph
healthcare · Pilot Complete

Real early-stage building, with the failure modes left in the write-up.

Healthcare Workflow Automation

Early-stage clinical workflow automation, piloted hands-on across three Nigerian clinics, cutting administrative documentation time by roughly a third.

Documentation time cut by roughly a third in pilot

Administrative time

PythonLangChainFastAPIPostgreSQL

View all projects →

05 What I bring

The stack under the hood.

Six working disciplines. Every tool below is in live use; none of it is aspirational.

T.01 In live use

AI systems design

A system you cannot score is a demo. Evals come before features.

  • Claude API
  • MCP servers
  • Agent orchestration
  • Evals
  • Spec contracts
  • Observability
T.02 In live use

Retrieval & knowledge systems

Answers should come from knowledge you own and can trace, not model memory.

  • pgvector
  • Supabase
  • Semantic search
  • Embeddings
  • LangChain
  • Markdown corpora
T.03 In live use

Agent & workflow automation

An agent gets the minimum context and authority the task needs, scoped before it runs.

  • n8n
  • Python
  • Cron agents
  • Relevance AI
  • Webhooks
  • Telegram bots
T.04 In live use

AI transformation strategy

Measure in time saved first, then convert to money. Technology that cannot answer that is a hobby.

  • Workflow mapping
  • Pilot design
  • Change management
  • Adoption plans
T.05 In live use

Business model analysis

What it sells, where the margin lives, which problems are worth solving and for whom.

  • Unit economics
  • Market analysis
  • Competitive analysis
  • Financial modeling
T.06 In live use

Investment & bid analysis

The numbers get rebuilt, not accepted.

  • Tender strategy
  • Bid pricing
  • Bottom-up estimates
  • Diligence
06 Contact & availability

Let’s build something real.

Open to founding and greenfield AI roles, forward-deployed and solutions engineering, and AI transformation work. Based in Toronto, remote works fine. Say what you are working on and you will get a straight answer.

¶ Send a note

Direct to my inbox

What is this about?
Email directly

The working inbox pipeline is on its way. Until then this opens the note in your own mail app, already addressed. Direct email works just as well.

¶ Correspondence

Three channels

Email nwogbo.gozzy@gmail.com LinkedIn linkedin.com/in/gozzy-nwogbo GitHub github.com/gozzy-nwogbo