MARKETING AI + OPERATIONS
Product systems, enterprise enablement, organic growth, and startup launches.
02 / HOW I WORK
Sources, decisions, review, and ownership stay attached to the work. AI speeds up research and variation. A person still owns the final call.
Company documents, approved claims, audience behavior, and product constraints.
The problem, owner, audience, proof, review path, and success condition.
The product, campaign, or workflow ships with review inside the process.
Adoption, response, and operating friction decide what changes next.
03 / PRODUCT SYSTEMS
The walkthroughs show the actual flow. The links open the live builds and public code where it is available.
01 / Compliance aware content product
A weekly content system for mortgage loan officers who need to show up online without spending every Friday writing, designing, and rebuilding the same compliance handoff.
THE ISSUE
A blank AI chat can write a caption, but it does not know the loan officer's brand, required disclosures, or the phrases that should trigger review. Graphics happen somewhere else. Compliance notes live in another document. The result is more copy and paste work and an easy reason to skip posting.
Solo mortgage loan officers and small teams that handle their own marketing.
Managers and compliance reviewers who need the final claim, visual, and source together.
Turn a weekly idea into branded content that is easier to review, save, and reuse.
WHY + HOW
At First Horizon Bank, I ran a program with 350+ advisors and a library of 500+ approved posts. The biggest improvement came from changing the workflow. We used standard inputs, known claims, repeatable review, and a clear handoff. I built CompliPost to make that structure usable by one loan officer without an enterprise platform.
A user starts with a weekly plan or a content idea. CompliPost creates platform ready copy and a branded visual, checks the draft against a federal baseline and workspace rules, then keeps the findings with the post. The user can revise, save, export, or create a review packet. The product does not auto publish or promise legal approval.
PRODUCT DECISIONS
The planner gives the user a useful starting point before asking them to write a prompt.
Name, colors, logo, headshot, NMLS details, and voice can carry across every export.
The prompt avoids known high risk phrases. A rules pass then shows matched text, severity, citations, and suggested fixes.
Final visuals, captions, findings, and disclosures can travel together in a reviewer ready packet.
Solo founder. Product direction, user research, positioning, workflow design, frontend and backend build, and launch.
A working vertical SaaS product that turns enterprise publishing lessons into a self service workflow for one loan officer.
CompliPost is a review aid. It does not claim legal approval or publish on the user's behalf.
02 / Guided learning product
A path based learning product that teaches nontechnical people how to use AI at work, practice the judgment, and save work they can reuse.
THE ISSUE
Prompt tips cover the easiest part. A beginner also has to learn where context goes, what stays private, when to check a claim, how to recover from a bad answer, and how to use the same habits in their role. A flat course list puts all of those decisions back on the learner.
Founders, operators, and nontechnical teams learning AI for real work.
Beginners enter through shared foundations. Experienced users can use placement and test out paths.
A visible route, one next lesson, role specific practice, and work saved locally.
WHY + HOW
The first version was a gamified AI course. Testing showed that the learner had too many tabs to choose from and could see advanced work before basic habits. I cut the interface back to Trail and Library, made the foundations cumulative, and kept future lessons visible without making them selectable.
The current graph has 226 production lessons, 32 shared modules, 12 role routes, and 6 specialties. Each lesson combines ordered teaching and practice with source and review metadata. Route rules keep prerequisites in sequence. Local progress and Workbench artifacts let someone learn without creating an account.
PRODUCT DECISIONS
Trail opens on the current lesson while the full route stays visible as a map of what comes next.
Context, files, memory, boundaries, skills, and connections come before specialty work.
Matching, ordering, fill-in, multi-select, and applied projects check whether the idea can be used.
Progress, bookmarks, and Workbench artifacts stay on the device with export and delete controls.
Product concept, curriculum architecture, interaction design, content system, and full implementation.
A complete learning system with route logic, lesson data, applied practice, local progress, source review metadata, and a working learner interface.
Standard page analytics are enabled. Learner progress and Workbench content stay local to the device.
THREE SUPPORTING SYSTEMS
GTM OS keeps sources attached to answers. SignalScope does the same for briefs. Flowcraft does it for content drafts.
03 / Source grounded GTM workspace
A cited answer workspace for scattered positioning, pricing, objection, and launch documents. The team can inspect the source before using an answer.
04 / Market intelligence pipeline
A live monitor that collects public market signals, filters the noise, and writes a cited brief that can be checked before it becomes a recommendation.
05 / AI content planning workflow
A campaign planning workflow that turns source material into themes, content pillars, and review ready channel drafts.
04 / FIRST HORIZON BANK
I inherited slow reviews, uneven adoption, and disconnected content. I rebuilt the operation from business intake through publishing, field enablement, and reporting.
THE SYSTEM I BUILT
Mortgage, Wealth, Commercial, Recruiting, Brand, and Community priorities.
Campaign themes, message frameworks, media formats, and reusable series.
Standard submissions, preapproval, auto review, manual review, and an audit trail.
500+ approved posts, publishing rules, campaign subscriptions, and content retirement.
Onboarding, live training, playbooks, FAQs, and one on one coaching.
Views, clicks, engagement, earned media value, adoption, and monthly business updates.
AUDIENCE + CHANNEL LOGIC
I changed the framing, approved detail, and call to action for each audience while keeping every version ready for field advisors.
PHYSICIAN MORTGAGE EXAMPLE
I narrowed the promise, used a direct headline, listed the approved benefits, and gave advisors a clear next step. The field could use the copy without rewriting the product claim.
PROGRAM RESULTS
05 / ORGANIC GROWTH + LAUNCHES
Product launches, creator partnerships, owned channels, and community programs built for a competitive VR audience.
01 / BETA TRAILER
I wrote the script, planned the shoot, captured gameplay, filmed, edited, published, and handled distribution.
29 releases / ~1.65M lifetime viewsWatch video
02 / STEAM EARLY ACCESS
I scripted, filmed, and edited the launch piece, then turned it into platform specific clips, posts, copy, and calls to action.
Watch videoAEXLAB CHANNEL SYSTEM
I ran two X profiles, managed the CEO's LinkedIn, built recurring series, coordinated cross promotion, and reviewed performance.
~500K monthly organic X impressions
~35K monthly CEO LinkedIn views
8 recurring content series
100,000+ combined social community
STUDIO MINT
10M+organic views in client niches through positioning, recurring formats, channel planning, testing, and performance review.
06 / CONTACT
Marketing AI and operations lead. Paris and US remote.