AI Strategy
AI that people actually use.
I design agentic workflows, get teams to adopt them, and build with AI myself. Every system below started with a real business problem, not a trend.
From prompt to workflow
Each step came from a problem the previous step could not solve. That path shapes how I advise on AI: start small, learn what the work actually needs, then build the next layer.
- 1. Prompt engineering
- On client engagements, I wrote prompts by hand for messaging, content, and SEO analysis. The repeated effort showed me that the work needed a dedicated agent.
- 2. A purpose-built agent
- I built an agent trained on copywriting practice, then added messaging, research, and brand guidelines. Using it showed me that one agent could not carry strategy, quality control, and production at once.
- 3. An agentic workflow
- I split the work across specialized agents with defined handoffs and independent quality gates, so each stage owns one kind of failure.
What I've built at work
At IDC, I lead content marketing and run these systems in production.
- A multi-agent content system
- Seven stages, from strategy through media guidance, SEO, writing, persona validation, answer-engine optimization, and final quality control. Three independent gates can send an asset back for rewrite, any agent can run on its own, and a full quarterly content plan takes 20 to 25 minutes. The whole content marketing team uses it.
- A synthetic persona panel
- AI versions of a CEO, CFO, CIO, CTO, CMO, industry analyst, and journalist that score content, critique what works and what does not, and vote between headlines and copy. The workflow reads the feedback and orders the rewrites.
- An enterprise thought-leadership agent
- Built on 300+ FutureScape predictions, IDC's messaging, and its brand guidelines, so analysts could turn research into blogs, social posts, client emails, and video scripts that stay on brand. About 500 employees had access. I designed it, trained it, launched it, and administered it.
- Answer-engine strategy
- An AEO and GEO approach that writes some content for people and some for AI assistants, plus a Canonical Definition Page program so AI can extract and cite IDC's research. See the live definitions library.
How I approach AI strategy
- Start with the process
- Map the work as it happens today, and find where a delay, a handoff, or a repeated judgment costs the most.
- Decide where an agent belongs
- Not every step needs one. Agents earn their place where the work is repeatable and the quality bar can be written down.
- Govern the output
- Give each failure mode its own gate, whether that is audience fit, accuracy, brand voice, or how readers and AI systems will receive the result.
- Plan for adoption from day one
- A tool nobody uses returns nothing. Training, access, and a clear first use case matter as much as the build.
What I build on my own time
I do not only advise on AI. I ship with it, across very different kinds of work.
- eko.4m
- Four albums plus EPs and singles on Spotify and Apple Music. Human-written songwriting, AI-assisted production, and a lyric-analysis agent I built across rock, country, pop, hip-hop, and EDM. Hear it at eko4m.com.
- Websites built with AI
- This site, eko4m.com, the luminousblue consulting site, and a multi-page student-athlete profile site for a family member covering soccer, cross country, orchestra, and academics. All designed with AI and hosted on GitHub.
- Native iPhone apps
- Two apps in development, a card game and a task-management app for a family member.
Hiring for AI strategy or workflow integration?
Let's talk.
Full-time roles.
Speaking.
Storytelling.
Collaborations.