It starts with a half-day workshop for your leadership team — mapping the work worth handing to AI. It ends with 1:1 installs for the operators who'll actually run it: each one walks out owning agents built around their real workflow.
Companies buy tools that sit unused. Or they hire consultants who build something nobody on the team can run six months later. The agents I install survive me leaving.
One process. The same shape every engagement. Workshop earns the trust. Starter pack lets your team self-serve. 1:1 Role Installs go deep with the people who need it most.
Half-day on-site with your exec team and department heads. We walk every workflow that AI could take over.
Whiteboard the team's pain points. Rank by ROI. Lock the highest-leverage workflow as the live build.
Your team watches the agent get built live, on your data. They reverse-engineer it. They learn how to do it themselves.
The In-Seat AI Starter Pack — the methodology, the templates, the worked example. Your team can spin up new agents themselves.
Per-attendee recommendations within 24 hours: who specifically would benefit from a Role Install, what agent we'd build, what the ROI looks like.
The workshop earns the room. The 1:1 installs do the work. The retainer maintains the leverage.
Your exec team and department heads in one room. Map every workflow AI can take over, prioritise by leverage, build one working agent live. Walk out with a roadmap, a prototype, and per-attendee 1:1 recommendations.
From R40,000
Workshop fee credits toward 1:1 installs booked within 14 days.
Book a discovery call →I sit with one person — your finance lead, your e-commerce manager, your CFO — and we build the agents that take over the repetitive 60% of their week. They walk out owning every agent. They keep extending it after I leave.
From R20,000
Transformation Pack also available — 2–3 agents over 6 sessions, for roles where one agent isn't enough.
Book a discovery call →Tuning agents across multiple roles. Building new workflows as the business evolves. Acting as the de facto AI lead without the headcount. Default upsell after 3+ Role Installs at one company.
Pricing on the discovery call
Book a discovery call →Four layers. Each one solves a specific problem you've already hit with AI. Together they're how an operator owns an agent that survives me leaving.
A one-page contract that names the job, the inputs, the outputs, the receiver. Operators don't engineer agents. They write briefs. We build the first one together; from then on the operator owns it.
Context lives in files the operator can read — not buried in a chat thread that resets when you close the tab. Open the laptop tomorrow and the agent picks up exactly where you stopped.
Outreach lists, first drafts, research compilations, formatting passes, QA. Your judgment stays on the 25% that actually compounds — the creative call, the client conversation, the strategic choice.
What the agent did, what it used, what it skipped, what it flagged for review. You audit your own AI labour the same way you'd audit a junior. No black box.
Four tools. No bespoke black boxes. Everything I install your team onto, your team can keep using when I'm gone — real tools with documentation your operators can read.
The conversational layer. Your operator's AI partner — the one they already know and trust. Everything in the stack starts here.
The engine. Where the agents that run your repetitive 60% actually live. CLI, not magic. Briefs in, work out, receipts attached.
For the creative roles. Brand, marketing, design, content — agents that understand visual work, not just text. Built for the people who think in mood, not in markdown.
The second opinion. When a Claude agent ships a build, Codex audits it. Two models, adversarial. The receipt at the bottom of every Proof layer.
A sports-betting intelligence product that would normally need a content team, two analysts, an editor, a social media manager, and a part-time designer. I run it alone. Real customers. Live revenue. Every match-day, the agents pull the data, write the analysis, generate the image cards, validate the verdicts, and publish across 23 channels — without me touching a keyboard.
This is the proof of the methodology. Every Role Install I sell is a fraction of what's running here.
ROLES REPLACED → ~5–6 FTEs · COST → token spend + 1 director (me)
"Paul was with us for an hour before everyone was trying to book one-on-ones with him." Three department heads booked 1:1 installs by lunch. The pattern working as intended.
Read the case study →
When buyers ask "have you done this before," I open MzansiEdge running live. The methodology shipping in real time. No deck. No theory. The receipt.
In-Seat AI isn't zero people. It's zero new hires. The methodology doesn't replace your team — it makes each person on it capable of two or three more.
Paul Domanski. Cape Town. I build AI systems for a living and I install them — role by role — inside South African scale-ups and mid-market companies.
I don't lead with credentials. I lead with what's running in production. Two AI products built and operated solo. The methodology that built them — In-Seat AI — is the credential. The install is the product.
I don't replace your team. I give each person on it the team they'd otherwise have to hire.
What's your team shape? What's the workflow eating your week? Who's the person whose job AI could change the most this quarter? If there's a fit, we book the workshop. If not, I'll say so.
Bespoke AI systems for creative production. Installed into your employees. Not in place of them.
