DWAI · workflow-first AI systems Founder studio · Cape Town

Choose your DWAI section.

Start here. DWAI Solutions is the product route. DWAI Consultancy is the paid audit, implementation, and organization rollout route.

Below the two sections, the rest of the homepage explains the thesis: workflow first, owner gates before automation, and no unsupervised risky actions.

Solutions: $297 Mac AI Operator Setup Kit · Consultancy: paid audits scoped manually · no outcome promises · no unsupervised risky actions

Two ways to work with DWAI

Buy the system when you want to run it yourself. Hire the consultancy when the business needs Dylan's judgement.

DWAI Solutions

Self-serve AI operator systems: setup path, templates, checklists, workspace structure, owner review queues, PASS/HOLD gates, a first-install issue loop, and a workflow guardian handoff you can run on your own Mac after the first workflow is established.

  • Best when you want the setup path and product materials.
  • Manual checkout and ZIP delivery.
  • Primary current offer: Mac AI Operator Setup Kit v0.2.
View DWAI Solutions →

DWAI Consultancy

Paid business/workflow audits and consulting: Dylan uses AI agents as analysis leverage to review your company, identify workflow bottlenecks, map safe agent opportunities, and decide whether the right next step is a roadmap, the Blueprint/Setup Kit, or an organization-wide employee rollout.

Use this route when the problem is bigger than “send me the kit.”

View DWAI Consultancy →

The thesis

AI doesn't fix a workflow you can't describe.

Random AI chats are not a system. A clever prompt is not a workflow. A SaaS subscription is not a process. Most "AI problems" founders bring us are actually undefined work: nobody owns it, nobody can describe it end-to-end, and nobody can tell whether the output is correct.

Tool-first automation makes this worse. You wire something fast, it touches accounts and money, and the first time it fails, you can't tell whether the failure was the model, the tool, the data, or the missing approval gate.

So we work the other way around. Workflow first. Manual first. Track what breaks on the first install. Patch the source. Then — only when the manual path is stable, owned, and testable — package the stable workflow into a local operator system with a workflow guardian and owner gates.

Why agents fail

Five failure modes we see every week.

01

Nobody owns the workflow.

An AI sits between three people and a folder. When something breaks, no human is accountable for fixing the inputs, reviewing the output, or deciding when to stop.

Ownership
02

Context is stale.

The agent runs on screenshots, half-pasted docs, and an outdated brand voice file. Nothing is the source of truth, so every output drifts.

Inputs
03

Output can't be verified.

There's no acceptance criteria. No checklist. No diff. So a "good" run and a "bad" run look identical, and quality regresses without anyone noticing.

Acceptance
04

Automation touches money, accounts, secrets, or public posting without a gate.

The model is one prompt away from sending an email, posting publicly, charging a card, or leaking a secret. There is no human approval step between draft and action.

Red lines
05

Every person runs the process differently.

The workflow lives in one founder's head. The minute a teammate, contractor, or future-you tries to run it, the result is unrecognisable.

Repeatability

The DWAI rule

One workflow. Seven days. Three manual tests. Zero unsupervised risks.

Every system we design or recommend has to pass this rule before automation gets near it. It's the cheapest filter we know for "is this AI idea ready, or are you about to set $200/mo on fire?"

1workflow chosen before any tools
7days to prove the manual path works
3manual test runs before automation
0unsupervised actions on money, accounts, secrets, or public posting

Choose your starting point · no guessing between packages

Pick the route that matches where you are today.

If you want the system and templates, start with DWAI Solutions. If you want Dylan to inspect the business, scope DWAI Consultancy. Either way, the rule stays the same: workflow first, owner gates before automation.

$49

No-install Blueprint

Choose this if you want to pick one workflow, define red lines, and run a 7-day manual test before installing or configuring tools.

View the $49 Blueprint → Clarity first
$297

Mac AI Operator Setup Kit

Choose this if you use a Mac and are ready to set up Claude Code, Codex, VS Code, GitHub, local operator guidance, Infisical-ready secret custody, NotebookLM, local reports, review queues, workflow guardian handoffs, and optional spec-driven build guidance with written review gates.

View the $297 Setup Kit → Primary offer
Later

Assisted Setup Clinic

Choose this only if you want help implementing after the self-serve kit. It is not open yet; the best next step today is the $297 Setup Kit.

Register interest → Not open yet
Custom

DWAI Consultancy

Choose this if you want Dylan to audit the business/workflow, identify AI-agent opportunities, map red lines, and choose the right implementation path: current-business roadmap, product purchase, or employee rollout.

View consultancy route → Paid consulting

Simple rule: choose $49 for no-install workflow clarity. Choose $297 when you are ready to build the actual Mac workspace. Choose Consultancy when you want Dylan to audit your business before deciding what system to build.

The primary offer

Start with the Mac AI Operator Setup Kit.

v0.2 is a $297 self-serve ZIP for founders who want Claude Code, Codex, VS Code, local operator guidance, GitHub, Infisical-ready secret custody, NotebookLM, local reports, review queues, workflow guardian handoffs, and optional spec-driven build guidance working together to build a local AI operator workspace over multiple sessions.

You buy the setup path, command prompts, templates, workflow packs, local report/review-queue patterns, QA gates, workflow guardian handoff, and NotebookLM Learning Pack that let your local AI tools set up as much as safely possible while red lines stop account, payment, secret, public, and destructive actions until approved.

Manual checkout. Manual ZIP delivery. 7-day satisfaction guarantee. No outcome promises.

Before / after

From scattered prompts to a workflow you can hand to anyone.

Before

  • Scattered prompts in seven chat windows
  • Random tools chosen by hype, not need
  • No owner — "the AI does it"
  • No approval gates on money or posting
  • Every run looks different
  • Nobody knows if last week's output was good

After

  • One documented workflow with a name
  • Defined inputs, outputs, owner, acceptance criteria
  • Reusable prompt pack and source-of-truth folder
  • Human approval gates on every risky action
  • A testing log with three real runs
  • A buyer-owned workflow guardian that keeps future changes visible, reviewable, and gated
  • An automation-ready plan — only after the manual path works

Workflow-first method

Six steps. In this order.

No "Discovery / Design / Deployment / Support" — that's a deck, not a method. This is the actual sequence behind the setup kit and the no-install starter tier.

01

Pick one workflow.

Not three. Not "all of marketing." One real piece of work that runs at least weekly and has a clear human owner today.

Scoping
02

Map the manual path.

Write the workflow as humans run it now, end-to-end. Every input, every decision, every place a person currently uses judgement.

Mapping
03

Define inputs, outputs, owner, acceptance criteria.

What does "good" look like? Who decides? What inputs are always present? What does the finished output have to contain?

Contract
04

Add human safety gates.

Anything touching money, customer accounts, secrets, or public posting requires explicit human approval before it leaves the system. Written down. Non-negotiable.

Red lines
05

Test manually — three times.

Run the workflow by hand or with AI as a co-pilot. Three real runs. Log every issue. If the manual path doesn't survive three runs, no tool will save it.

Proof
06

Create the guarded operating loop.

Once the manual path is stable, owned, and verifiable, package the stable parts into a local operator system with a workflow guardian, review queue, and PASS/HOLD gates.

Automation

The offer ladder

Primary kit first. Assistance later.

DWAI Solutions are the self-serve product path. DWAI Consultancy is the manually scoped service path when you want Dylan to review the business itself.

01

Mac AI Operator Setup Kit v0.2

$297 · self-serve primary kit. Mac setup path, tool routes, workspace structure, optional spec-driven build guidance, NotebookLM Learning Pack, workflow packs, workflow guardian handoffs, and QA gates for a review-first operator workspace.

View the $297 Setup Kit → Primary
02

AI Agent Starter Blueprint v0.1

$49 · no-install starter tier. Choose this only if you want workflow clarity before setup. It does not include the Mac setup path, NotebookLM Learning Pack, or Hermes coordinator layer.

View the $49 Blueprint → No-install
03

Assisted Setup Clinic

Not open yet. For now, buy the $297 Setup Kit first if you are ready to build. Register interest only if you may want help implementing later.

Register interest for later → Later
04

DWAI Consultancy

Custom · manually scoped. Paid business/workflow audits, AI-readiness reviews, agent opportunity maps, implementation roadmaps, and organization rollout planning for founders who want Dylan's judgement on the business.

View DWAI Consultancy → Consulting

Product checkout and consultancy intake are both manual email routes. No external checkout, payment scripts, subscriptions, public-send automation, or hidden upsell are added on this site.

Build the workspace. Or scope the audit.

Use DWAI Solutions when you want the self-serve system. Use DWAI Consultancy when your business needs Dylan to find the workflow, risk, implementation path, or employee rollout plan first.

Solutions are product-led. Consultancy is manually scoped. Both stay workflow-first, security-gated, and guardian-owned only after a workflow proves itself.

A validated workflow can become a workflow guardian only through owner-approved promotion.