See exactly how AI should run in your business.
Most AI plans fail because of how work moves through the business, not because of the technology. Answer eight short questions and get a plan sized for a business like yours and where it is on its journey: which repeating tasks to hand to AI, which tools to use, where a person should still make the call, and what order to do it in.
Works right in your browser · Nothing you type is sent anywhere · Same answers always give the same plan
- 1 Answer eight questions Your business, where you are now, your team and budget. About three minutes.
- 2 Read your plan The loops, the tools, the people and the checks, drawn out.
- 3 Take it with you Copy it into a document, or refine it with your own AI.
CommonAdd AI to one step
From 15 days to 14 days
The work itself gets six times faster, then waits in a queue for the same people to check and sign it off.
AI-nativeRedesign how the work moves
From 15 days to 5 days
AI drafts and checks at every step. A person decides only where judgement matters, so nothing waits.
Your answers
Tier 2 of 4 Guardrailed build
- AI steps
- 0
- People
- 0
- Approvals
- 0
- Tools
- 0
- Timeline
- 0wk
Plain meanings for words you'll see across this plan. No need to memorize them — come back here any time.
- Level (or "Tier")
- How much AI is trusted to work on its own, from 1 (a person checks every step) to 4 (AI runs approved routines with no one watching in real time).
- Approval point (or "gate")
- A checkpoint where work must be signed off before it moves on — by a fixed rule, a second AI check, or a person.
- One person in charge (or "DRI")
- One named owner is accountable for a result, instead of a group deciding by committee.
- Second AI check (or "adversarial pass")
- A second, independent AI reviews the first AI's work before it reaches a customer.
- Playbook (or "canon")
- The small set of official documents — answers, rules, procedures — that AI reads before doing any task.
- Shared memory (or "context layer")
- One searchable place holding your records, documents and decisions, which every tool and AI step can read from.
- Accuracy test (or "eval")
- An automatic check that keeps confirming the AI is still getting things right, the way a test drive checks a car.
- Summarizing agent (or "distillation")
- An AI that turns a large pile of calls, tickets and notes into a short, useful summary.
- Build-and-test pipeline (or "CI")
- The automatic process that checks and assembles software every time it changes, before it goes live.
- Direct tool connection (or "connector"/"MCP")
- A link that lets an AI tool read from and update another app directly, such as your CRM or support desk.
- Company-wide safety rules (or "managed settings")
- Guardrails set once, centrally, that no individual AI session can turn off or bypass.
- Spending limit (or "spend envelope")
- A cap agreed in advance so AI costs never creep up unnoticed.
Company
- Done by AI
- Done by a person
- Approval or policy
- A system or signal
- A document or tool
Point at any box to see what feeds it and what it feeds. Click to read what it is, who owns it, and why it is there. On a small screen, swipe the diagram sideways to see the rest.
Shared context, read by every lane
Who does what
The same plan as a working team. The circles along the top are the agents doing the repeated work. The circles below are the people who still decide. Everything they send each other passes through one shared memory in the middle, and the label on each line names what travels it.
Click any circle to read what it does and what it reads and writes. Drag to move the picture, and hold Ctrl (or Cmd) while scrolling to zoom.
Build it in this order
Dependency order, not lane order, and phases overlap. Each ends in something you can point at. Click any part to open it in its view.
How you know it worked
Compare leading operational signals (weeks) against lagging business results (quarter). Every metric is calibrated to your generated plan with concrete targets, calculation formulas, and systems of record.
Try it — change the three numbers and watch the result
Plan Outcome & Value Simulator
Simulate capacity returned and business impact based on your team size and task overhead.
| Stage & Focus | Leading Indicator (Early Operational Signal) | Lagging Indicator (Quarterly Business Outcome) | Source of Truth |
|---|
Failure modes & operational risks
Contextual risk exposure derived from your team size, autonomy tier, industry constraints, and primary goal.
Take it with you
Plain text with simple headings, ready to paste into a document, an email, or a chat with your own AI assistant.
About the sample prompts. Each prompt here is a starting point, not a finished script: it shows the level of detail and the constraints worth mentioning. Rewrite it in your own words, for your own tools, before you rely on it. A prompt you haven't actually tried yet is only a guess.
A plan is not a result
Most of these plans die as a document somebody meant to come back to. Activating yours turns it into a roster and a checklist: which parts run without you, where a person still decides, and what to do in the first week.
On the tools. Named products are common choices at each stage, not endorsements, and the landscape moves faster than this page. The category, what it emits, and how that data reaches your context layer are the durable parts. Swap the product.
Plan activated
Your business with the plan switched on
The same blueprint, read as work rather than as a diagram: the parts that run without you, the places a person still decides, and what to do in the first week. Nothing here is new advice — it is your plan, sorted by who has to do something about it.
Click any part to read it right here, see what comes before and after it, and step through the rest with the arrow keys. You stay on this page.
Run by AI
Run by your people
Before you start
Four decisions a person has to make. None of them need a tool, and skipping them is what turns the rest into shelfware.
Switch-on checklist
The front of your rollout, small enough to finish. The Rollout section holds the rest of the sequence. Ticking boxes here is just for this sitting — nothing is saved.
Questions people ask before they book
- What is AI-native business deployment?
- It means running your business so that AI does the repeated work and your people make the decisions, instead of bolting an AI tool onto the way you already work. Most AI projects fail because a company buys software and changes nothing about how work flows. Deployment is the part that changes the flow: what gets written down, who decides what, what has to be checked before anything reaches a customer.
- Who runs the plan with us?
- If software is not what you sell, a fractional Chief AI Officer: a senior lead who maps how work flows through the business, puts AI on the repeated parts with guardrails and spend caps, and sets up the loops that improve it. If you build software, a Forward Deployed CTO: a senior engineer who works inside your team rather than from the sidelines. They set the plan, build the first version with you, and stay until it is running. For a startup or a small business that cannot hire a full-time CTO, it is the cheapest way to get senior technical judgement on the decisions that are expensive to get wrong.
- Is the planner free, and is my information stored?
- It is free and nothing is stored. The whole planner runs inside your browser, so your answers are never sent to us or anyone else. The same answers always produce the same plan.
- Do I need a technical team to use this?
- No. The plan changes depending on whether you have developers or not. If you have nobody in house, it recommends a different shape of work, starting with the day-to-day processes that never needed software in the first place.
Make your business AI-native, starting this month.
You speak to the senior lead who would do the work. If we are not the right fit, we will say so.