{title}
{children}
What is the AI Readiness Compass?
A free, 6-minute self-assessment that tells you whether your company's systems and data can actually support AI that does real work — and exactly what to fix first.
The problem it solves
Most AI projects don't fail because the model isn't clever enough. They fail because the systems underneath can't give the AI anything trustworthy to work with.
Useful AI isn't a chatbot writing marketing copy. It's an assistant that can answer "where is order 4471?", update a customer record, flag an invoice, or brief your team on an account. Every one of those needs the AI to read the truth out of your systems and, sometimes, act on it safely.
What it measures
Six weighted dimensions, 100 points in total.
{d.desc}
There is also a small bonus score for MCP readiness — whether your systems could safely be opened to an AI assistant through one secure, controlled doorway.
Your result
Your total puts you in one of five stages, each with a different first move.
What you'll get
The two layers it talks about
Almost every recommendation points at one of two things. They are different jobs, and the assessment works out which one you actually need.
Integration & Orchestration Layer
The nervous system. Sits at the centre of your systems and connects each one once, in a reusable, governed way — so a change means updating one place, not ten. Recommended when you run several core systems (ERP, CRM, warehouse) with nothing joining them up.
AI Gateway
The control point every AI request passes through: who may call what, spend limits, policy, data masking and a full audit trail. Recommended when your landscape is small, or added on top when you already run an integration platform.
If you already run something like n8n, Zapier or MuleSoft, the report will not sell you a second platform — it tells you to keep what you have and add governance on top.
Who it's for
Pick your role at the start and the questions, the wording and the recommendations all change:
-
{[['Business owners and leadership','Strategic language, cost and ROI, a roadmap you can act on.'],
['Sales, marketing and operations','Your day-to-day tools and manual handoffs, in plain language, with no technical knowledge assumed.'],
['IT managers and architects','Full technical depth on APIs, integration and data governance.']].map(([t,d])=>(
- {t} — {d} ))}
Built for small and mid-sized companies. Nothing to prepare, and you don't need to know what's under the bonnet — "I'm not sure" is a valid answer throughout, and every question starts at "none".
The whole assessment runs in your browser. No sign-up, no form, no database, no cookies, no tracking. Your answers vanish when you close the tab unless you export the PDF yourself.
No product is named or sold. Recommendations point to the kind of capability you need, never a specific vendor — the point is an honest picture, not a sales pitch.
Ready to find out where you stand?
Six minutes, no signup, and an honest answer at the end.
AI Readiness
Compass
Find out whether your business is really ready for AI — and what to fix first.
Why this exists
Every business is being told to "do something with AI". The hard part is knowing whether your foundations can carry it. AI that actually does work — checking an order, updating a record, answering a customer properly — has to read the truth out of your systems and act on it safely.
If your CRM, your ERP and your spreadsheets only talk to each other through people, no amount of AI will fix that. This assessment scores that foundation, shows you where the gaps are, and tells you what to do first — without selling you any particular software.
How it works
-
{HOW_IT_WORKS.map(([ic, title, desc], i) => (
-
STEP {i + 1}
{title}
{desc}
))}
Ready to find out where you stand?
Six minutes now can save months of building AI on foundations that cannot carry it.
Questions, or want to talk it through?{' '} {CONFIG.contactEmail}
Tell us about your company
This tailors the questions and your result. Nothing you enter is sent anywhere.
What best describes your role?
We adapt the questions, language and recommendations to match.
Your AI vision & ambition
{a.persona==='exec'?'Where do you want AI to move the business?':a.persona==='ops'?'What would make your daily work easier?':'What outcomes are you targeting?'}
Listed from the most focused scope to the broadest.
{cat.hint}
} {cat.tools.length>0 && (Which systems do you use?
Expand a category, pick your tools (or add your own), and rate each one. Everything defaults to None.
How are your systems connected?
The heart of agent readiness: can data move safely and automatically?
{/* Asked of everyone — a middleware platform is the strongest single signal of readiness, and it was previously invisible for two of the three roles. */}Platforms that move data between your systems automatically — for example Zapier, Make, n8n, Power Automate or a larger integration platform.
Select every one you use.
Tell us about your data
Agents are only as trustworthy as the data underneath them.
Automation & AI maturity
Where are you on the journey today?
{simple ? 'Most companies have none of these yet — that is the normal starting point. Tick anything you do have.' : 'Guardrails that keep AI safe, compliant and affordable.'}
Your AI Readiness Score
{TIER_HEADLINE[res.tier]}
{res.hasMiddleware ? <>You already route through {res.middlewareName}, so this is not a pure point-to-point landscape — a real head start. What it does not yet do is govern AI access. : <>Brittle point-to-point connections. Costly to maintain, hard to secure — AI can't run reliably on this. If one system changes, the connections around it break.} {!res.hasLlm && <> You have no AI running against these systems yet, so this is simply the ground AI would have to stand on.} {apiNodes>0 && <> {apiNodes} of your systems already expose APIs — that groundwork counts.}
{res.architecture === 'platform-gateway' ? <>Keep {res.middlewareName} as your Integration & Orchestration Layer at the centre, and add a separate AI Gateway above it — authentication, rate limiting, policy and logging for AI specifically. : res.architecture === 'gateway' ? <>With {res.totalSystems} system{res.totalSystems===1?'':'s'} and no ERP or data warehouse behind them, you do not need a full integration platform. An AI Gateway sitting directly under your AI, connecting out to each system, is enough. : <>An Integration & Orchestration Layer at the centre of your systems — one governed, reusable connection each — before AI governance is added on top.}
{res.architecture === 'middleware' ? 'The nervous system first — then agents can be trusted to act through it.' : 'Governs how agents access enterprise systems and data.'}
{t}
{d}
{res.architecture === 'platform-gateway' ? <>You already run {res.middlewareName}. Treat it as your Integration & Orchestration Layer — the nervous system at the centre of your systems — and add a separate AI Gateway above it. They are two different jobs: one moves and governs data between systems, the other governs what AI is allowed to do with it. : res.architecture === 'gateway' ? <>Your landscape is light enough that connecting it is not the bottleneck — control is. An AI Gateway directly beneath your AI, calling each system, gives you authentication, spend limits, policy and logging without a full integration programme. : <>You run {res.bigSystems.length>=2 ? <>{res.bigSystems.length} core systems : <>{res.totalSystems} systems} with nothing joining them up. The Integration & Orchestration Layer comes first: one governed connection per system, sitting at the centre rather than on top, exposing each as a clean reusable API. AI governance is added onto that afterwards — not instead of it.}
-
{roi.without.map(x=>
- ✕{x} )}
-
{roi.with.map(x=>
- ✓{x} )}
{byRole(a.persona, { it: <>Your systems already expose programmatic interfaces, so the integration half of the problem is largely solved. Building a full iPaaS on top would duplicate work you have already done., exec:<>Your systems can already be reached automatically, so you probably don't need a full integration platform to connect them. The expensive groundwork is largely done., ops: <>Your systems can already talk to other software automatically — the hard plumbing is mostly there., })}
What you're missing is the layer above them: an AI Orchestration Layer — a single control plane that governs traffic, decides which agent may call which API, enforces security and limits, and logs everything.
This is considerably less work than a full middleware programme — you are building governance on top of what already exists, not rebuilding the plumbing.
Some of your systems are reachable through APIs, others are not. The efficient path is to API-enable the remaining core systems first, then put a single Integration & Orchestration Layer across all of them — rather than building one-off connections for each.
Your systems are not yet reachable programmatically, so AI has no safe way in. Start by making your two or three most important systems accessible, then govern them through one Integration & Orchestration Layer instead of wiring point-to-point.
{res.hasMiddleware ? byRole(a.persona, { it: 'You have a platform to build on, but the AI-specific controls are not in place. Each gap is an unbounded call path, an unlogged action, or an unmetered cost.', exec:'You have a platform to build on, but several controls are still missing. Each one is a place where AI can cost more than expected, or reach data it should not.', ops: 'You already have a tool that connects your systems — but nothing yet that limits what AI can see, do or spend.', }) : byRole(a.persona, { it: 'Without a gateway, every AI tool holds its own credentials and calls your systems directly. There is no central place to enforce scope, apply rate limits, or reconstruct what happened.', exec:'Without a governing layer, every AI tool connects directly to your systems with its own credentials and its own rules. There is no single place to see what it accessed, cap what it costs, or prove compliance.', ops: 'At the moment each AI tool connects on its own terms. Nobody can see what it looked at, stop it going too far, or say what it cost.', })}
Missing today:
-
{res.securityGaps.map(g => (
- ✕{g} ))}
Why this matters now
- •Cost control. Without token and consumption limits, AI usage scales with no ceiling — the bill arrives after the fact.
- •GDPR & the EU AI Act. You need to show which data AI processed and on what basis. Masking personal data before it reaches a model, and a data processing agreement with your provider, are the usual expectations.
- •Access boundaries. A documented decision on what AI may and may not read — salary data, customer contracts, health records — enforced in one place rather than trusted to each tool.
- •API policy. Rate limiting and OAuth scopes stop one misbehaving agent from overwhelming a production system or reaching beyond its remit.
- •Audit trail. If you cannot show what the AI did, you cannot defend it — to a customer, an auditor, or a regulator.
These are exactly the controls an Integration & Orchestration Layer centralises. Without one, each has to be solved separately in every tool — or not at all.
{byRole(a.persona, { it: <>An AI Gateway is the enforcement point every agent call passes through. It can be a standalone layer or a capability of your Integration & Orchestration Layer — either way, it is where identity, limits, policy and logging are applied once instead of per integration., exec:<>An AI Gateway is the single point every AI request passes through. It can sit inside your integration platform or stand alone. Its value is that control, cost and evidence live in one place — which is what makes AI safe to scale beyond a pilot., ops: <>An AI Gateway is a checkpoint that every AI request goes through. It is what turns "we are trying an AI tool" into something you can rely on day to day without worrying about cost, mistakes or data leaking., })}
{byRole(a.persona, g.body)}
Three ways it pays for itself
- 1.Lower running cost. Caching repeated prompts and routing simple queries to cheaper models cuts blended cost per request — automatic once the gateway exists, instead of a decision made project by project.
- 2.Incidents avoided. Guardrails and personal-data controls at the boundary reduce the chance of a costly leak. Priced against even one avoided incident, it usually dwarfs what the layer costs to run.
- 3.Faster delivery. The next team building the next agent inherits auth, limits, logging and routing instead of solving them again — real engineering weeks saved per use case, compounding as you build more.
Why now: autonomous agents break the assumption traditional API infrastructure was built on — predictable, human-initiated requests. That exposes gaps in token visibility, policy coverage and audit trails. Adding a standalone gateway per tool deepens the fragmentation; one shared policy layer across whatever gateways you already run is what closes it, since governance gaps form at the seams between tools.
{res.hasMiddleware ? <>You already have {res.middlewareName} to build on — the gateway is the layer you add on top of it, not a replacement. There is no case for a second integration platform. : <>None of this exists by default. Every AI tool you add without it is another set of credentials, another uncapped bill and another blind spot.}
{INDUSTRY_RULES[a.i_industry].intro}
{t}
{d}
Practical consequence: in your sector, the governing layer is not an optimisation — it is where masking, access rules, logging and provider contracts are actually enforced. Regulators ask you to evidence these, and evidence is much easier to produce from one place than from a dozen tools.
General orientation, not legal advice — confirm the specifics with your data-protection officer or counsel.
{r.body}
Think of it as a secure power socket for AI. Instead of a custom wire to every AI tool, it gives any AI assistant one safe, controlled doorway into your business systems — and you decide exactly what it may see and do.
It's the difference between AI that talks about your business and AI that actually works in it — looking up a real order, updating a record, answering from today's numbers instead of guessing. Without it, every AI project is rebuilt from scratch.
{res.mcpReady ? 'Your systems are structurally ready: you can open this doorway and let AI work with your live data under your rules.' : 'To get there: make your key systems reachable automatically, put one secured gateway in front of them, and write down which data AI is allowed to use.'}
The orchestration layer is the brain
It plans, reasons and decides what to do next. That is what an AI Gateway and an agent framework give you.
The integration layer is the nervous system
It executes reliably and securely against real systems, and exposes them as clean, reusable, governed APIs — which then simply look like "tools" to the brain.
{byRole(a.persona, { it: <>Connecting an LLM straight to raw systems works — until you hit scale, security review, or the first breaking schema change. Most teams that start direct end up building a middleware layer anyway within 12–18 months, usually without calling it that, and usually under time pressure. Orchestration does not stop working without it; it works better and more safely when it is calling governed APIs instead of raw systems., exec:<>Going direct is a legitimate way to start, and for your landscape it is the right call today. The honest caveat: almost every company that starts this way rebuilds an integration layer 12–18 months later — usually when growth, an audit, or a system change forces it. That is not a reason to build it now. It is a reason to keep the option open and avoid decisions that make it harder later., ops: <>Starting simple is the right call for you. Be aware, though: as you add systems and more people rely on it, most companies reach a point where they need a proper connection layer underneath. That is normal, and nothing you do now has to make it harder later., })}
Practical guidance: build the gateway now, and keep your integrations behind clean interfaces so they can move under a proper integration layer later without rewriting your agents.
Ready to take the next step?
Book a free 30-minute AI Architecture Review to turn this assessment into a concrete, vendor-neutral roadmap.
When to use what: direct point-to-point is fine for 1–2 simple systems. An integration layer pays off at 3+ systems, complex processes, or when you need security, monitoring, reusability — and AI on top.
Part {n}
{title}
{sub}
{icon &&
{subtitle && {subtitle}
} {!subtitle && } {children}{missingHint}
}{CONFIG.brandName} · vendor-neutral AI readiness assessment