01 / About
A technical diagnostic, before the sales narrative
Can I Hire an AI? turns a role description into a structured report in under five minutes — automation score, task-level breakdown, 36-month ROI projection, and regulatory readiness. No consultant. No €5K discovery phase.
02 / The problem
Three questions that deserve real answers
AI automation is happening now, in companies your size, in roles that look a lot like yours. The barrier isn't capability — it's clarity.
“Which role do I start with?”
“What would it actually cost to automate this?”
“Is it worth it? When do I break even?”
03 / Scope
Analyze a role, a department, or a process
You might be thinking about a single role, a whole department, or a specific process. The tool works at all three levels — you pick the scope that matches the question you're trying to answer.
A specific role
"Sales Development Representative", "Finance Manager", "Customer Support Agent". The analysis breaks the role into its component tasks and assesses each one independently.
A full department
"Customer Success team of 8", "Finance & Accounting department". Useful when you want to understand the automation landscape across a function before deciding where to focus.
A task or process
"Monthly invoice reconciliation and supplier payment cycle". Useful when you already know the process you want to target and just want a rigorous assessment of what's automatable. This scope produces hours-recovered estimates — not monetary ROI — since the work spans roles and can't be tied to a single salary baseline.
Step 1 of 5
What do you want to analyze?
Step 1 of 5 — choose the level of analysis. The rest of the form adapts accordingly.
04 / The form
Designed for humans, not for technical people
We invested as much engineering effort in the form as in the analysis engine behind it. Designed for the person who knows the role — not the person who knows AI.
- No AI jargon, no dropdowns full of terminology
- One plain-language question at a time
- Most users finish in under a minute
- Describe the work — the system handles the rest
05 / Ethics
Not a tool to decide who to fire
That framing gets the whole thing backwards. The question isn't "can a machine replace this person?" The question is: how much of this person's time is spent on repetitive, low-judgment tasks — so they can spend more time on the work that actually requires a human?
Not this
This
Evaluating whether a person's job should be eliminated
Identifying how much of a role is repetitive, low-judgment work that automation can absorb
A headcount reduction tool
A capacity-recovery tool — freeing human hours for higher-value work
A verdict on whether a person is replaceable by AI
A task-level analysis of which hours follow predictable patterns vs. require human judgment
A justification for layoffs
A starting point for a conversation about where your team's time is actually going
Behind this tool
kaixlab.com
Built by a lab in AI engineering and cybersecurity.
Can I Hire a Bot? is an open experiment by KAIX LAB. The same team that designs production AI architectures, audits models against OWASP GenAI, and helps organizations meet the EU AI Act stands behind this diagnostic.
If what you see here is useful, the rest of the lab — training, audits, and implementation projects — lives at kaixlab.com.
Visit KAIX LAB06 / Method
A structured analysis method
The role description is normalized into a structured profile, specialist nodes run in parallel across four independent dimensions, and only then is the result assembled. Financials stay deterministic and inspectable — not hidden inside generated prose.
Context-aware profile
Role, country, sector, salary, and workload are treated as one working brief that informs every downstream step.
Parallel specialist nodes
Task decomposition, regulatory screening, solution design, and rollout planning run as separate reasoning steps before synthesis.
Deterministic financials
Savings, ROI, and payback use a fixed model. Inputs and assumptions remain inspectable in the result view.
07 / Governance
Security and regulation by design
Defense-in-depth
Free-text input from anonymous users is an attack surface. Three independent layers run before any LLM sees your data.
01
Deterministic screening
Regex guardrail scans for PII signatures (emails, phones, URLs) and prompt injection patterns. Pure code — no model, no inference cost.
02
Sanitization
Unicode normalization, control character stripping, length limits. The model sees a clean, bounded string — not raw user input.
03
OpenAI Content Safety
OpenAI runs independent PII and jailbreak detection at the infrastructure level. A separate backstop if anything slips through.
The same controls apply to every AI solution we build for clients. Data handling, prompt safety, and infrastructure-level guardrails are engineered in from day one — not retrofitted.
Regulatory screening runs in parallel, not as an afterthought
The most avoidable failure mode: discovering regulatory constraints after the system is built. Most agencies skip this. We screen against the full context: country, sector, tasks, and organization type.
GDPR
Data protection obligations
Personal data in automated pipelines triggers obligations around lawful basis, data minimisation, and human oversight. Flagged where relevant.
EU AI Act
High-risk classification + documentation
Employment decisions and safety-critical systems trigger compliance requirements. The screening checks classification and documentation obligations.
National rules
Sector and jurisdiction-specific constraints
German co-determination, French union consultation, healthcare data rules — jurisdiction-specific constraints checked against your declared country and sector.
08 / Outputs
Five outputs, one coherent readout
The result helps an operations or leadership team decide whether a workflow deserves deeper investigation — not to close an implementation deal on the spot.
How the economic model works
The model runs month-by-month projections over 36 months. Every month: (annual productivity value / 12 × adoption rate) − monthly recurring cost. Year-1 ROI accumulates months 1–12. Steady-state ROI uses months 25–36. Payback is the first month where cumulative position turns positive.
Adoption rate ramp
Cumulative value · 36 months
Illustrative · typical SDR profile
Realization rate
100%
Single role
Full projected productivity value is assumed realizable.
70%
Department / team
30% absorbed by reclassification overhead, QA processes, and change management — consistent with real deployment data.
Task or process
Hours only
When you select a task or process as the scope, the analysis shifts to a productivity diagnosis: instead of economic ROI, the output measures hours recovered per week. The work spans multiple roles and cannot be tied to a single salary baseline, so no payback period or NPV is calculated.
Demo
See the kind of report you get back
The walkthrough shows the actual output format: score, task evidence, financial assumptions, risks, and a practical implementation path. For more inspiration, browse real example roles on the homepage.
AI-Readiness Score
An overall percentage of the work that AI could handle, derived from task-level scores weighted by time. Not a guess — each task is scored independently.
Task breakdown
Every task scored individually. You see which parts of the work AI can handle and which remain distinctly human — and why. The question stops being "can AI do this job?" and starts being "which hours of the day?"
Economics
Hours recovered, capacity value, estimated savings, implementation cost, ROI, and payback period. Built on a non-linear adoption curve over 36 months. Inputs and assumptions are visible in the result.
Implementation plan
A phased timeline with concrete steps: discovery and design, supervised pilot, full deployment. It tells you what to do in what order.
Regulatory readiness
GDPR obligations, EU AI Act requirements, and sector- or country-specific rules that apply to the proposed automation.
09 / References
Based on traceable sources, with explicit assumptions
Every model assumption, adoption curve, and regulatory flag traces back to something specific — consultancy studies, peer-reviewed papers, real deployment data, regulation, and honest reporting on where AI still falls short. These are the sources that inform the tool.
Run a free analysis on any role or process
Describe a role, a department, or a set of tasks. Get your automation score, ROI projection, and payback period in under 5 minutes. No signup required.