🎮 The Next Input — Issue #210

In partnership with

⚡ The Briefing — 60 sec

• Stripe will reportedly acquire AI gateway startup OpenRouter for $7B Isn’t Stripe STILL private?! Yes. Yes it is. Apparently being private is no impediment to casually dropping $7 billion on an AI company. At this point Stripe is less “payments startup” and more “financial infrastructure empire that happens not to have an IPO ticker.”

• Queensland alternative schools aren’t rushing to bring AI into classrooms Fair enough. Not every classroom needs to become a prompt-engineering bootcamp overnight. There’s probably something healthy about asking what AI actually improves before throwing it at every child with a Chromebook.

• Employees turn to AI for workplace advice in disputes If you treat your employees like crap? GPT will lawyer the fuck outta you on their behalf. The information asymmetry is disappearing fast. Policies, PIPs, entitlements, complaint procedures—your employee can now interrogate the whole thing before HR has finished writing the email.

🛠️ The Playbook — Workplace Policy Stress Test

Mission
Use AI to proactively identify weak, contradictory or outdated workplace policies before employees, regulators or lawyers find them first.

Difficulty
Intermediate

Build time
3–5 hours

ROI
Reduces workplace-dispute exposure while making policies clearer, more consistent and easier for managers to apply correctly.

0) Why This Matters

Employees now have an employment-relations researcher sitting in their pocket.

That changes the balance.

A worker can take:

• their employment contract

• a performance plan

• company policies

• meeting notes

• emails

• Fair Work guidance

…and ask an AI system to identify contradictions, missing process or potential entitlements in minutes.

That is not inherently bad.

In fact, it creates a useful incentive for employers:

Make your shit defensible before somebody asks the model to inspect it.

The answer is not trying to stop employees using AI.

It is ensuring your policies, processes and management decisions can withstand the same scrutiny.

1) Architecture

2) Workflow

1. Gather current employment contracts, policies, procedures and standard HR templates.

2. Identify authoritative legal and regulatory sources relevant to each policy.

3. Run structured AI reviews for ambiguity, contradictions and missing procedural steps.

4. Compare findings against historical complaints, disputes and manager questions.

5. Route material policy changes through qualified HR or legal review.

6. Publish approved versions centrally and retire obsolete documents.

3) Example Prompts

Policy Adversarial Review

You are reviewing an Australian workplace policy from the perspective of an employee challenging how it has been applied.

Policy:
[PASTE POLICY]

Relevant employment context:
[PASTE CONTEXT]

Identify:
- ambiguous language
- contradictory clauses
- unclear employee obligations
- unclear manager obligations
- missing procedural steps
- inconsistent definitions
- areas vulnerable to competing interpretations

Do not provide legal conclusions.

Return:
1. issue
2. why it could cause a dispute
3. clarification required
4. recommended policy improvement
5. appropriate human reviewer

PIP Stress Test

Review the following performance improvement process.

Assess whether the documentation clearly establishes:
- expected performance
- evidence supporting concerns
- measurable improvement criteria
- reasonable timeframes
- employee support offered
- review checkpoints
- consequences
- escalation process

Identify wording that is vague, inconsistent or difficult to apply objectively.

Do not determine whether the process is legally compliant.
Flag areas requiring qualified HR or legal review.

Dispute Pattern Analysis

Review the following anonymised workplace disputes and employee complaints.

Identify recurring themes across:
- unclear policies
- inconsistent management behaviour
- performance management
- leave and entitlements
- communication failures
- role expectations
- escalation procedures

Return:
1. recurring root causes
2. policies implicated
3. process changes recommended
4. training opportunities
5. priority for remediation

4) Guardrails

• Never treat model output as substitute legal advice.

• Ground reviews in current authoritative workplace sources.

• Remove employee-identifying information where possible.

• Require qualified review before material employment decisions.

• Keep approved and draft policies clearly separated.

• Version-control every policy change.

• Audit manager templates as well as formal policies.

• Do not use AI analysis as the sole basis for discipline or termination.

5) Pilot Rollout — 3 hours

1. Select three commonly used workplace policies.

2. Gather the current approved versions and relevant authoritative guidance.

3. Run an adversarial AI review against each policy.

4. Identify the five highest-risk ambiguities or procedural gaps.

5. Have HR or legal validate the findings and approve amendments.

6. Publish revised versions and document what changed.

6) Metrics

• Percentage of policies reviewed annually

• Ambiguities detected before disputes

• Policy-related employee questions

• Repeat dispute categories

• Number of obsolete documents removed

• HR escalation frequency

• Manager policy-error rate

• Policy review turnaround time

• Employee complaint resolution time

• Percentage of policy changes with recorded approval

Pro Tip: Assume every policy you write will eventually be pasted into an LLM by somebody actively looking for holes. Write accordingly.

🎯 The Arsenal — Tools & Platforms

• Fair Work Ombudsman · authoritative Australian workplace information and guidance · Link

• SharePoint · centralises approved workplace policies and version history · Link

• Azure AI Search · retrieves policies and approved workplace guidance for grounded analysis · Link

• Microsoft Purview · supports governance, audit and sensitive-data management · Link

• Airtable · tracks recurring disputes, policy gaps and remediation actions · Link

Copy-paste prompt block:

You are designing an AI-assisted workplace policy assurance system for an Australian organisation.

Organisation:
[DESCRIPTION]

Current policies:
[LIST]

Employment agreements:
[LIST]

Recurring employee questions or disputes:
[LIST]

Authoritative sources:
[LIST]

Existing HR and legal review process:
[LIST]

Design a system that:
- centralises approved workplace documents
- identifies contradictory or ambiguous language
- stress-tests policies from both employee and employer perspectives
- retrieves current authoritative guidance
- detects recurring dispute patterns
- routes material findings to qualified human reviewers
- preserves privacy and confidentiality
- maintains policy version history
- never substitutes AI output for professional legal advice

Return:
1. architecture
2. policy-review workflow
3. risk taxonomy
4. review prompts
5. human approval matrix
6. dispute feedback loop
7. rollout plan
8. operational metrics

💡 Free Office Hours

AI is changing workplace relations from both sides. Employers can use it to tighten policies and processes; employees can use the same capability to challenge them. The sensible move is making sure your organisation survives scrutiny rather than hoping nobody asks the right question.

Book here: https://calendly.com

AI Is Moving Fast. Here's How to Keep Up.

AI is moving faster than any other technology. New models. New tools. New claims. New noise.

Most people feel like they're behind. But the people that don't, aren't smarter. They're just better informed.

The Future Today is a daily briefing for people who want clarity. In one concise email each day, you'll get the most important AI and tech developments, learn why they matter, and what they signal about what's coming next.

Written for operators, builders, leaders, and anyone who wants to sound sharp when AI comes up in the meeting.

One email. Five Minutes. Stay ahead of 99% of the world.

Join Free Today →

🕹️ Game Over

Stripe is privately spending public-company money.

Some schools are pumping the brakes.

And your employee just turned ChatGPT into an HR cross-examiner.

Interesting Monday.

— Aaron Automating the boring. Amplifying the brilliant.

Subscribe: link