What becomes possible when your role changes?
As routine work becomes automated, space opens for judgement, creativity and connection. We help people and organisations explore that space and build what comes next.
Explore the possibilities Scroll to walk the lab ↓Your operation. A coordinated
AI counterpart.
We built an agent counterpart to our own lab first, with defined roles, workflows and human approval points. We call it our digital twin, and we build one for your business.
Defined roles. Coordinated work. Human approval.
See how it worksExplore our products and partnerships. Systems for your operation.
Our products & partnerships- 1Yield · plant A$3.4M
- 2Labour hours per unit$2.7M
- 3Freight cost per tonne$2.2M
- 4Energy per unit$1.3M
- 5Scrap rate$1.0M
W-0217 Standby → Packing
Fills the gap on line 2W-0142 Packing → Processing
Certified for the open positionW-0175 Prep → Processing
Frees a senior positionW-0088 Processing
Locked by supervisor
Targets, actuals and ranked gaps for every unit, on one page.
- Gaps ranked by financial impact
- Monthly approval built in
Every position covered and every rule respected, in seconds.
- Skills and certifications checked
- The supervisor approves the plan
Every charge checked against the contract before you pay.
- Mismatches flagged with the reason
- The same engine for accounts and receivables
A machine-learning concept: every cut measured against its spec.
- 3D measurement from a camera
- The operator sees exactly where to trim
A rental contract, read on your side.
Five specialist agents check it against the law and score every part in minutes.
Visit contratoclaro.com.br ↗Shown in English for illustrationMore clients for wellness businesses.
Strategy, content and ads: the output of a marketing team, with a person checking every piece.
Visit lumenagency.com.au ↗AI for retail: buying, stock and pricing.
Their product, built with a partner team we work alongside.
Visit chicoai.com.br ↗Shown in English for illustrationFollow the work. Learn what's next.
Want to follow the work, learn and stay current on AI in real operations? The community opens in the next few weeks. Join the waitlist and you will be the first to know.
Taught by Rodrigo Souza · 11 modules
See the Academy
Four questions, answered in order.
Solutions
Enterprise systems, our own products and partnerships. Each one is real and we can show it.
Problem: agents do work nobody can see or check. Result: lines with stations, independent review and a person on every final call.
See how it works →
Problem: wellness businesses have no time for marketing. Result: the output of a marketing team, month after month.
Visit lumenagency.com.au ↗
Problem: people sign rental contracts they can't read. Result: a scored, plain-language reading in minutes.
Shown in English for illustrationSee Contrato Claro →
Problem: home maintenance on the Gold Coast means chasing quotes. Result: one fixed labour price shown before you book, and a verified tradie.
About GetSOLVD →
Problem: producing content at scale without losing accuracy. Result: a channel produced by a pipeline, approved by a person.
See clareIA ↗More products & ideas
How we work
We build a coordinated AI counterpart of your operation. People keep every final call.
Agent orchestration: your operation, mirrored by a coordinated team of AI agents.
We built an agent counterpart to our own lab first, with defined roles, workflows and human approval points. Every line is explicit: one agent per role, its skills, its instructions, and the approvals it cannot skip.
The agent does
- drafts, builds and checks
- works one station, one instruction
- is reviewed by an agent that didn't build it
A person decides
- the goal and the criteria
- whether the output is approved
- what ships
Our agent counterpart, in numbers
Learn & community
Start by following the work. Go deeper with the Academy.
Built in public, alongside the lab.
This site shows what the lab builds. The community is where Rodrigo talks about how: the decisions behind the work, the misses and the methods, while they're still fresh.
More about the community →Human-Agent Orchestrator Academy
Learn to direct work done by people, AI agents and software together, in 11 connected modules with Paperclip as the lab.
Follow the journey from left to right. B02 sits between L02 and L03: it turns repeated work into something you can reuse.
Rodrigo Souza
- Executive, management consultant and founder of Kaikaku Labs.
- Ten years in management consulting, including as a partner.
- International executive career leading teams in Brazil, the US, Europe and Oceania.
- Food Engineering; MBAs in Data Science, AI & Analytics, in Finance, Auditing & Controlling, and in Valuation.
About & contact
The people behind the lab, and one conversation to see if we fit.
Kaikaku means radical change. We practise it on ourselves.
Founded by Rodrigo Souza, Kaikaku Labs brings together the products we create, the solutions we deliver and the partnerships we build.
Rodrigo is an engineer who grew up inside real companies. He started on the quality and audit side (ISO 9001, ISO 14000, GMP, HACCP), then became a management consultant and later an executive leading operations, commercial strategy and finance across countries and cultures.
The lab was born when he hit the limit of continuous improvement alone. A whole generation built ISO systems, PDCA and better processes, yet many companies still live in noise. Coordinated agents, with people in charge, are the next step.
We exist to give people and organisations a place to move through this transition together and grow exponentially (tenfold), building the structures, skills and communities that let a small team deliver outsized results.
- Founded by Rodrigo Souza
- Remote-first, clients worldwide
- Kaikaku Labs (ABN 63 852 840 846)
Fair questions, straight answers
What do humans actually supervise?
Direction and judgment. Agents draft, build, check and report; people set the goal, review the output and sign off before anything reaches a client.
What happens when an agent gets something wrong?
The system assumes it will happen. Nothing ships on an agent's word alone.
Will agents replace my team?
The design assumes your experts stay. Their judgment is what the approval points are for.
What happens to our data?
Agents only see the sources the engagement names, and nothing you share is used to train models.
How long before we see something working?
We start small on purpose: one process, validated against real cases.
Put agents to work on a real problem, with your people in charge.
Bring us one process you want off your team's plate. We'll tell you if coordinated agents fit it, and what it takes to run them at a standard you can sign off on.