Insights

Real systems we designed, built and run ourselves. The same engineering goes into yours.

Two companies run by AI employees

situation
One founder, two businesses to run: an ecommerce store and this agency, with no staff for either.
task
Operate both companies daily without hiring, from market research to client replies.
action
Fifteen AI employees across research, design, publishing, sales, mail and content, each behind a human approval gate.

The AI receptionist

situation
A cleaning company where enquiries sat unanswered for hours while the owner was out on jobs.
task
Answer every enquiry instantly, qualify the work and book it.
action
A chat receptionist on the website that quotes, qualifies and books, with every lead pushed to the owner.

The self-cleaning inbox

situation
Years of promos and newsletters burying the mail that actually mattered.
task
Clean it once, keep it clean forever, with zero daily effort.
action
Every incoming email labeled on arrival, stale promotions swept on a schedule, one morning digest.

The job scout that reads like a human

situation
Finding good roles meant scrolling four job boards every day.
task
Surface only the postings worth applying to.
action
A daily agent scans four job markets and scores every posting against the candidate's profile.

Brynth OS: an AI fitness coach

situation
Fitness apps track numbers. Nobody coaches: no context, no memory, no judgment about what is safe to say.
task
Build a mobile coach that knows your training, nutrition and health data, and never invents numbers.
action
A React Native app on a NestJS platform: HealthKit integration, deterministic nutrition targets, safety guardrails that refuse instead of guessing.

Showrunner: festival operations SaaS

situation
Festival teams coordinating artists, stages and schedules across spreadsheets and group chats.
task
One platform for the whole production office.
action
A multi-tenant SaaS with scheduling, team onboarding and Stripe billing, shipped through six product phases.

Ranwae: finance for physical brands

situation
Physical product brands flying blind between production costs, stock and actual margin.
task
Make unit economics visible and decisions testable.
action
A financial platform with a purpose-built calculation engine, backtested against real brand data.