BerlengaTur
A growth system for maritime tourism, combining acquisition, technical health, multilingual content and digital experience.
- +20% sales YoY
- 97/100 technical health
- 6 languages and around 70 articles
- Live case
Work
Selected work, with context and explicit status.
Cases and experiments where strategy, systems and execution meet. Only approved public work enters this collection.
Berlenguinha — archived
An experimental travel assistant disabled after client validation revealed critical errors. Approximate 90% average accuracy was not sufficient for critical intents.
- Archived experiment
- No public demo
- Learning preserved
- Not presented as a success
Marcoo
A scheduling platform for salons, barbershops and clinics in Portugal. Cofounder, with Marketing and GTM responsibility.
- Cofounder · Marketing & GTM
- B2B scheduling SaaS
- Portugal — salons, barbershops, clinics
- Closed beta
Agent OS — internal system
An internal operating system connecting prospecting, commercial pipeline, project execution and invoicing in a single cadence, with specialised agents per function and a human approval gate before any external action. Not a client case: proof that the same systems thinking applies to the business proposing it.
- Prospecting → pipeline → execution → invoicing, in a single cadence
- 6 modules, with specialised agents per function
- Mandatory human approval before any external action
- Internal system · in production · no public demo
How to read the work
A useful case explains decisions, not only a result.
This collection shows how I think under real constraints: the problem, which part of the system needed to change, what I measured and what I learned.
Problem before solution
A campaign, website or assistant only makes sense when the problem is clear. Context determines architecture, priority and sometimes the decision not to build.
Explicit responsibility
I distinguish strategy, execution, collaboration and external dependencies. Outcomes rarely belong to one intervention, so role and scope need to be clear.
Evidence with limits
Metrics need a source, period and context. When that information is not documented, the claim remains marked as pending validation.
Learning also counts
Archiving an insufficiently reliable experiment can be the right decision. Berlenguinha shows why average accuracy cannot replace reliability in critical intents.
Publication criteria
What enters — and what stays out.
Only public work with approved context enters this collection. Private tools, commercial data, credentials and interfaces containing real information are not published to make the portfolio look larger.
- Explicit project status
- Results are not transformed or rounded
- Limits and dependencies described
- Private data excluded or anonymised
- PT/EN parity
Questions about cases and experiments
Why are there only a few public cases?
Because quantity cannot replace permission, context and evidence. Confidential work or work without approved content remains outside the public collection.
Is a prototype presented as a product?
No. Prototypes, experiments, internal tools and archived projects keep that status. They are not presented as SaaS or validated businesses without evidence.
How are results without a complete source treated?
The original number is preserved but marked as a previously published claim pending source or measurement-period validation before amplification.
Explore related context
BerlengaTur caseWebsite, SEO, acquisition, measurement and experience in one system.
Marcoo caseMulti-tenant scheduling SaaS built from scratch.
ServicesCapabilities and engagement formats.
LabHypotheses, experiments and current learning.