Execution over advice
Our systems perform work — they do not produce recommendations for someone else to implement. An assistant that drafts, checks, files and reports is worth more than a dashboard that describes the backlog.
About the Company
Augmented Ingenuity
TICNES AI is a technology company that builds applied artificial intelligence, data science systems and information-security engineering for organisations operating in complex, regulated and low-connectivity environments. We design systems that execute, and we leave the decisions where they belong: with the people accountable for them.
Who We Are
A company built around execution, not demonstrations
TICNES AI was founded in 2015 by systems engineers working on multi-agent coordination and reinforcement-learning frameworks. What began as research into how autonomous components could cooperate without a central controller became a working method for building software that runs operations rather than advising on them.
Today the company operates a portfolio of digital products under the ticnes.ai domain, each one addressing a concrete problem: matching students to scholarship funding, securing and organising personal media archives, explaining Angolan labour legislation, measuring public sentiment across Portuguese-language sources, and training small and medium businesses to automate their own workflows.
We are deliberately small and deliberately technical. Every engagement starts with a systems audit, moves into a scoped build, and ends with a handover of a running system — documented, monitored and owned by the client. We do not sell strategy decks, and we do not deliver pilots that never leave the pilot stage.
2015
Founded
20+
Systems Deployed
6
Live Platforms
3
Engineering Disciplines
Our Mission
Bridge human creativity and autonomous execution
Software should absorb the repetitive work and return judgement to the people who carry responsibility for it. That single principle shapes every system we ship.
Humans hold the decision
Automation proposes and prepares. It does not approve, sign, disburse or sanction. Every consequential step in our architectures requires a named human authorisation.
Sovereignty by default
Client data stays inside infrastructure the client can inspect and revoke. Encryption in transit and at rest, tenant partitioning and auditable access are baseline requirements, not upgrades.
Built for real constraints
Intermittent connectivity, mixed device fleets, Portuguese-language content and thin engineering teams are design inputs. A system that only works on a fast connection is not finished.
What We Do
Three engineering disciplines
Multi-agent systems & retrieval pipelines
Agent swarms with bounded tool access, retrieval-augmented generation over private document sets, prompt and output guardrails, and evaluation harnesses that measure whether an assistant is actually correct before it reaches production.
Predictive analytics & signal processing
Forecasting, classification and natural-language measurement built on real operational data, with attention to Portuguese-language sources, sparse datasets and the reporting formats regulators and boards actually read.
Zero-trust architecture & applied cryptography
Identity and access design, namespace isolation for autonomous workloads, sandboxing for generated code, TLS 1.3 transport, AES-GCM-256 at rest and post-quantum key exchange planning for long-lived records.
Platforms & Initiatives
What runs under ticnes.ai
Each platform is an independent application on its own subdomain. This page is the index; the systems themselves are at the links below.
Scholarship Hackathon 2026
A scholarship programme for university students, run as an open competition: candidates submit applied work, submissions are evaluated against published criteria, and awards are disbursed to the winners.
Open platform →Photo Vault
A local-first desktop photo library with on-device vision captioning, EXIF-aware metadata, folder-scoped full-text search and editable tagging. Images are indexed locally, not uploaded.
Open platform →Student AI Tutor
A study assistant that works through problems with a student step by step, in Portuguese or English, built for the syllabus and bandwidth realities of secondary and tertiary learners.
Open platform →Q-Mvutu
An intelligent social-research engine that collects and classifies public opinion across Portuguese-language sources, returning sentiment and theme distributions rather than raw keyword counts.
Open platform →LGT-AO
An assistant for the Angolan General Labour Law (Lei n.º 12/23). It locates the governing article, quotes it, and explains the obligation in plain language — for HR teams and for the workers on the other side of the contract.
Open platform →TICNES AI BootCamp
Practical automation training for small and medium businesses: identifying the tasks worth automating, running the first automations safely, and keeping the result working after the trainer leaves.
Open platform →Chronology
How the company developed
2015 — Foundations
Early multi-agent coordination models and reinforcement-learning frameworks. The company is founded by systems engineers and applied researchers.
2018 — First production agents
An automated route-scheduling deployment moves from research prototype to daily operation, with measurable efficiency gains in scheduling overhead.
2021 — Neural guardrails
Proactive defence work: output filtering, prompt-injection containment and sandboxed execution for generated code become a standing part of the architecture rather than a bolt-on.
2024 — Agent swarms at organisational scale
The unified automation stack connects language models to secure sandboxes and internal systems, allowing multi-step business processes to run end to end with human checkpoints.
2026 — Sovereign integration
A portfolio of six live products and more than twenty enterprise deployments, with data-sovereignty requirements treated as an architectural constraint from the first meeting.
Working With Us
How an engagement runs
1. Systems audit. We map the process, the data and the failure modes before proposing anything. If automation is the wrong answer, we say so.
2. Scoped build. A defined deliverable with acceptance criteria, built against the client’s own data and infrastructure wherever possible.
3. Deployment and handover. The system goes into production with monitoring, documentation and a named owner inside the client organisation.
4. Support. Production monitoring with a defined response window, and a documented escalation route.
To start a conversation, write to dr.jorgeclemente@proton.me or use the contact page. We respond within one business day.