Talent · hire, review, develop
AI ranks. People decide.
Top Applicants puts your shortlist in order, on resume match blended with the hiring manager’s own evaluations. It never picks from it. Your people do.
With reviewee and reviewer weighting
With course-to-skill mapping
No quota logic anywhere in the product
Trusted by the Caribbean’s best-known employers
Airlines, telecoms, utilities, hotels, credit unions and the National Insurance Board itself.

Recruitment
From a blank job post to a ranked shortlist
It starts where hiring usually stalls. A hiring manager, looking at an empty job post.
AI-assisted job posts
Help that gets a manager from nothing to something they can edit.
A manager types a job title and gets a draft to edit. Nothing goes anywhere until they say so. It’s aimed at the blank page, because that’s where most hiring delays really start. The requisition is approved. Then the post sits unwritten for a fortnight, because nobody has an hour to start it.
The person keeps the pen. Same principle as the ranking below. The machine does the mechanical part. A person makes every decision that matters.
Ranking, described exactly
A ranking you can’t see inside is hard to trust. So here’s how ours works.
Resume analysis produces a match score, and Top Applicants ranks on a weighted blend of that match score and hiring-manager evaluations. It runs on Textkernel plus a local sentence-transformer model.
The blend is what matters. A ranking that only read resumes would ignore everything the people doing the hiring have already learned. AI ranks, people decide.
With custom questions and auto-marking
With placeholders, so a batch is not impersonal
With resume auto-fill
Job posts and applicants, with no quota logic
Performance reviews
Configurable enough to match how your organisation really reviews people. And there are exactly two places a model gets involved.
The cycle
- Templates with six question types
- Reviewee and reviewer weighting
- Up to five approval stages
- Overall commenters
- Staged visibility, so nobody sees a rating before they should
- Self-assessment, and company KPIs
Where AI is involved, exactly
Claude on AWS Bedrock summarises the Feedback section of a cycle, cached for 24 hours.
AWS Comprehend scores sentiment on feedback answers.
That’s all of it. There are no AI performance-trend alerts. The flight-risk score in people analytics is rule-based, and that’s better. Rules you can read are rules you can argue with.
What this is not
It isn’t all-round or multi-source review, so we don’t use the shorthand for it. You get multiple raters and overall commenters, and that covers much of what people want from the idea. What you don’t get is peer, direct-report or external rater roles, or anonymity handling. The shorthand would promise something the cycle doesn’t do, and you’d find that out during configuration.
Goals and OKRs live on the employee record and connect into the cycle, so a review can reference what was agreed.
Learning and skills
A library people browse, and training an employer can direct. Two different jobs. Most systems only do the first.
The library
- Courses and lessons
- Quizzes with retry
- PDF certificates
Employees browse and enrol themselves. That’s the right default for development people choose for themselves.
The skills bank
Skills held at levels 1.0 to 5.0, with course-to-skill mapping. Finish a course and a skill moves. And you can actually see a capability gap across a team.
Training an employer can direct
This is what turns a library into something you can use for onboarding and compliance-style training.
An employer can assign a course to a person or to a group, and track who has completed it.
That matters more than it sounds. Self-serve enrolment answers “what can I learn”. Assignment answers “did everyone who needed this actually do it”. And that second question is the one an employer has to be able to answer about a safety briefing, a policy update or a new starter’s first week.
It pairs well with onboarding templates on Core HR, where the rest of a new starter’s first week gets set up.
What customers say
“As a very well established company we use Triblock to automate many of our existing processes from payroll to recruitment to employee data management and it does the job very well.”

Nina MaynardHR Director, The Myers Group of Companies
“As a growing company, it was becoming increasingly difficult to keep up with our employees across all of our divisions and departments. TriblockHR provided us with an all-in-one solution that ticked many boxes.”

Brent BurrowsSenior VP, Commonwealth Building Supplies
Implementation
Live in six weeks. Implementation included.
- 1
Send us your current payroll export.
A spreadsheet or an export from your current platform. We map it.
- 2
We configure your territory engines and migrate your records.
Employees, year-to-date figures, leave balances and statutory registrations, with training for your team on every module you take.
- 3
Parallel run, then go live.
One cycle side by side with your current process. Then you switch.
Standard implementation is included with every subscription for go-lives within six weeks of signature, and monthly stakeholder meetings continue after launch.
Articles
Further reading
What we mean when we say AI ranks and people decide
Where the model stops and the hiring manager starts.
Read the article ArticleScoring a candidate without letting the score make the decision
A ranking is an ordering.
Read the article ArticleWhy a rule-based score beats a model you cannot read
Rules you can read are rules you can argue with.
Read the articleQuestions talent leads ask us
Does the review cycle support all-round, multi-source feedback?
What exactly does the AI do in performance reviews?
How does applicant ranking work?
Can we require a course?
Is there a limit on job posts or applicants?
How do goals connect to reviews?
See it on your own data
Bring a real job post and a handful of resumes. We’ll run the ranking in front of you, including the ones it puts at the bottom.
- Resume matching and the Top Applicants blend
- A review cycle with its approval stages and weighting
- Assigning a course and tracking completion
- 18,000+ employees paid across six territories
- SOC 2 Type II attested
- AES-256 at rest
- 99.9% uptime
- Built in The Bahamas
One product, four parts
Everything here shares one employee record, one permission model and one audit trail. The modules are how we license it. Underneath, it’s one system.
Security: AES-256 at rest via AWS KMS, TLS 1.2 and above in transit, SOC 2 Type II attested against the Security criteria with the report available under NDA, and AWS US multi-AZ as standard with on-premises or Caribbean regional cloud on request, subject to scoping. Sign-in is by Microsoft Entra ID single sign-on where you use it; multi-factor authentication comes through Entra ID. See also Core HR, Payroll and Workforce · the six territories · the HR and payroll glossary · clients, blog and contact.
