While the room fills up · two questions in the chat

Tell us where you are coming from.

Drop your answers in the Google Meet chat. We will read what comes in before we begin.

1 The subject you teach and your years at SALCC.

2 One word: how do you feel about AI in your students' coursework right now?

Chani Saied, Bridge Point Scholars · Tuesday, 19 May 2026 · 10:00 AM AST

Tuesday, 19 May 2026 · A conversation with SALCC educators

Where we are.
Where we are going.

Bridge Point Scholars Ltd. · Castries, Saint Lucia
A working session with the educators of Sir Arthur Lewis Community College.

Section A · The hour

How we would like to use the time

1

Who we brought, and where we begin

Introductions, mission and vision, and how we read this moment in Caribbean education · 5 min

2

What students will see at the July launch

Platform walkthrough interleaved with three demos: Diagnostic + Quick Study, AI Tutor on a Caribbean item, SBA Helper · 25 min

3

CAPE in August

Phase 1 rails built, Phase 2 content fills in through summer, where lecturers shape what we build · 5 min

4

Bridge Point, Moodle, Keystone

Where the host LMS is thin, what we fill at launch, and what the institutional layer would add · 10 min · Jeff

5

Open discussion

Questions, concerns, and where this could go · 15 min

Section A · Introductions

Who we brought to the conversation

Four of us are with you today, each speaking to a different lens on what we are building.

Chani Saied

CEO & Co-Founder

Leads the conversation today and owns the academic vision, pedagogy, and institutional partnerships.

Jeffrey Saied

CTO & Co-Founder

Leads platform engineering, AI infrastructure, and the learning domain. Takes the architecture block (academic integrity, the Moodle comparison, and where this is heading).

Minelva Adrien Prosper

COO & Co-Founder

Runs operations, commercial structure, and the partnership side of any collaboration we agree on today.

Melicia David

Operations Officer

Capturing what we agree on so the follow-up is clean. Shares notes after the meeting.

Section A · What we are trying to do

Our mission and our vision

Two sentences that have not changed since we started. Everything else negotiable.

Mission

Make quality academic support accessible to every student in the Caribbean, regardless of location or income.

Vision

To be the Caribbean's academic infrastructure: built here, calibrated to our curricula, and trusted by the educators who shape the next generation.

Founded by Saint Lucian educators. The brief from day one was Caribbean curriculum first, not US tools adapted for the region.

Section B · Why we built it this way

Most EdTech is American.

Bridge Point is built where the students are.

We are aligned to CSEC and CAPE from the ground up rather than retrofitting US content for the region, and every question, setting, name, and currency reference defaults to Saint Lucian and broader Caribbean context. That brief has been the same since day one and is not something a future client can configure away.

Section B · What the data and the lecturers have surfaced

The diagnosis we are designing against

Two lenses. Dr. Sancha Meliat's empirical study of SALCC outcomes on the left. What lecturers across the region have flagged to us in the past eighteen months on the right.

SALCC institutional data · Meliat (2023)

What the regression analysis shows

Roughly one in four students ends semester one on academic probation, nearly twice the tertiary benchmark.
Foundation preparation predicts sustained GPA (logistic regression, OR = 4.17, p < .001).
Term-time employment is a school-side intervention point, not just a student-side problem.
Attendance does not predict sustained GPA. Standard LMS engagement alerts would misfire here.

What Caribbean faculty have told us

The pattern across the conversations

Mark schemes drift between lecturers who are each building their own quizzes inside the same course.
Student writing standards have declined since the pandemic and have not recovered to the level lecturers used to expect.
Generative AI use among students is outpacing institutional policy.
Foundation-level gaps surface late in CAPE Year 1, after the early signal would have been most useful.
Faculty time is going to grading rather than to teaching.

Our CAPE build is calibrated to both: Meliat's evidence on what predicts SALCC outcomes, and the practical pains lecturers have surfaced in working conversations across the region.

Meliat, S. (2023), pp. 269–273 (predictors); p. 228 (null attendance finding). Office of Institutional Effectiveness data, five-year window, n=180.

Section B · The literature behind our response

What the research has settled in the last six years

Four post-2020 findings that align with Meliat's evidence and shape what we built. Full citations on the references slide.

Nickow, Oreopoulos & Quan (2020)

NBER Working Paper 27476 · 96 randomised evaluations of tutoring

Personal tutoring is the most consistently effective intervention in the K–12 / post-secondary evidence base. Pooled effect size around 0.37 standard deviations, sustained across reading and mathematics. The challenge is delivering it at cost.

Kestin, Miller, Klales et al. (2024)

Harvard Physics · Scientific Reports · randomised controlled trial

A guided AI tutor produced learning gains roughly double those of in-class active learning, in less time. The intelligent-tutoring effect that earlier work suggested replicates with modern LLM-backed systems, provided the tutor is bounded and grounded.

Carpenter, Pan & Butler (2022)

Nature Reviews Psychology · synthesis of three decades of evidence

Spaced practice and retrieval practice are among the most reliable findings in cognitive science. Mastery without scheduled re-exposure decays. Our adaptive review system implements both directly.

Kraft (2020)

Educational Researcher · effect-size framing for education research

Recalibrates how to read effect sizes in education. Sizes above 0.05 SD warrant attention; above 0.20 SD are large in this field. Helps interpret what Meliat's predictors imply at scale, and what tutoring effects mean for cohort outcomes.

Meliat shows the SALCC-specific problem. The post-2020 literature shows that tutoring works, that AI-mediated tutoring replicates the effect, and that retention requires structured re-exposure. Bridge Point operationalises all three.

Section C · The product

Bridge Point Scholars, today

Built and ready for the July 26, 2026 public launch. The figures below are verified as of 5 May 2026. The platform is in pre-launch with a growing waitlist.

35,607

Questions in the bank

30

Subjects · CSEC and primary

554

Instructional videos

33

Full study guides

87

Embedded PhET simulations

36

CAPE syllabi indexed (Phase 1)

The platform is built at bridgepointscholars.com, headquartered in Castries, and opens to students across the OECS and the broader Caribbean at the July 26, 2026 launch.

Section C · The student-facing surface

What a student does on Bridge Point

Seven areas of the platform that are built and ready for the July launch, none of them on a roadmap slide.

Question bank

35,607 questions with worked answers and step-by-step explanations

CSEC and primary content is built and ready for the July launch, with CAPE following in August. Every item carries a model answer and a student-facing explanation rather than a marker's key alone, and this bank is the spine that the diagnostic, Quick Study, and the AI tutor all draw from.

Diagnostic

Intake diagnostic

A timed diagnostic at sign-up identifies gaps and builds the student's personal study plan before they ever touch a tutor.

Adaptive

Quick Study

A mastery-driven practice mode in which a Bayesian model decides the next-best question for this student, on this topic, today.

AI

AI Tutor

Caribbean-context and mastery-aware, grounded in our content database. It scaffolds the student through the problem and refuses to hand back the answer.

AI

Writing Corner

Five-dimension essay feedback covering strengths, structure, ideas, vocabulary, and one concrete suggestion. The tool never rewrites the student's prose.

SBA

SBA Helper

A Socratic coach for CSEC, CAPE, and CPEA School-Based Assessments that questions and guides but never generates the work. It auto-builds the CXC AI Disclosure Form from the audit log.

Access

Accessibility layer

A single layer of accessibility features the student can toggle on at any time, including the OpenDyslexic font, Reader View, math-to-speech, click-to-define, high-contrast and distraction-free modes, and a low-bandwidth toggle for the parts of the region where that matters.

Section C · Demo

Demo 1 · Diagnostic plus Quick Study

Recorded directly on the platform using a test account. The account runs the diagnostic, then the personal study plan it produces drives Quick Study.

Diagnostic

Quick Study

Speaker notes · ~45 sec before the videos, ~30 sec to close

You just saw the seven areas of the student-facing platform. Demo one puts the first two of those into action: the diagnostic and Quick Study. This is the student's first contact with the platform, before any tutor has met them. We open a brand new CAPE-aligned test account, run the diagnostic, and the result drives Quick Study. Two short videos back to back.

Watch the first video for what the diagnostic actually does. It is a timed CAPE-level set, but the question the platform is really answering is structural. What does this student know, with what confidence, and where are the gaps that matter. The output is not a score. It is a personal study plan.

The second video picks up where the first ends. Quick Study is the platform serving the next item the mastery model thinks the student should see right now. Bayesian shrinkage, paired-Elo difficulty, forgetting decay. The student sees one question, then the next one. The model is making that decision behind the scenes.

After both videos. From the student's perspective they answered a diagnostic and got handed a stream of relevant questions. From the platform's perspective the mastery model has been updated on every item, the next-best-question is being picked in real time, and a lecturer in a Keystone-enabled course would already be seeing the cohort distribution build up. Same surface, much deeper machine. Next, how the AI Tutor behaves on a Caribbean item.

Section C · How the AI behaves

The AI Tutor as it actually behaves

Three design choices that distinguish a curriculum-native tutor from a generic chatbot.

1

Grounded in our content database, not the open internet

The tutor answers from our CSEC question bank, our study guides, and (since April) every CAPE syllabus we have ingested. RAG retrieval, not free generation. The student gets material the institution can trace.

2

Mastery-aware, not stateless

The tutor knows what this student has mastered and what they have not, from the same Bayesian model that drives Quick Study. It scaffolds where the gap is, not where the student is already strong.

3

Caribbean by default, with the prompt boundary set high

Every question we generate uses Caribbean settings, names, currency, and context. Dr. Phulgence's framing of AI as a valuable educational tool aligned with regional values is built into our prompt architecture, not surfaced as a configurable option.

Phulgence, W. (2024). AI Global South Summit 2024 keynote on inclusive education and Caribbean heritage preservation.

Section C · Demo

Demo 2 · AI Tutor on a Caribbean Studies item

A Caribbean Studies or Communication Studies prompt. We focus on refusal patterns, citation grounding, and how the tutor responds when a student asks for the answer outright.

Speaker notes · ~60 sec before the video, ~30 sec to close

You just saw the three design choices behind our AI Tutor. Demo two is what those choices look like in a real conversation, on a Caribbean Studies prompt. This is the slide where the integrity properties show up in dialogue, not in a policy document.

Three things to watch for. One, the tutor refuses the obvious request. When the student asks for the answer outright, the tutor does not hand it back. It asks a question of its own, it scaffolds, it points at the part of the syllabus the item is testing. Two, citations are grounded. Every response that draws on CAPE content traces back to a specific syllabus reference or item in our content database, not to the open web. Three, when the student is already strong on a topic, the tutor does not over-scaffold. It moves the conversation forward.

One more thing. We deliberately tried to break it during the recording. That is the point. A tutor that only behaves on cooperative students is not a tutor a CAPE faculty member can hand to their class. Watch how it responds under pressure.

After the video. The two tools coming next, the Writing Corner and the SBA Helper, are where these same refusal patterns appear in essay and SBA contexts. Refusal is structural, not stylistic.

Section C · Caribbean integrity at the centre

The two AI tools a Caribbean lecturer should look at first

Both are built and ready for the July launch, and both are calibrated to coach a student through their thinking rather than write for them.

Writing Corner

Essay feedback that respects the writer

Five dimensions of feedback: strengths, structure, ideas, vocabulary, and one concrete suggestion to act on next.
The tool never rewrites the student's prose. It identifies issues and asks the student to repair them.
Every submission and its structured critique are stored against the student so the same feedback can surface in a lecturer view in the institutional rollout.
It is useful across Communication Studies, Caribbean Studies, History, English A, and English B drafts.

SBA Helper

A Socratic coach with an audit trail

A guided chat and per-dimension structured feedback for CSEC, CAPE, and CPEA School-Based Assessments.
The system prompt enforces a hard boundary: the helper never generates SBA content, and every interaction is logged.
At submission time the platform builds the CXC AI Disclosure Form automatically from an immutable log of the conversation.
The lecturer sees the student's process, not just the product, which makes plagiarism structurally harder.

Architecture aligns with CXC (2025) Generative AI Policy Framework. Implementation notes in docs/context/content-database.md.

Section C · Demo

Demo 3 · SBA Helper and the auto-generated disclosure form

A student opens an SBA dimension, asks for guidance, and the disclosure form the platform auto-generates from the interaction log at submission time.

Speaker notes · ~60 sec before the video, ~45 sec to close and hand to the next section

You just saw the two AI tools described side by side. Demo three is the second of those, the SBA Helper, in action. This is the architectural answer to the question every administrator asks when AI shows up in coursework: what happens when the student tries to get the tool to write the assessment for them.

Two things to watch for. One, the tool refuses to generate SBA content. The student in the video opens an SBA dimension, asks for guidance, and at one point tries to get the tool to write the introduction. Watch how the SBA Helper responds. It coaches, it asks Socratic questions, it points at structural gaps. It will not produce a paragraph the student can paste in. Two, every interaction is logged with a tamper-evident hash. The student cannot edit the log.

At the end of the video we open the auto-generated disclosure form. This is the document the lecturer receives at submission time, populated from the audit log, ready for the faculty member to countersign. The form reflects the real conversation, not a self-report.

After the video, I will pivot to what is shipping in August before Jeff takes the architecture story. The thread is simple: this is what AI in coursework will look like inside the platform we open in July, this is what lands next in August, and Jeff will then show how it sits alongside the LMS that SALCC already runs.

Section D · The build that lands in August

CAPE in August 2026

Phase 1 rails are built. Phase 2 content fills in through the summer with input from faculty at institutions that deliver CAPE.

Phase 1 · Engineering complete

The rails are ready

A selection of CAPE subjects is being onboarded across the summer, with the goal of opening Phase 1 of the CAPE content layer in August 2026.
Every CAPE syllabus uploaded and RAG-indexed; the AI tutor will know the syllabus content from day one of the August opening.
All three subscription tiers configured with a CAPE add-on at EC$5 above the current tier price: Explorer CAPE EC$14.99 / mo, Learner CAPE EC$34.99 / mo, Scholar CAPE EC$54.99 / mo.
CAPE tutoring sessions at EC$38 per session.
Reading Corner CAPE tab in place, ready to populate with CAPE-aligned readings.

Phase 2 · Through summer 2026

The content layer fills in

Per-subject topic tree: units, modules, specific objectives extracted from each syllabus.
Study guides per module, written and calibrated against the CAPE mark schemes.
Specimen papers ingested as practice. Applied Mathematics, Logistics, and Performing Arts ship with their specimens.
CAPE-aligned question generation across priority subjects, item-reviewed.

We are in active communication with CXC about content licensing for CAPE materials. That conversation, together with the lecturer feedback we are seeking from rooms like this one, is the deliberate reason CAPE Phase 1 opens in August rather than alongside the July public launch. The rails are ready. The content waits on the right partners and the right feedback.

Reference: docs/ROADMAP.md, CAPE Phase 1 completed 2026-04-29. Migration 20260523000000_cape_phase1.sql.

Speaker notes · ~90 sec · Chani, closing the demo block

CAPE is the centrepiece of why we are in front of you today. There are roughly eleven weeks between this conversation and the August launch, and that window is where lecturer input has the most leverage.

Left side is what we have built. The rails. Every CAPE syllabus is uploaded and indexed for the AI tutor. The three subscription tiers each carry a CAPE add-on at EC$5 above the current tier price, which puts the highest tier with full CAPE access at fifty-five Eastern Caribbean a month. CAPE tutoring sessions land at thirty-eight EC. The Reading Corner CAPE tab is built and ready to populate. When CAPE Phase 1 opens in August, a SALCC student will be able to subscribe to the CAPE tier and start using the AI tutor against the CAPE syllabus from day one.

Right side is what fills in over the summer. The per-subject topic tree, broken out into units, modules, and specific objectives. Module-level study guides, calibrated against the CAPE mark schemes. Specimen papers ingested as practice. Applied Mathematics, Logistics, and Performing Arts ship with their specimens. CAPE-aligned question generation across the priority subjects, item-reviewed.

The sentence at the bottom is the one I want to draw your attention to. CAPE Phase 1 lands in August, one month after the public launch, and that delay is deliberate. Two things are in motion. We are talking with CXC about content licensing, because if we are going to ship CAPE materials at this scale we want the right relationship with CXC behind it. And we are looking for lecturer feedback from rooms like this one, because a CAPE study guide that has not been read by a CAPE lecturer is not a CAPE study guide we want to ship.

The rails are ready. The content waits on the right partners and the right feedback. That is the August picture. Jeff is going to take the architecture block from here: how the integrity story is built into the platform itself, where the LMS that SALCC already runs is thin against what your own institutional data calls for, and where this can lead institutionally on SALCC's timeline. Over to Jeff.

Section E · Academic integrity by architectureJeff

Integrity by architecture, not by policy.

Detection has become a losing arms race, but architecture has not. We have built the integrity story into the platform itself: the coach is structurally prevented from generating the deliverable, the audit log cannot be edited by the student, and the disclosure form the lecturer receives stands up as evidence of how the work was actually produced.

UNESCO (2023) Guidance for generative AI in education and research calls for institutional infrastructure, not policy alone. CXC (2025) Generative AI Policy Framework is the regional anchor. Vanderbilt University (2023) retired Turnitin's AI detector after false-positive rates compromised its use. Detection failed; designed-in integrity is the alternative.

Speaker notes · ~75 sec · Jeff opens the architecture block, taking the handoff from Chani's CAPE slide

Thank you Chani. CAPE in August is what is shipping in front of students. The question every administrator asks alongside that, and the question we hear in every working session with educators, is what happens when AI shows up in coursework. So before we talk about how the platform sits alongside the LMS that SALCC already runs, one slide on the principle the architecture is built on.

There are two ways an institution can respond to AI in student work. The first is policy plus detection. Write a rule, scan submissions, try to catch the cheating after it happens. That is the path most institutions defaulted to in 2023 and 2024, and it has not worked. Vanderbilt is the headline example. They retired Turnitin's AI detector when the false-positive rate was high enough to be unsafe to use against a student.

The second response is architecture. Build the integrity properties into the tool itself, so the student cannot get the disqualifying behaviour out of it in the first place. That is the line we have taken.

Three architectural properties you just saw in the demos. One, the coach refuses to generate the deliverable. The SBA Helper discusses the student's ideas, asks Socratic questions, points at structural gaps. It will not write a paragraph the student can paste in. Two, the audit log is tamper-evident. Every interaction is logged with a cryptographic hash that the student cannot edit. Three, the disclosure form is auto-generated from the log. When the student submits, the platform produces the CXC AI Disclosure Form on their behalf, populated from the audit log. The lecturer reads a document that reflects the real conversation, not a self-report.

We agree with UNESCO and we agree with CXC. The institutional response to AI in coursework is not a stronger detector. It is a tool whose architecture makes the disallowed thing structurally hard to do, and produces evidence of how the work was actually produced. That is the principle the next three slides apply to Moodle and to the institutional path through Keystone.

Section E · How to read what comes nextJeff

Three things on the table, named clearly before we compare them

For the next few slides we are going to talk about Bridge Point, Moodle, and Keystone in the same breath. Naming them up front so the comparisons land cleanly.

1 · What you just saw

Bridge Point Scholars

The student-facing platform from the demos. Mastery model, AI Tutor, SBA Helper, Writing Corner, content database, accessibility layer. Built and ready, waitlist open now.

Built and ready · opens July 26, 2026

2 · What SALCC runs today

Moodle

The institution's LMS. Course shells, enrolment, the gradebook, proctoring. Excellent at course delivery, thin on the academic-quality layer Meliat's data calls for.

Already in production at SALCC

3 · What we are building

Keystone by Bridge Point

An LTI 1.3 institutional layer. Surfaces Bridge Point's mastery model, AI tutor, SBA Helper, and lecturer dashboards from inside the Moodle course shell. Standards-based, reversible.

In build · pilot-ready 2026/27

The next slide names where Moodle alone is thin against Meliat's evidence and the research, what Bridge Point will deliver for the student at the July launch, and where Keystone surfaces the same capability inside the institution's existing Moodle.

Speaker notes · ~60 sec · Jeff opens the architecture block

Chani has handed off to me for the architecture block. Before we get into the comparison table, one slide on terminology, because we are about to put three named things side by side and the comparisons only land if everyone in the room is reading the same labels.

Three layers. Read left to right.

Left is Bridge Point Scholars. That is what you saw demoed. Mastery model, AI Tutor, SBA Helper, Writing Corner, content database, accessibility layer. It is built and ready, a waitlist is filling now, the platform opens publicly in July, and that value does not depend on anything SALCC does institutionally.

Middle is Moodle. Your LMS. Course shells, enrolment, the gradebook, proctoring. Excellent at what it does. The two things I want to be honest about on the next slide are where it is thin against the evidence we walked through earlier, and how those gaps map to what your own institutional research has surfaced.

Right is Keystone. That is the institutional layer we are building on top of Bridge Point. It is an LTI 1.3 tool, the same standard your Moodle already speaks. Installs in roughly thirty minutes. Surfaces Bridge Point's capabilities from inside the Moodle course shell so students never log into a second platform and lecturers see a familiar interface.

With those three labels on the table, the next slide is the comparison itself. Where Moodle is thin, what Bridge Point will fill at the July launch, and what Keystone adds for the institution.

Section E · Where Moodle alone falls shortJeff

Six gaps named, mapped to what we deliver at launch and what Keystone adds

Read across each row: what the evidence calls for, where Moodle is thin today, what the student-facing Bridge Point delivers at the July launch, and what changes when the same capability lands inside Moodle through Keystone.

What the data calls for

Moodle today

What Bridge Point delivers at launch

What Keystone adds inside Moodle

Personal study plans driven by mastery, not snapshots

Static quizzes. Score-only attempts. No mastery model.

Ready at launch. Bayesian mastery, paired-Elo difficulty, exponential forgetting decay drive Quick Study and the AI Tutor.

Surfaced inside the course. Lecturers see mastery distribution per cohort, per CXC objective, on the same Moodle page.Carpenter, Pan & Butler 2022; Nickow, Oreopoulos & Quan 2020

At-risk alerts that match SALCC's reality

Engagement-based only. Fires when a student has not logged in.

Ready at launch. Content-level alerts driven by Bayesian mastery rather than login frequency.

Pushed to the lecturer. Mastery-based at-risk alerts appear in the lecturer's Moodle course view and trigger re-teach recommendations.Meliat 2023, p. 228 · null attendance finding

CAPE / CSEC-aligned content corpus

No regional curriculum. Lecturers build from scratch.

35,607 questions built and indexed. All 36 CAPE syllabi RAG-indexed. Writing Corner, SBA Helper, AI Tutor all draw from the same corpus.

Pull into Moodle. Lecturers attach Bridge Point items, study guides, and AI activities directly into Moodle course shells.Phulgence 2024 · Caribbean by default

AI tutor with academic-integrity audit

Generic text completion. No SBA guardrails, no mark-scheme calibration.

Ready at launch. SBA Helper refuses to generate content. Auto-builds the CXC AI Disclosure Form from an immutable audit log.

Embedded in course shells. The SBA Helper and its disclosure form attach to Moodle submissions automatically.CXC 2025; UNESCO 2023; Vanderbilt 2023

AI grading against real CAPE / CSEC rubrics

Manual grading. No rubric-aware AI. Lecturer time goes to marking, not teaching.

Built into the platform. Five-dimension Writing Corner feedback and AI grading calibrated against CXC-style rubrics.

Lecturer-tunable rubrics. Lecturers edit the rubric inside Moodle, AI grading recalibrates, scores pass back to the Moodle gradebook.Kraft 2020 effect-size framing

Per-objective mastery breakdown to the gradebook

A single numeric score per activity. No per-objective detail.

Ready at launch. Mastery tracked per CXC specific objective inside the learning domain.

Structured AGS passback. Per-CXC-objective mastery delivered to the Moodle gradebook alongside the numeric score.Meliat 2023, p. 257 · 54% lecturer concern

We are not asking SALCC to replace Moodle. Moodle keeps doing what it does well. Bridge Point fills the student-facing gaps at the July launch, and Keystone is what makes the same capabilities visible to lecturers inside the Moodle courses they already run.

Speaker notes · ~110 sec · Jeff

Six gaps. Read across each row. The first column is what Meliat's data, the post-2020 research, or both, calls for. The second column is where Moodle is thin against that. The third column is what Bridge Point delivers for the student at the July launch. The fourth column is what changes when that same capability lands inside Moodle through Keystone.

Row two is the one that comes straight out of your own data. Moodle's at-risk alerting fires on engagement. Meliat is explicit that attendance does not predict sustained GPA at SALCC. The most common alert Moodle produces is firing on a signal your own institutional research has told you to discount. Bridge Point's alerting fires on mastery instead. Keystone is what pushes those alerts into the lecturer's Moodle course view so they actually act on them.

Row one is the mastery model itself. A Moodle quiz attempt is a snapshot. Score, done. Bridge Point has the Bayesian model, paired-Elo, and forgetting decay built into the learning domain. Keystone surfaces the distribution per cohort and per CXC objective on the same page the lecturer is already on.

Row four is the integrity question. Bridge Point's SBA Helper plus the auto-generated CXC AI Disclosure Form is the architectural alternative to detection. Keystone is what attaches that disclosure form to Moodle submissions automatically.

I will not walk every row. The columns make the through-line explicit: Moodle is thin in named ways, Bridge Point delivers each of those things for the student at the July launch, and Keystone is the path that puts the same capabilities inside your existing Moodle without asking the institution to migrate.

Section E · Where this is headingJeff

Where this is heading

Read bottom up. Bridge Point is the student-facing platform that opens on July 26, 2026. Moodle is the LMS the institution already runs. Keystone is the institutional layer that closes the gaps the previous slide named, inside the Moodle SALCC already runs.

Moodle

SALCC's existing LMS continues to handle course shells, the gradebook, enrolment, and proctoring. Nothing about that workflow needs to move.

Keystone

Bridge Point exposed as an LTI 1.3 external tool, installable in about thirty minutes. Students launch it from inside their existing Moodle course and never log into a second platform. Lecturer dashboards with cohort mastery and at-risk indicators, AGS gradebook write-back so mastery scores land in Moodle automatically, the SBA Helper and its disclosure form embedded directly in course shells, AI grading calibrated to lecturer-edited rubrics, and re-teach recommendations the lecturer can act on in the next class.

Bridge Point

The mastery model, AI Tutor, SBA Helper, Writing Corner, content database, and accessibility layer. An out-of-school enrichment platform built and ready, opening publicly on July 26, 2026. A SALCC student can join the waitlist now and subscribe at the July launch to use it alongside their existing lectures, with or without an institutional agreement in place.

The student-facing layer is available today. The Keystone layer is reversible and standards-based, so if SALCC ever uninstalls the LTI tool the Moodle environment reverts cleanly. This is the direction the conversation can lead, when the institutional side is ready, on SALCC's timeline.

Speaker notes · ~75 sec · Jeff closes the architecture block

One more slide before I hand back to Chani for the lecturer ask. The same three layers from a few slides ago, now drawn as a stack so it is clear where each one sits.

Read it bottom up. The bottom layer is Bridge Point as we have built it. Mastery model, AI Tutor, SBA Helper, Writing Corner, content database, accessibility layer. A SALCC student can join the waitlist this week, subscribe when we open publicly in July, and use it alongside their lectures. That value does not depend on anything SALCC does institutionally.

The top layer is Moodle. Your LMS. It stays where it is. Course shells, enrolment, the gradebook, proctoring. We are not asking the institution to move any of that.

The middle layer is Keystone. Bridge Point exposed to Moodle as a standards-based LTI 1.3 tool. Roughly thirty-minute registration on your side. Students launch Bridge Point from inside their Moodle course and never log into a second platform. The features in the box are the ones the gap table on the previous slide called for. Lecturer dashboards, AGS gradebook write-back, the SBA Helper and disclosure form embedded in course shells, rubric-calibrated AI grading, re-teach recommendations.

Two things to land before I hand back. The institutional layer is reversible. If SALCC ever uninstalls the LTI tool, the Moodle environment reverts cleanly. There is no platform lock-in by design. And the institutional conversation is not what we are asking for in this hour. The ask in this hour is the lecturer collaboration around the August CAPE rollout that Chani walked you through. The Keystone layer is the direction this can lead, when SALCC is ready, on SALCC's timeline.

Back to you, Chani.

Section F · Where you fit

Where SALCC lecturers can shape the August rollout

Roughly eleven weeks of CAPE build between this conversation and the August launch. Faculty input inside that window is high-leverage.

1

Subject prioritisation

CAPE is thirteen subjects across two blocks. Tell us which units should deepen first and we will sequence the content build accordingly.

2

Mark-scheme review

Help us calibrate AI grading on real CAPE rubrics. Two hours of a lecturer's time per subject shapes how thousands of student attempts get marked.

3

Sample-question feedback

As we generate CAPE-aligned items, we want lecturers to flag what is off. The feedback loop tightens item quality faster than internal review alone.

4

Specimen-paper context

Where the CXC specimens are thin, your lecturers know which past sessions and which exam patterns matter. We capture that as classroom-grade signal.

This is the closest equivalent of co-authorship without contracts. Your lecturers' fingerprints end up on the content their students will eventually learn from.

Section F · Pass it on

Help us reach more Caribbean educators

Our public launch for CSEC and primary lands in July, and we are aiming to bring CAPE online in August. Here are four ways SALCC educators can help us reach the students who would benefit, before any commercial conversation begins.

1

Share this deck across SALCC

Send the slides to colleagues, other departments, and educators at sister institutions. The deck is self-contained and opens in any browser.

2

Mention us to your students

CSEC and primary students can use Bridge Point today, and CAPE students will be able to join when the August rollout opens. A short word from a lecturer travels further than any advertising we can do.

3

Refer fellow educators across the region

Anyone who wants to follow the August CAPE rollout, contribute lecturer feedback, or just stay in the conversation as the build progresses is welcome.

4

Stay connected

Subscribe for product updates, follow our channels, and reach out directly for any follow-up. We answer every email a SALCC educator sends.

Whichever of these routes works for you, we are committed to addressing any question a SALCC educator raises. Email, a follow-up call, a working session with a department, or a written response that we send across SALCC: pick the format that fits and we will meet you there.

Section G · Take it with you

Two scans, everything you need

Save Chani's contact directly to your phone, or open the resources hub for the flyer, the one-pager, this deck, the tutor application, and student sign-up.

Scan 1 of 2

QR code containing Chani Saied's contact information as a vCard

Save Chani's contact

Adds Chani Saied (CEO & Co-Founder) to your phone contacts with email, phone, and website.

Scan 2 of 2

QR code linking to bridgepointscholars.com/salcc resources hub

All the SALCC resources

Flyer, one-pager for SALCC educators, this presentation deck, the tutor application, and student sign-up — all on one page.

Thank you.

For your time, your questions, and for spending part of your morning thinking with us about what CAPE delivery could look like in August.

partnerships@bridgepointscholars.com +1 758 712 5252 bridgepointscholars.com

References

Sources cited in this presentation

A combination of recent post-2020 work alongside Meliat's SALCC-specific evidence and CXC's regional AI policy.

Carpenter, S. K., Pan, S. C., & Butler, A. C. (2022). The science of effective learning with spacing and retrieval practice. Nature Reviews Psychology, 1, 496–511.

Caribbean Examinations Council (2025). Generative AI Policy Framework. CXC, Bridgetown, Barbados.

Kestin, G., Miller, K., Klales, A., Milbourne, T., & Ponti, G. (2024). AI tutoring outperforms in-class active learning: a randomised controlled trial. Scientific Reports (Harvard Physics).

Kraft, M. A. (2020). Interpreting effect sizes of education interventions. Educational Researcher, 49(4), 241–253.

Meliat, S. (2023). Investigation of the Influence of School and Student-Based Factors on Sustainable Academic Performance at Sir Arthur Lewis College, Saint Lucia. PhD thesis, Unicaf University.

Mollick, E. R., & Mollick, L. (2023). Student use cases for AI: start of the school year. Harvard Business Publishing / SSRN.

Nickow, A., Oreopoulos, P., & Quan, V. (2020). The impressive effects of tutoring on PreK–12 learning: a systematic review and meta-analysis. NBER Working Paper No. 27476.

Phulgence, W. (2024). Keynote address. AI Global South Summit 2024: inclusive education and Caribbean heritage preservation.

UNESCO (2023). Guidance for generative AI in education and research. United Nations Educational, Scientific and Cultural Organization, Paris.

Vanderbilt University Office of the Provost (2023). Guidance on AI detection tools in academic work. Public communication on the discontinuation of Turnitin's AI detector.

Tell us where you are coming from.

Where we are.Where we are going.

How we would like to use the time

Who we brought to the conversation

Our mission and our vision

Most EdTech is American.

Bridge Point is built where the students are.

The diagnosis we are designing against

What the regression analysis shows

The pattern across the conversations

What the research has settled in the last six years

Bridge Point Scholars, today

What a student does on Bridge Point

Demo 1 · Diagnostic plus Quick Study

The AI Tutor as it actually behaves

Grounded in our content database, not the open internet

Mastery-aware, not stateless

Caribbean by default, with the prompt boundary set high

Demo 2 · AI Tutor on a Caribbean Studies item

The two AI tools a Caribbean lecturer should look at first

Essay feedback that respects the writer

A Socratic coach with an audit trail

Demo 3 · SBA Helper and the auto-generated disclosure form

CAPE in August 2026

The rails are ready

The content layer fills in

Integrity by architecture, not by policy.

Three things on the table, named clearly before we compare them

Bridge Point Scholars

Moodle

Keystone by Bridge Point

Six gaps named, mapped to what we deliver at launch and what Keystone adds

Where this is heading

Where SALCC lecturers can shape the August rollout

Subject prioritisation

Mark-scheme review

Sample-question feedback

Specimen-paper context

Help us reach more Caribbean educators

Share this deck across SALCC

Mention us to your students

Refer fellow educators across the region

Stay connected

Two scans, everything you need

Save Chani's contact

All the SALCC resources

Sources cited in this presentation

Where we are.
Where we are going.