View On
tagtagtag

Kasti AI

Kasti AI is a multi-tenant K-12 platform that allows teachers to generate lesson plans using AI based on their institutional curriculum. Districts upload their own standards and curricula to the portal rather than selecting from a predefined library. Kasti processes district curricula to keep lesson plans aligned while adapting them to teacher preferences and student accommodations. Teachers can customize plans for special education students, organize them into folders, and share copies with colleagues. Kasti reports that teachers save ten or more hours a week on lesson planning. OCloud Solutions merged external code into Kasti mid-build and launched it across Dev, UAT, Demo, and Prod environments. The engagement then grew to include Clever and ClassLink SSO, pipeline stabilization, and a twice-yearly Learning Commons standards sync.
Kasti AI
POPULAR FEATURES
Feature | Multi-Tenant Platform
Multi-Tenant Platform
A single platform hosts multiple institutions, each managing its own curriculum, users, and lesson plans for teaching and collaboration.
Feature | Role-Based Access
Role-Based Access
Permissions are assigned based on user roles, and platform-level roles have broader access than tenant-level roles.
Feature | Custom Curriculum
Custom Curriculum
Institutions upload their own curricula, and the platform generates customized lesson plans aligned with their district standards.
Feature | Lesson Generation
Lesson Generation
AI drafts each plan from uploaded district curriculum, and educators adapt it by topic, hook, or special education need.
Feature | SSO Integration
SSO Integration
Institutions that use Clever or ClassLink can sign in through their existing systems without additional credentials.
Feature | Lesson Management
Lesson Management
Instructors can organize lesson plans in folders, add notes, and share or copy plans with other instructors within their institution.
Feature | Document Processing
Document Processing
Google Document AI performs OCR on standards, curricula, and lesson plans, and Claude converts those results into structured output.
Feature | Standards Library
Standards Library
Learning Commons provides K-12 academic standards that are normalized and matched with extracted lessons for standards alignment.
Feature | Sync Controls
Sync Controls
Superusers can view sync history, trigger imports, or reprocess files, and a job refreshes standards every January and July.
PROJECT REQUIREMENT

The client needed an engineering partner to stabilize a live codebase, integrate it with district systems, and continue improving it.

  • Integrate new code into the existing Kasti codebase and set up independent Dev, UAT, Demo, and Prod environments.
  • Refine the AI parsing and processing logic for curriculum processing and lesson plan preparation to ensure more consistent outcomes.
  • Deliver enhancements and fixes as requirements change, including lesson library improvements driven by instructor and client feedback.
  • Connect Clever and ClassLink for centralized authentication and rostering across participating educational institutions.
  • Bring in Learning Commons standards with automatic updates twice a year to maintain an accurate public catalog.
  • Build an admin portal for platform-wide visibility, content management, and simpler day-to-day administration.
Kasti AI project requirement diagram — stabilize, integrate, and improve a live codebase: merge code, admin portal, refined parsing, SSO and rostering, enhancements, and standards sync.
OUR APPROACH
OCloud Solutions stabilized the platform first and introduced a structured sprint process once the codebase was dependable.
Agile Stabilization
Agile Stabilization
The team spent five months testing the platform, fixing issues, and integrating Clever and ClassLink before adding new features.
Sprint Cadence
Sprint Cadence
Weekly sprints with a Thursday UAT release gave the client a predictable point to review every change.
Environment Separation
Environment Separation
Dev, UAT, Demo, and Prod operated as separate environments, so testing ran alongside development before any change reached production.
Feedback Loop
Feedback Loop
Instructor and client requests shaped the GitHub-tracked backlog, so every enhancement and fix addressed a need they had raised.
Cloud Logging
Cloud Logging
Google Cloud Logging monitored core processing and surfaced failures in Firebase Functions early, which shortened resolution time.
Parallel Extraction
Parallel Extraction
The team designed parallel per-lesson extraction to save an estimated 10 to 20 minutes on a 400-page curriculum.
PROJECT SUCCESS

OCloud Solutions delivered Kasti across six milestones and solved each technical challenge along the way.

  • Clever and ClassLink each require approval before a platform goes live. The team worked directly with each provider to get Kasti approved.
  • Standards and curriculum parsing broke on real district documents. The team traced each issue, shipped fixes when ready, and lifted parsing accuracy from 70% to 92%.
  • Lesson plan generation had failures in its core pipeline. The team addressed each issue individually, which reduced generation failures by 40%.
  • Kasti needed external code merged into an unfinished codebase. The team integrated it without pausing development or breaking live features.
  • Importing the same Learning Commons release twice would duplicate standards. The sync now checks the version and imports each release only once.
  • Instructors and clients kept asking for changes as they used the platform. The team turned those requests into UI and user experience improvements.
Kasti AI project success — Clever and ClassLink go-live approvals, parsing accuracy raised to 92%, 40% fewer generation failures, code merge, import-once standards sync, and UI/UX refinements.
RESULTS
OCloud Solutions turned a fragile, in-progress build into a dependable AI platform now serving 51 teachers.
Generated 100+ lesson plans from 50+ district curricula and pushed the successful generation rate above 97%.
Made Learning Commons into a self-updating catalog for four subjects, reducing manual standards work by an estimated 60 to 80%.
Opened access for any Clever or ClassLink district and gave operators admin controls to reprocess syncs without engineers.
Automated deployments across several environments to CI/CD pipelines, reducing manual deployment time by 50-70%.
Established four dedicated environments, reducing test setup and delivery time by 30–40% while enabling parallel testing.
06
Delivered 3 to 4 major features monthly, about 9 to 12 per quarter, alongside fixes and changing requirements.
Kasti AI results — 100+ lesson plans generated from 50+ district curricula, 97%+ generation success, 60–80% less manual standards work, and 50–70% faster releases across four environments.
TECHNOLOGIES WE USE
Technology | Next.js
Next.js
Technology | Firebase Authentication
Firebase Authentication
Technology | Firebase Firestore
Firebase Firestore
Technology | Firebase Functions
Firebase Functions
Technology | Google Cloud Document AI
Google Cloud Document AI
Technology | Google Cloud Logging
Google Cloud Logging
Technology | Anthropic Claude API
Anthropic Claude API
Technology | LangGraph
LangGraph
Technology | Learning Commons Knowledge Graph
Learning Commons Knowledge Graph
Technology | Clever
Clever
Technology | ClassLink
ClassLink
Technology | GitHub Projects
GitHub Projects

Explore Our Success Stories

Slide 1 of 24
Contact Information
Phone
ocs logo
We are a trusted software development agency focused on setting the future standards of the industry. OCloud stays true to their values and always aims to exceed our clients' expectations.
ISO 9001:2015 CertifiedISO 9001:2015 Certified
social logossocial logossocial logos
United States
1309 Coffeen Avenue,
Sheridan, Wyoming, United States
+1 973 922 9512
Pakistan
6/B Block XX (CCA, Phase 3 Khayaban-e-Iqbal), DHA Cantt, Lahore, 54000
+92 323 700 0555
© 2026 OCloud Solutions | All Rights Reserved