recallai
Trust CenterSign in
Grounded in cognitive science

Reduce ramp time and compliance risk across support, operations, clinical, and field teams.

RecallAI turns your SOPs and training docs into AI-generated flashcards with FSRS-powered review so people retain critical procedures and improve 30-day recall. FSRS has been benchmarked on hundreds of millions of real review events and consistently outperforms classic SM-2 style scheduling on predictive recall accuracy.

Evidence behind FSRS
  • Open Spaced Repetition benchmark reports FSRS variants beating SM-2-family baselines across 10,000 collections and 519M+ evaluation reviews (with same-day reviews).
  • FSRS4Anki wiki documents the algorithm, optimization process, and links to peer-reviewed spaced-repetition research by the same research line.
  • The Anki manual explains how FSRS schedules by predicted recall probability (retrievability) and desired retention, replacing older manual interval tuning.
Book demoSee platform overview
How it works

From document to knowledgeable team, in minutes

Step 01
Upload your knowledge base
Drop in PDFs, Word docs, or paste text. RecallAI parses your SOPs, policies, and training materials automatically.
Step 02
AI generates flashcards
Claude reads your documents and extracts the key facts, procedures, and policies, turning them into targeted Q&A cards ready for review.
Step 03
Your team reviews daily
5–10 minutes a day. FSRS surfaces the right cards at the right time for each person, building lasting knowledge across your whole team.
5–10 min
Daily review time
~80–90%
30-day retention (pilot cohort, n<50; preliminary)
FSRS 5
Adaptive scheduling algorithm
Any doc
PDF, Word, or plain text

How measured: Early indicators based on internal pilot usage and product telemetry; methodology and sample details will be published at Methodology.

The science of remembering

Why most training doesn't stick, and why FSRS does

Without reinforcement, people forget up to 70% of new information within 24 hours. Workshops, slide decks, and one-off sessions create the illusion of learning without producing lasting results.

FSRS (Free Spaced Repetition Scheduler) is an open-source memory algorithm built on decades of cognitive science research. It tracks retention at the individual level (for each card, for each person) and estimates the next review window before predicted forgetting occurs.

Adapts to each person
Review intervals adjust to individual memory curves, not a fixed schedule.
Timing-aware reviews
Reviews are scheduled based on recall probability estimates, not a fixed interval.
Knowledge that compounds
Each review strengthens the memory trace. Retention improves over time, not in spite of it.
SM-2 baseline (Anki legacy scheduler)
Benchmark metrics on 4,632,965 held-out reviews (lower is better)
26%
Log Loss 0.7317
41%
RMSE 0.4066
21%
RMSE(bins) 0.2079
FSRS v4 (best in benchmark)
Measured benchmark results from ankitects/fsrs-benchmark
Dataset: 71 users, 4,632,965 evaluation reviews. Lower bars indicate better calibration and recall prediction error.
61%
Log Loss 0.3874
67%
RMSE 0.3347
95%
RMSE(bins) 0.0459
Talk pricing and rollout
For team leads

Full visibility over your team's knowledge

Manage your knowledge library
Upload documents, review AI-generated cards, and approve them before publishing to your team.
Onboard new members instantly
Invite by email. New members receive a setup link and are reviewing within minutes.
Spot knowledge gaps early
Retention scores per topic show exactly where the team is confident and where they are not.
Mandatory decks
Mark critical policy decks as required; they auto-assign to every team member on join.
SL
Sarah L., Team Lead
Acme Support
Team members
24+3 this week
Active decks
82 pending review
Avg. retention
87%↑ 4% vs last month
Reviews today
14268% completion rate
Enterprise readiness

Built for enterprise operations

Align rollout, governance, and identity with your existing operating model.

Identity and access
  • SSO authentication for managed enterprise sign-in.
  • SAML and OIDC identity provider integration.
  • Role provisioning and deprovisioning workflows for admins and managers.
Data flow and lifecycle
  • Ingests documents you upload (PDF, Word, plain text) and parses them into reviewable card content.
  • Stores extracted knowledge, card history, and review telemetry scoped by organisation.
  • Retention and deletion controls for content lifecycle and account closure workflows.
Admin controls
  • Hard organisation isolation on data access paths to prevent cross-tenant exposure.
  • Role-based permissions for Agent, Manager, Admin, and Super Admin actions.
  • Admin-facing audit trails for content changes, role updates, and policy-sensitive actions.
Rollout model
  • Pilot: start with one team and a bounded set of SOP decks.
  • Team rollout: expand to adjacent teams with manager-level oversight and required decks.
  • Org expansion: scale with formal governance, identity integration, and security review.

Need a deeper architecture walkthrough or help mapping controls to your procurement checklist?

Talk to security/sales engineering
Security & privacy

Enterprise-grade security, by design

Your SOPs and training materials are sensitive. We organize our security program into clear trust pillars with concrete controls and auditable proof artifacts.

Data Protection
Your content stays confidential and segregated from other organizations at every stage of processing.
  • TLS 1.3 encryption for data in transit.
  • Extracted text is encrypted at rest and scoped to your organisation.
  • Hard organisation ID scoping on all data-access paths.
Proof artifacts: data flow diagram, encryption key management summary, tenant-isolation test evidence.
Access Control
Only the right people can access the right actions, with permissions aligned to role and risk.
  • Role-based access control across Agent, Manager, Admin, and Super Admin roles.
  • Managers and above control document upload and card publication actions.
  • Signed httpOnly session cookies and single-use password reset tokens (1-hour expiry).
Proof artifacts: role-permission matrix, auth/session control spec, access-review checklist.
Application Security
The platform is designed to reduce abuse and malicious payload risk before it reaches business logic.
  • Magic-byte file signature verification blocks disguised uploads before parsing.
  • Per-user rate limiting on all API endpoints with tighter limits for login/reset/upload routes.
  • Password hashing with bcrypt (12 salt rounds).
Proof artifacts: secure upload validation test logs, API abuse-throttling policy, password storage standard.
Compliance & Governance
Security operations are backed by documented policies, accountability, and customer-facing transparency.
  • Documented incident response process with defined escalation and communication paths.
  • Published privacy and data-processing terms for enterprise review.
  • Versioned security documentation and control ownership records.
Proof artifacts: incident response summary, policy register, governance review cadence and owners.
Trust Center
Need details for procurement or security review? Request our trust documentation package.
Security overview
DPA / privacy terms
Subprocessor list
Incident response summary
Certification status
Review security

Ready to build a team that remembers?

Get in touch to see RecallAI in action with your own documents and team.

Book demo
recallai© 2026 RecallAI. All rights reserved.
Trust CenterSecurityPrivacy PolicyTermsData Processing AddendumSubprocessorsSign in