📣 Fall 2026 Lyceum Institutional is now open for university enrollment — see the pilot partnership program →
🤖 Anatomy of the Agent

Plans. Tools. Memory.
Action.

Lyceum isn't a chat box wrapped around an LLM. It's a full agentic system that plans multi-step study paths, calls verified tools, retrieves from your textbooks, and stays with each student across semesters — exactly the way a great human tutor would.

🧭
01 · PLANS

Multi-step planning

Given "I'm stuck on series convergence," the Agent doesn't just answer. It diagnoses prerequisite gaps, builds a 4-step path through the right concepts, and schedules a follow-up.

⌖ diagnose 📐 plan path 🎯 execute 🔁 schedule retry
🛠️
02 · TOOLS

Real tool use

The Agent calls Python (sympy / numpy) to verify derivations, retrieves from your uploaded textbooks via vector search, runs Wolfram for symbolic math, and queries the curriculum graph — every call is logged and auditable.

🐍 Python sandbox 📚 Textbook RAG 🧮 Wolfram 🌐 Web search 📊 Knowledge graph
🧠
03 · MEMORY

Cross-semester memory

The Agent remembers Maya struggled with LIATE in week 7, that she prefers visual derivations, that she always reviews on Sunday nights — and weaves all of it into how it teaches her in week 14.

📝 episodic 📋 semantic 🎨 preference ⏱ scheduling
04 · ACTION

Autonomous workflows

Overnight, the Agent runs jobs without prompting — generates next week's practice set, drafts a class digest for Dr. Hartman, and pages the dean of students about a struggling cohort it identified at 02:14 AM.

📄 Practice generation 📨 Faculty digest Advisor alerts
🔁
Multi-turn agentic loops, not single-shot prompts. A typical hard tutoring session at Lyceum is a 14-step trace: plan → retrieve → reason → tool-call → verify → respond → wait → diagnose → re-plan → ... The student sees a teacher. Internally, an Agent is running.
🧬 The frontier stack

Six frontier models,
one routing brain.

The Agent picks the right brain for each request — deep reasoning for proofs, fast clarifications for follow-ups, long-context for full-textbook ingestion. Multi-vendor by design — never a single point of failure for your students.

Tokens · last 30 days
12.4B
+18% MoM
Frontier models routed
6
across 4 providers
Median first-token
380ms
P95 · 1.2 s
Inference uptime
99.97%
90-day rolling
🧬 Active routing ⚙️ Failover topology 📊 Cost per task
All providers healthy 6 models
Aa Model ⚙️ Provider Routed for Share P50
🧠Claude Opus 4.7
Anthropic
Multi-step proofsLong humanitiesAmbiguous tutoring
28% 2.1 s
Claude Sonnet 4.6
Anthropic
Everyday tutoringSocratic dialogueConcept explanation
34% 820 ms
🌬️Claude Haiku 4.5
Anthropic
Fast clarificationsFollow-upsSmall steps
14% 320 ms
💻GPT-5.5
OpenAI
Code generationCS walkthroughsDebugging
11% 1.3 s
📚Gemini 3.1 Pro
Google
Long-context (whole textbooks)Deep Think · exam analysis
9% 1.6 s
Muse Spark
Meta
Multimodal reasoningHigh-volume FAQContemplating mode
4% 210 ms
🛡️
Multi-vendor by default — never a single point of failure. When one provider has an outage, the router auto-fails over to a backup. In Q1 2026 we maintained 99.97% inference uptime through three documented provider incidents.
🎓

Give every student
a teacher at 2 AM.

A 30-minute demo is enough to tell whether Lyceum fits your campus. We'll walk through three of the most common student help-requests at your institution — and show you exactly how Lyceum handles them.

Avg. response time < 4 hours · serving institutions across U.S., Canada, and Europe · 50+ partner campuses