Nemesis — Secure Learning Environment
Nemesis
Nemesis — Enemy to Ignorance Nemesis owl logo icon — owl in graduation cap on grey shield with gold border
NEMESIS
Enemy to Ignorance

A plain-English guide

Every school in America needs to protect its students online. Most can't afford to. Nemesis changes that.

Every piece of this already exists somewhere, sold by a different company, to schools that can afford it.

Network security

Cisco, Fortinet, and similar

Cost$3,000–8,000/yr
Self-hostedNo
Works at homeNo
Covers personal phonesNo
Who has itLarge districts only

Content filtering

Securly, GoGuardian, and similar

Cost$2,000–5,000/yr
Self-hostedNo
Student dataStored by vendor
Covers personal phonesNo
Who has itWell-funded schools

AI learning assistant

Khan Academy, Khanmigo, and similar

Cost$5,000–15,000/yr
Works with school's own curriculumNo
Adapts to each studentLimited
Who has itWealthy districts

Teacher lesson-planning tools

Chalkie, and similar

What it doesHelps teachers generate lesson plans and worksheets faster with AI
Who has it1M+ teachers, 100+ countries ($4M funded)
Device/network enforcementNone
Built-in curriculumNone — teacher still writes the content
On-premise optionNo

A teacher productivity tool, not a school platform — it doesn't compete with Nemesis so much as it could one day feed content into it.

Translation services

Various vendors

Cost$10,000+/yr
AutomaticNo — manual process
Who has itLarge urban districts

Emergency response

Raptor, Navigate360, and similar

Cost$3,000–8,000/yr
Integrated with the networkNo
Who has itWell-funded schools

Personal phone control, without confiscation

Partial: LockedIn (app, launched March 2026)

Requires an app on the phoneYes
Cost$5–20 / student / yr
Student dataStored by the startup
PrivacyGPS tracking, 24/7
ReviewsPoor
Network-level, no appNo — nobody does this yet

And the learning platforms, on top of all of that

These are the tools schools already buy to teach with. They are good at what they do. None of them touches security, and none of them is self-hosted.

Adaptive practice — math & language arts

IXL

Cost$24.50 / student / yr
A 500-student school pays$12,250 / yr — for IXL alone
Network securityNone
ESL reading supportNone
AI safety curriculumNone
Career guidanceNone
On-premise optionNo — data leaves the school

Strong drill-and-practice engine, used by millions of students. Price verified: Wichita Falls ISD board approval, 2026 — 9,200 students, $227,895 total.

Grammar & writing

NoRedInk

CostCustom quote — contact vendor
Network securityNone
ESL reading supportNone
AI safety curriculumNone
On-premise optionNo

Popular with ELA teachers, strong writing instruction. No published price — we are not estimating one.

ESL / newcomer language learning

Summit K12

CostCustom quote — ESL focused
Network securityNone
AI safety curriculumNone
On-premise optionNo

Strong ESL instruction for newcomers. No published price — we are not estimating one.

The Nemesis difference. A school using IXL, NoRedInk and Summit K12 is paying for three separate platforms, three separate logins, three separate data silos — and still has no network security solution.

Nemesis replaces all three education platforms and the network security product — in one installation, on the school's own hardware, with one teacher dashboard and one student login. At a price point well below what schools currently pay for these tools separately. Stated as the positioning claim it is: the AI-safety curriculum, translation and security are working today; the reading and full-subject curriculum that would replace an ELA or math platform outright are the planned phases in section 08, not shipped capability.

COMPETITOR pricing — what a school pays today for all of the above

$23,000 – $51,000 / year

These are competitors' figures, not Nemesis pricing. Nemesis pricing is in section 06.

  • 5–6 different vendors
  • Student data stored at every one of them
  • Protection stops working the moment a device leaves the building
  • None of it is self-hosted

One device, about the size of a small book. It plugs into the school's existing network. The school buys it once. They own it. They control it. Student data never leaves the building.

NETWORK $$ FILTERING $$ LEARNING $$ TRANSLATION $$ EMERGENCY $$ PHONES (partial) $$ 6 vendors, your data with each Nemesis Protected students Informed parents Focused teachers
✅ Working today 🔨 Being built now 📋 Planned, after funding

Network protection

✅ Working
Internet Nemesis Safe content reaches students Dangerous content blocked automatically

Blocks harmful websites automatically, updated daily with new threats. Works whether the student is at school, or at home on a school-owned laptop.

Firewall rule management

📋 Planned

IT directors get full visibility and control over every firewall rule on the Nemesis appliance, from the dashboard — no command line needed.

  • View. All active rules in a clean dashboard table, grouped by allowed/blocked, inbound/outbound, and protocol.
  • Create. A simple form for new rules — source, destination, port, protocol, action.
  • Manage. Enable, disable, or delete rules from the dashboard; changes take effect immediately.
Not yet built — the backend design work is only just starting as of this writing. The underlying firewall chokepoint it will sit on top of (firewall.py) is real and already enforcing rules today; this is a management UI for it, not a new enforcement mechanism.

AI learning assistant

✅ Working
QUIZ AI assistant works WITH the student — never hands over quiz answers

An AI helper that works with students on their lessons — not doing the work for them. Students must complete their lesson to earn access to it.

At a time when students are being told AI will eliminate their career options, Nemesis teaches them to be the people who build, secure, and direct AI systems. The students who understand AI will give instructions to AI. The students who don't will receive instructions from people who do. Every lesson connects to real IT careers that are growing because of AI, not despite it.

For students — focused learning time

✅ Working

Students are limited to educational tools during learning time — no social media, no distractions, no off-topic browsing while lessons are active.

Beyond dangerous content, unprotected school networks allow:

  • Social media during class (Instagram, TikTok, Snapchat)
  • Entertainment and gaming sites
  • AI tools that complete homework and essays for students — undermining learning and academic integrity
  • Any distraction a student can find on the open internet

Nemesis blocks all of it during learning time. Students stay on task. Teachers maintain classroom focus. Parents know their child is doing their own work. Academic integrity is enforced at the network level — not by asking students to comply, but by making non-compliance impossible.

Scope, stated because the rest of this document is specific about it: "at the network level" means traffic that crosses the school network or an enrolled school-owned device. A personal phone on its own cellular data is outside that path until the network-level phone control below is built — which is exactly why that item is on the roadmap.
Nemesis enforces academic integrity at the network level — not just in the classroom app. Social media, entertainment, and gaming sites are already blocked on enrolled devices during learning time — not by an honor code, but by the network itself. Named AI chat services (ChatGPT, Character.AI, Claude, Gemini, Copilot) are blocked the same way, today. The one AI a student can reach on an enrolled device is Nemesis's own lesson assistant — gated behind finishing the lesson, and built to work with a student rather than hand over answers. 🔨 Building, not yet shipped: catching an AI-misuse attempt and redirecting the student straight back into the curriculum, instead of just blocking the site. That's the homework-integrity system already on this document's own roadmap (see "Homework assistant" below) — real and scoped, not a working feature yet, and this document won't claim otherwise before it is.

For teachers — data-driven instructional reporting

🔨 Building

Data-driven instructional reporting built into every lesson — teachers see exactly what each student struggled with and what to teach next.

Quiz scoring is live today. The reporting layer that turns it into a teach-next recommendation is in build for the October demo.

Unlimited language translation

✅ Working
Lesson Español Tiếng Việt 中文 العربية Kreyòl Tagalog Français + any other

Nemesis translates into any language Claude knows — 200+ languages and dialects. Schools configure which languages they actively use.

A Vietnamese-speaking student gets the same lesson as an English-speaking student — automatically, at no extra cost. Adding a language is a configuration change, not a development project: there is no language pack to install and no per-language engineering work.

⛔ Current state, found 2026-09-19 — the architecture works; the stored catalog does not, right now. All 43 currently-stored translations (17 Spanish, 17 Vietnamese, 9 Arabic) fail the reader's own paragraph-alignment check, because a lesson-content rebuild (Sept 18-19) restructured every lesson after the translations were written (Sept 15). A student selecting a language today sees "This translation is out of date with the lesson and cannot be matched to the slide on screen" instead of the translated text — the safety check working correctly, not a bug in it. Not fixed overnight, deliberately: re-translating resets the 17 approved Spanish lessons to unreviewed status, discarding Angie's approval — a content decision for Paul and Angie, not an engineering one. Full detail: PUNCHLIST.md.

Device protection at home

✅ Working
School Home Anywhere else

When a school laptop goes home, the protection goes with it — same blocking, same monitoring. The protection lives on the device, not on the building.

Applies to enrolled, school-owned devices.

Component monitoring

✅ Working

Nemesis monitors every device for missing or compromised security components. ClamAV removed by a student. A security engine that goes absent after a Windows update. Component inventory that changes unexpectedly. Nemesis detects it on the next heartbeat and alerts the administrator immediately.

The administrator sees which device has the problem, exactly what is missing, and how long it has been missing. Detection is instant and certain — the administrator knows before the teacher does.

Remote recovery

📋 In development — roadmap

Automated remote recovery is on the active development roadmap. When complete, administrators will be able to trigger a full component reinstall from the dashboard, delivered silently to the device.

Expected: pre-release. Detection (above) is live today; remote recovery is designed, not yet built.

Parent visibility

🔨 Building

Parents will see what their child is doing on school devices — in their own language, on their phone.

Student social platform

🔨 Building

A school-safe version of social media. Students can only access it after completing their lessons.

Emergency response

🔨 Building

One button locks down the school. First responders scan a QR code for safe access.

Homework assistant

🔨 Building

Students get AI help with homework. The system watches for integrity concerns and notifies the teacher — without accusing the student.

Bring your own lesson

🔨 Coming Soon

Teachers will be able to upload their own lesson content directly into Nemesis. Upload a text file or document — Nemesis processes it automatically:

  • Key points marked with the Nemesis owl callout
  • Audio narration generated
  • Quiz questions created from the content
  • Translated into all 5 languages

The teacher reviews and approves everything before it goes live to students. The teacher writes the lesson. Nemesis handles the rest.

Not shipped yet — described here so it isn't confused with what students and teachers can use today.

Multi-platform agent coverage

One protocol, one dashboard, multiple platform implementations. The Windows agent solved every hard design problem — attestation, tamper detection, telemetry, dashboard integration; the remaining platforms are translation work against that protocol, not new design work.

  • Windows. ✅ Working Full agent — TPM-backed binary attestation, memory injection detection, tamper detection and alerting, persistent tray icon (the owl), heartbeat telemetry, dashboard integration. 1.0.14 build pending staging. Peer-to-peer enforcement: 🔨 in active development (WinDivert layer).
  • Linux. 🔨 Building Core components exist today (DNS enforcement, TPM attestation, memory scanning, behavioral monitoring deploy scripts, all confirmed present in the repo). Full feature parity with the Windows agent is not yet verified — an audit is in progress. Target: demo-ready October 6. Peer-to-peer enforcement: 📋 designed (privileged helper), not built.
  • Android. 🔨 Building Zero code in the repo as of this writing. Full build starting this week: enrollment + heartbeat (week 1), Play Integrity attestation (week 1, hardware-backed, cannot be spoofed), persistent notification (week 1), full tamper detection (week 2). The Windows agent protocol is the spec; this is a port, not a redesign. Peer-to-peer enforcement: 📋 designed (reporting only at this stage).
  • Chrome extension. 📋 Planned Zero code in the repo as of this writing. Specification complete; build starts once the Android build stabilizes. Target: October 6 demo. Deployment path: Google Admin Console. Peer-to-peer enforcement: 📋 designed (the appliance enforces for this platform, not the agent).
  • iOS. 📋 Planned Apple certification in progress. Expected Q1 2027.
"Working" is reserved for what the repo actually contains as of this writing, checked directly — not asserted from a build plan. Where a platform's timeline is stated, that is a plan, not a completed fact.

Personal phone control

📋 Planned
TODAY: APP-BASED (e.g. LockedIn) APP Battery drain GPS tracked 24/7 NEMESIS: NETWORK-LEVEL (planned) No app installed School WiFi applies rules

The infrastructure to control personal phones on school WiFi — without requiring any app installation — is designed and planned. Unlike current solutions that require an app on every student's phone, Nemesis will enforce school rules at the network level, automatically. This is in development.

Reading and language learning

📋 Planned

Nemesis includes a built-in reading teacher powered by phonics — the evidence-based approach at the center of the Science of Reading movement.

Students get pronunciation support, phonics instruction, and vocabulary building integrated directly into every lesson. The same AI that teaches AI safety teaches reading.

For older students and ESL learners: a song lyrics library where students can bring their own songs or choose from Creative Commons licensed music. The AI teaches phonics, vocabulary, and cultural context through music.

Students can read aloud and receive real-time phonics feedback. The AI hears what they say, identifies the specific error, teaches the rule, and asks them to try again.

A language learning module lets any student practice a new language using the same phonics engine, the same reading library, and the same song library — with AI feedback in the target language.

All of this runs on the school's own hardware. Student voice data is never stored. Audio is processed in real time and discarded.
NEMESIS VOICE. Every lesson is narrated by Nemesis the Teacher — a custom voice built exclusively for this platform.

Audio is generated once and cached permanently on the school's appliance. No cloud calls during student use. No per-school cost. No per-student cost. One generation serves an entire district.

The Nemesis voice runs on the school's own hardware. No third-party licensing. No vendor terms that can change. Owned outright.

Available today: English narration (native voice), plus Spanish, Arabic, Vietnamese, and Mandarin as an accented comprehension aid. Native-speaker recording for each non-English language is next — see the language roadmap below.
LANGUAGE ROADMAP. Nemesis speaks English today. English is the primary narration voice; other languages are provided as a comprehension aid alongside translated text, not yet full narration parity. The phoneme mapping approach we're building extends the Nemesis voice to any language without rebuilding the curriculum — the curriculum is written once, the voice is recorded once per language, and every school benefits immediately.

Shipped today (2026-09-20): Arabic, Vietnamese, and Mandarin narration is now available (accented) — ahead of the native-speaker-recording roadmap below. Vietnamese and Mandarin also carry a measured, genuinely flatter pitch profile consistent with a tonal-language voice, on top of the accent. "Tonal" here means the underlying voice model shows this real, measured signature — not that per-syllable tone is synthesized; that capability is built and tested but intentionally not yet connected, since the English lesson text carries no tone markers to key it off.

Cost to add a language: one native-speaker recording session, approximately $1,000 per language, serving every school on the platform permanently.

Native-speaker recording roadmap (replacing today's accented voice with a recorded native speaker, language by language): V1.0 English (native, complete) · V1.5 Spanish (native recording) · V2.0 Arabic + Vietnamese (native recording) · V3.0 Mandarin + French (native recording) · Beyond: any language, on demand.

"We're building the engine. We'll just need readers. Every language we add costs about $1,000 and serves thousands of students — forever."

Contrast with competitors: adding a language to a competitor platform means rebuilding the entire curriculum in that language — hundreds of thousands of dollars, years of development. Nemesis adds a language in one recording session. One curriculum. Every language. Any student. English is the only NATIVE-recording narration today. Spanish/Arabic/Vietnamese/Mandarin narration live now is accented (phoneme-mapped), not a native-speaker recording — the roadmap above is what closes that gap, one recording session per language.
READING LEVEL MEASUREMENT. Current implementation: Flesch-Kincaid Grade Level — an industry-standard formula, public domain, no licensing cost. It reports in grade-level terms every teacher and parent understands, plus a descriptive label (e.g. "Intermediate Reader") and an approximate Lexile reference band for context.

ROADMAP — Nemesis Reading Level (NRL), post-demo: a proprietary reading level metric owned outright by Nemesis, calibrated against Flesch-Kincaid and Lexile so districts can map it to whatever system they already use. No licensing fees. No dependency. No vendor terms that can change. Nemesis owns the metric — forever. The Flesch-Kincaid score is live today, on-demand per lesson in the teacher review tool. Automatic scoring for every lesson at publish time, and teacher/parent trend views over time, are next — not yet live. NRL itself is not built; it's the metric planned on top of this once FK is fully wired in.
SERVING EVERY LEARNER. Nemesis is built for the full range of learners in a modern classroom.

ESL students get a translation hover panel on every lesson screen — read the lesson in your home language alongside English. A student who speaks Spanish at home follows along in Spanish while learning to read in English. Nemesis bridges the gap. Now live: lesson and vocabulary audio narrated in Spanish, Arabic, Vietnamese, or Mandarin using the same consistent Nemesis voice character — accented quality today, with native-speaker recording for each language next; see the language roadmap above.

SPED students get audio narration so a student who struggles with reading can still access the curriculum, hints available on demand so a student who needs more support can request it without raising their hand, independent mode so they work at their own pace on their own timeline with no class-wide pressure, and a clean one-concept-per-screen design with no autoplay and no unexpected sounds.

IEP alignment: teachers assign specific lessons tied to IEP goals. Time on screen, quiz scores, and hint usage are tracked today and available as progress data.

Many ESL students also have IEPs. Nemesis serves both needs at once — home-language support and pacing/reading support — in one platform.

Nemesis doesn't have a separate ESL mode or a separate SPED mode. It serves every learner in the same classroom with the same platform — because that's how real classrooms work. Not yet built: reading-level-matched lesson assignment (Flesch-Kincaid isn't wired into lesson assignment yet), a teacher-facing view of in-progress student status with completion notifications, and home-language audio narration.

District management

📋 Planned

A district administrator sees all schools in their district from one dashboard.

This isn't an idea. It's a working system — and this is exactly what is and isn't built, checked against the actual project, not a wish list.

✅ Working today

  • Network security platform running (site blocking, threat detection, alert processing)
  • Agent heartbeat monitoring for enrolled school-owned devices
  • Over-the-air device updates — verified September 17, 2026
  • Device protection at home (enrolled school-owned devices)
  • AI learning assistant, active
  • Unlimited language translation, working (10 configured as shipped defaults)
  • 80+ lessons loaded, in review by the educator partner
  • Basic enrollment flow

🔨 Being built now (target: before the October demo)

  • Parent visibility portal
  • Student social platform
  • Emergency response system
  • Homework assistant with integrity monitoring
  • Bring your own lesson — teacher upload pipeline
  • Student learning profiles
  • Quiz generation (AI-assisted, reviewed by the educator partner)
  • Installing Nemesis on a school device is being built down to three steps: download the installer, double-click it, click Yes when Windows asks for permission. The installer is designed to handle everything else automatically — including installing any software it needs to work — so there's no IT training and no prerequisites to manage. Target: this week. Not yet verified end-to-end; today's install still needs a working zip file, not a single click.
  • Linux agent — core components exist, full parity with Windows not yet verified

📋 Planned, after funding

  • Personal phone control on school WiFi — network-level, no app required. This is the architecture, not yet built.
  • Full district management
  • Chrome browser extension
  • A research study with real schools
  • Collaboration with an educational psychologist

The live demo

Every device at the demo location connects to Nemesis running remotely via Tailscale — no local server, no special setup, the same architecture a real school deployment uses. The IT director, or investor, in the room watches in real time as their own devices appear in the dashboard; an unmanaged device attempting to reach a managed one gets blocked; a new device enrolls in about 30 seconds; the AI assistant responds to a real student question; and the cost tracker increments by a fraction of a cent for that call.

This is not a simulation. It is the product, running live, in the room.

Figures (30-second enrollment, per-call cost) are illustrative of a typical demo run, not independently re-measured for this document. As of this writing, this describes the demo as designed. The production dashboard has been running continuously since before today's firewall-rule and AI overage-pool work landed — a restart is required to bring that work online, and a full rehearsal against a freshly-restarted dashboard carrying today's changes has not yet happened.
Nothing above is overstated on purpose. If something looks unfinished, that's because it is — the plan is to be exactly this specific with the lawyer, too.

Today — COMPETITORS (average school)

$23,000 – $51,000 / year
5–6 different vendors
  • Student data stored at each vendor
  • Protection stops at the school door

Nemesis

$8,400–$21,000 one time, by school size
Everything included, one time$8,400–$21,000
  • Hardware, software, curriculum, translation, and narration — all included
  • Student data stays in the school
  • School owns it completely — no subscription, no renewal

Nemesis costs less than one year of what schools pay today for inferior, fragmented solutions — and there's no year two.

✅ Pricing decided, 2026-09-20 — a one-time purchase, not a subscription

Supersedes every earlier per-student and per-year figure in this document.

School sizePrice
Small (under 500 students)$8,400 — one time
Medium (500–1,500 students)$14,000 — one time
Large (1,500+ students)$21,000 — one time
District (5+ schools)Contact for custom pricing

Optional support plan

$1,200–$3,000/year, by school size. Covers platform updates, new curriculum additions, priority support, and the hardware replacement program.

Completely optional — the platform runs without it.

Nemesis is priced at roughly 70% of the average annual cost of comparable solutions — as a one-time purchase. The average school spends about $20,000 per year on the combination of tools Nemesis replaces (Section 02). For less than that single year's cost, Nemesis is owned outright — forever.

No subscription. No renewal fees. No vendor lock-in. No price increases. No access held hostage.

Year 1: you save 30%. Year 2: you save 100%. Year 3 onward: pure savings.

Cost comparison

Security tools$5,000–10,000/yr
Curriculum platform$12,000–25,000/yr
Translation$2,000–5,000/yr
TTS narration$1,000–3,000/yr
Competitors, total — yearly, forever$20,000–43,000/yr

Nemesis, once, forever: everything above for $8,400–21,000, one time.

Grant pathway

Nemesis works alongside schools to identify education-technology grants that cover the full purchase price. E-Rate, Title I, Title IV-A, and state technology grants frequently apply.

Many schools pay nothing out of pocket. We help you find out if yours qualifies.

⚠ Not independently modeled this session: whether these tiers cover Nemesis's own hardware and support cost at these volumes, or what margin they leave. A real unit-economics pass is still owed before this is quoted to an investor as a margin claim rather than a customer price.

Every school starts where it's comfortable and goes as deep as it chooses. No all-or-nothing commitment, no district-wide decision required to get started.

Level 1 — Security only

✅ Available now

Blocks harmful content automatically. Protects enrolled school-owned devices at school and at home. No curriculum changes, no teacher training.

Who decides: IT director. Time to deploy: one afternoon. Cost: hardware + base license.

Level 2 — AI Safety curriculum

✅ Available now

Pre-built lessons on AI safety, internet safety, and digital citizenship — already built, translatable into any language the school configures, already tested. Students complete lessons before accessing the AI assistant.

Who decides: principal or IT director. Time to enable: ~10 minutes. Additional cost: included.

Level 3 — Build your own curriculum

📋 Vision, not yet built

A teacher describes a lesson on any subject; the AI builds it; the teacher reviews and publishes it. No district approval needed, no commitment beyond that one lesson.

Who decides: individual teacher. Depends entirely on the Curriculum Builder (section 08 below), which does not exist yet.

Level 4 — Subject integration

📋 Vision, not yet built

A department or school builds curriculum for a full subject, aligned to state standards, reviewed by their own teachers, replacing one subscription at a time.

Who decides: department head or curriculum director. Same Curriculum Builder dependency as Level 3, one tier up.

Level 5 — Full platform

📋 Vision, not yet built

Curriculum, assessment, AI tutoring, parent portal, and translation in one self-hosted system — Nemesis as the school's primary education platform.

Who decides: superintendent and school board. This is the Phase 3/4 vision in section 08, not a current option.

No school has to commit to Level 5 to get the benefits of Level 1. A school that starts with security and never builds a single lesson still saves money and protects students today. Nemesis is a tool, not a mandate — the depth of integration is entirely the school's choice.

Nemesis today is a working security platform with one curriculum built on top of it. The company's ambition is bigger — stated plainly here, phase by phase, so it's never confused with what exists right now.

What schools pay today, across a full EdTech stack

Separate vendors, per category, per year

Reading/ELA platform$15,000–40,000
Math platform$15,000–40,000
Science curriculum$10,000–30,000
Social Studies$10,000–25,000
Foreign Language$10,000–30,000
LMS$20,000–50,000
Assessment platform$10,000–25,000
AI tutoring$10,000–30,000
Translation services$10,000–20,000
Content filtering$5,000–15,000
Network security$10,000–30,000

Total today: $125,000–285,000 per school, per year, forever. These figures are the founder's own working estimates, not independently sourced this session — worth a real citation before quoting them to a grant board. The vision: hardware ($400-600, one time) plus a license, replacing all of it — eventually. Only Level 1 and 2 of that stack (network security, one curriculum) exist today; everything else is the roadmap below.

The market reframe this implies: not a ~$3B school-security niche, but the ~$89.49B global K-12 EdTech market (2023 figure, ~13.4% annual growth, operator-supplied and not independently verified this session) — every school, every subject, every grade, if the full vision is realized.

The four phases

  • Phase 1 — current, demo-ready: AI Safety curriculum, unlimited language translation, network security. The only phase with shipped, verified capability.
  • Phase 2 — grant funding: Curriculum Builder for any subject, IT/CS curriculum expansion, a research study, first 10 school pilots. Zero code exists yet.
  • Phase 3 — Series A: Full LMS features, multi-subject curriculum library, a curriculum marketplace, district management, an all-50-state standards database. Not scoped or designed.
  • Phase 4 — scale: International standards, university/community-college curriculum, corporate training, government/military education. Not scoped at all.

Three-tier architecture — roadmap, post-demo

Nemesis is designed to scale to district deployments. Multi-school management is not built today — it's parked on the roadmap. Three deployment tiers share one codebase: a school buying Tier 2 today receives Tier 3 capability when it ships, at no additional license cost.

  • HOME (Tier 1): Parents connect to their child's school data from home via Tailscale. The data stays at the school. The home dashboard is a secure window — nothing is copied, nothing is stored remotely.
  • SATELLITE (Tier 2 — available today): One appliance per school. Complete security and education platform. Fully autonomous. Works with or without district connectivity.
  • PRIMARY (Tier 3): One district dashboard, receiving reports from each satellite school. Raw student data stays at each school — the primary sees aggregated reports only. District-wide policy override: push rules, curriculum standards, and emergency lockdowns to every school simultaneously. All communication over Tailscale VPN, encrypted end to end, always.

The Nemesis data volume is encrypted at rest.

Nemesis protects student data with hardware-backed encryption. Student work is reviewed by teachers, not algorithms or a technology company — the school owns the hardware, the school owns the data.

✅ AVAILABLE TODAY: TPM-backed volume encryption (LUKS2 + TPM2, live as of 2026-09-21).

Nemesis protects student data with hardware-backed encryption:
  • The Nemesis data volume is encrypted using LUKS2 with AES-256 encryption.
  • The encryption key is sealed to the appliance's TPM 2.0 chip — the same hardware that verifies every agent binary on the network. Physical disk removal cannot decrypt the data without the original TPM.
  • Normal restarts are designed to be fully automatic — no passphrase, no manual step.
  • A break-glass recovery key is stored separately for disaster scenarios.
  • Backup copies are GPG-encrypted before writing to removable media.

What this protects: disk theft or loss; appliance RMA or disposal; lost or stolen backup media.

Honest boundary: this covers the Nemesis data volume. System logs and configuration files are outside this boundary — documented honestly rather than obscured. ✅ Live as of 2026-09-21 — running on the production appliance today.

For districts with multiple schools: data travels between school appliances and the district dashboard exclusively over Tailscale VPN — encrypted end to end, never touching the public internet.

AI-assisted features

Hint comprehension, lesson translation, and the teaching assistant process content via Anthropic's API.

What is sent: lesson text and anonymized student responses.
What is NOT sent: student names, IDs, grades, or any identifying information.

API calls are governed by Anthropic's data processing agreement. Under current terms, API inputs are not used to train Anthropic's models. Schools should review Anthropic's API privacy policy for their own compliance requirements.

Transparency note. The Nemesis teaching assistant sends student questions to Anthropic's API verbatim. No student name, ID, or identifying information is included in the request. Schools should review this against their FERPA obligations and Anthropic's API data processing terms.

AI feature controls

AI features can be disabled globally today. Per-feature controls (Lesson Assistant, Hint Comprehension Check, Lesson Translation, Auto-Marker Placement, each independently) are in active development and will be available before release.

AI usage — always on for students

✅ Working

Nemesis uses a three-pool budget system so students are never refused a hint or lesson response because of a budget limit.

  • Base pool — 2,000 calls/hour, default. Every AI request served instantly.
  • Overage pool — 200 calls/hour, default. When the base pool is exhausted, AI keeps serving students from the overage pool. Students notice nothing; IT is alerted immediately.
  • Hard stop. Only when both pools are exhausted do students see: "Nemesis is taking a short rest. Try again in a few minutes!" IT gets an urgent alert and can add budget instantly from the dashboard.
At a 3,000-student school, expected peak demand is roughly 1,600 calls/hour — comfortably inside the 2,000/hour base pool default. Both pool sizes and every alert threshold are IT-configurable dashboard settings, not fixed in code.
The base/overage split, the settings and the three status states above are live and tested today. Alert checkboxes are real, persisted settings, but do not yet send email/SMS to IT — that delivery step is not wired yet, and this document says so rather than implying it.

AI usage intelligence

✅ Working

Nemesis tracks every overage event: which AI feature triggered it, time of day and day of week, how many calls went over the limit, and its estimated cost and duration.

Patterns are surfaced automatically from that history — peak usage hours and days, the most common trigger feature, average overage duration, and monthly overage cost — so IT directors can tune the base limit from real data instead of a guess.

Illustrative example of what that data supports: "Your peak AI demand is 11:30 AM–12:30 PM on weekdays. Consider raising the base limit to 2,500/hour during that window. Estimated extra cost: $0.43/day."
The underlying tracking — 30-day overage history, peak-hour detection, monthly cost totals — is live and tested. The recommendation shown above illustrates what that data makes possible; it is not yet auto-generated copy in the dashboard today.

Enterprise Readiness Roadmap

Nemesis is built to enterprise standards from day one. Here is our honest assessment of where we are and what we are actively building:

✅ Available today

  • Single-school deployment
  • Device management (design target: 2,500+ per school)
  • Role-based access control
  • Audit logging
  • OTA agent management
  • AI Safety curriculum
  • AI teaching assistant
  • Multi-language translation
  • Global AI spending cap
  • TPM-backed volume encryption (LUKS2 + TPM2, live as of 2026-09-21)

🔨 In active development

  • Per-feature AI controls
  • Translation approval workflow
  • Job engine (load balancing)
  • Per-feature spending controls
  • Production web server (nginx + gunicorn)

📋 Pre-release (before first sale)

  • FERPA compliance documentation
  • Multi-school data isolation
  • District RBAC scope
  • Database query optimization for 50,000+ devices

📋 District scale

  • Primary dashboard (Tier 3)
  • Cross-school reporting
  • District-wide policy override

Schools that deploy Nemesis today receive all future capabilities as they ship — at no additional license cost.

Device segmentation

✅ Live today

Unknown devices are automatically placed in a restricted network segment. They cannot reach school resources through the guaranteed enforcement path — that specific guarantee is what the Visitor Network Policy below adds. Full visitor policy controls are on the roadmap.

What's confirmed live today: a device without an enrollment tag is placed into a distinct, restrictive DHCP scope. What's on the roadmap: the linked firewall/routing enforcement that formally guarantees a segmented device can't reach internal resources — see Visitor Network Policy, below.

Network architecture roadmap

📋 Planned
  • Zero-trust network — four tiers. Every device is classified and gated automatically; nothing gets default access.
    • Tier 1 — Agent devices (Tailscale). Full Nemesis agent, verified identity, full platform access.
    • Tier 2 — Known non-agent devices. IT-registered by MAC address — IoT, printers, cameras. Device-type rule templates (PRINTER, CAMERA, CLASSROOM_AV, STAFF_PERSONAL). Cannot reach school resources or other devices.
    • Tier 3 — Unknown devices. No agent, not on the approved list — fully blocked by default. IT grants specific access; every grant is logged, with an expiry.
    • Tier 4 — Quarantine. Triggered by an alert or suspicious behavior. Complete isolation — zero internet, zero LAN — until an IT director manually releases the device.
    Each tier has its own rule set. Unknown devices get nothing until IT deliberately opens access. Implementation: post-demo.

This is a genuine differentiator, not a slogan: every competitor sends student data to their own servers. Nemesis doesn't — it's self-hosted, with hardware-backed encryption live today and granular AI controls in active development.

Building all of this in one product requires expertise across network security, artificial intelligence, education technology, and translation — all at the same time.

Large companies build one piece well and charge a lot for it. They have no incentive to combine everything into one affordable, self-hosted package.

Small developers rarely have the security expertise to build the network protection layer correctly.

Nemesis was built by someone with experience across all of these areas, motivated by one question: why should only wealthy schools be able to afford to protect their students?

We believe Nemesis improves outcomes. We need funding to prove it.

  1. Do students who earn AI access through curriculum completion use AI more responsibly than students with unrestricted access?
  2. Do students whose AI assistant adapts its pace and depth to individual performance show better outcomes than students with a one-size-fits-all assistant?
  3. Does consistent protection — at school and at home — produce better safety outcomes than protection that stops at the school door?
  4. Does network-level phone control reduce classroom disruption more than app-based approaches?When built — this is a study nobody can currently run, because no network-level solution exists yet.

These are answerable questions. No one has answered them yet, because no one has had a platform that could ask them. Until now.

P

Paul Lozelle — independent developer

Paul is an independent software developer based in Houston, Texas. He built Nemesis because he believes every school — regardless of budget — deserves the same protection as the wealthiest districts.

A

Angela Campbell-Lozelle — educator partner

Certified ELA and ESL Teacher and Instructional Coach, M.Ed., Texas A&M University 2026. She has tested and reviewed the curriculum platform throughout development, and her feedback has shaped every education feature in the system.

Nemesis is not venture-funded. It was built by one person, with one goal: make enterprise-level school protection accessible to every school in America.

The platform you are reading about was built by one person.

Paul Lozelle designed and architected every system in Nemesis — the security platform, the education center, the AI assistant, the data-driven teaching engine. Every product decision, every standard, every rejection of work that didn't meet the bar. That was Paul.

Every line of code was written by Claude AI. Not assisted by AI. Not reviewed by AI. Written by AI — under continuous human direction, judgment, and quality control. 15,000+ lines of Python, JavaScript, and SQL. A Windows agent with TPM-backed attestation. A quiz system with server-side scoring. 57 lessons across four grade bands. A unified AI assistant that works across every surface of the platform.

The curriculum was designed by Angela Campbell-Lozelle — a Certified ELA and ESL Teacher and Instructional Coach, M.Ed., Texas A&M University 2026. The research basis, the lesson structure, the age-appropriate content decisions for each grade band — that is her expertise. Claude AI built 57 lessons to her standard. Every lesson was reviewed against her design rules. Her judgment was the quality gate.

This is not a story about AI replacing human expertise. It is a demonstration of what happens when human expertise and AI capability work together correctly.

Nemesis teaches students that AI is a tool that amplifies human capability — not a replacement for human thinking, judgment, or creativity. Nemesis itself is the proof of that principle.

The platform you are evaluating was built exactly the way it teaches. The method and the message are the same thing.

That is intentional.

Students who complete the Nemesis curriculum read this same statement before their first lesson — in language appropriate for their age.

They begin their AI education already experiencing the principle they are about to learn.

Nemesis's current build sprint — one developer and three AI sessions working toward the October 6 demo — runs eleven days. (The platform itself started earlier than that; this is the sprint that carried it from mid-build to demo-ready.)

Every decision about what to build first was a prioritization decision. We built what students need to learn. We built what teachers need to teach. We built what IT directors need to trust the platform. We built what parents need to understand their child's safety.

We did not build district-level reporting, because there is no district yet. We did not build fleet-scale validation, because there is no fleet yet. We did not build full malware corpus validation, because there is no corpus budget yet.

Those are not omissions. They are correct sequencing: build what proves the concept, validate with real users, scale with real funding.

Everything that is here works. Everything that is not here has a plan, a timeline, and a funding requirement. That is not a weakness. That is how good software gets built.

Takes 2 minutes. Nothing here is stored on our end until you hit send in your own email — see the note below.

(Don't look back — just write what you remember.)
(optional)
(optional)
(optional — only if you'd like a follow-up)

Please answer the required questions (marked without "optional") before submitting.

This opens a pre-filled email in your own email app, addressed to Paul — nothing is sent until you hit send there. There's no server collecting responses automatically.

Want the full technical details and architecture documentation?
Request NDA-protected access.

Link copied