// APPLIED INTELLIGENCE — LOCAL FIRST
Two products. One platform. Zero data leaves the building.
01 // OUR CORE CONCEPTS
Built-in indicators show exactly how much you save versus cloud-hosted solutions — live, on every query. The dashboard tracks your spend against per-seat AI subscriptions and metered API pricing, so finance sees the return without asking engineering.
MONTHLY AI SPEND · 40-SEAT ORG
▼ 0% AVG. SPEND REDUCTION — $19.3K BACK PER YEAR
Pulse watches how your AI is used and auto-generates agents for the tasks your team repeats every day. Every recommendation ships with the evidence behind it — accept one and observation becomes automation in a single step.
A TYPICAL WEEK ON PULSE
Loki-1 turns documentation, tickets and runbooks into consistent answers — onboarding without the overhead. New hires stop shoulder-tapping seniors and start pulling verified answers from a single source of truth on day one.
ONBOARDING WITH LOKI-1
FIGURES SHOWN ARE AN ILLUSTRATIVE EXAMPLE — YOUR DASHBOARD COMPUTES THESE LIVE FROM YOUR OWN USAGE AND SEAT COUNT.
02 // PRODUCTS
Greyspan.ai combines Loki-1, which answers questions, and Pulse, an Agent Discovery Engine that shows you what to automate next. Together, they turn everyday work into a continuous loop of answers and automation.
LOKI-1 — LOCAL ONLY KNOWLEDGE INFRASTRUCTURE
Our Local Only Knowledge Infrastructure learns from your documentation, tickets, and runbooks so teams can get consistent answers without exposing sensitive data. Empower your organization with a single source of truth that instantly synthesizes complex documentation into verified, actionable answers.
you ▸ where does our data go?
Nowhere. Every prompt is processed on this machine.
Network egress: 0 bytes. That's the whole point. ▊
PULSE — AGENT DISCOVERY ENGINE
An observability and recommendation layer that connects to however your organization already runs AI. It watches how your AI is actually used, surfaces the patterns worth automating, and turns them into purpose-built agents — without your prompts ever having to leave your environment.
Every recommendation is backed by the evidence behind it. Accept one and hand it off in a single step to create a new agent — turning observation directly into automation, not just another dashboard.
03 // THE PLATFORM
Fully on-premises against local models (e.g. Ollama). Sensitive prompts never leave the building.
Connects to your self-hosted chat interface, such as OpenWebUI or LibreChat.
Connects to a hosted model through an API when speed and quality matter most.
Greyspan gives your organization two products that work together: one answers your questions, the other learns what you keep asking and helps you automate it — all while your data stays under your control. The platform keeps getting more useful the more you use it.
YOUR TEAM USES → PULSE OBSERVES → AGENTS GET CREATED → THE PLATFORM GETS SMARTER ⟳
04 // RECOGNITION
Greyspan is a member of NVIDIA Inception, a program that nurtures startups building in AI and data science with access to NVIDIA's technology, technical expertise, and go-to-market support.
For us, that means building our local-first platform — Loki-1 and Pulse — on a foundation engineered for high-performance, privacy-first AI infrastructure.
© 2025 NVIDIA, the NVIDIA logo, and NVIDIA Inception are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
MEMBERS ONLY
Sign in with your Microsoft account to reach your license, downloads, guides & FAQ.
Verified by Microsoft. No passwords are stored here.
MEMBER PORTAL
you@company.com
LICENSE KEY
GREYSPAN.AI
••••-••••-••••-••••
Your key is unique to this account, activation-limited, and revocable. Keep it private — sharing it burns your own activations.
Local Only Knowledge Infrastructure. Includes Hamr companion.
macOS: M-series only · 907 MB | Windows x64 · 596 MB | signed, time-limited link
Agent Discovery Engine. Connects to your existing AI stack; runs on-prem.
SHA-256 · e9a1…puls · 64 MB · secure link
Prerequisites: Node.js 20+ and Ollama with at least one model pulled (e.g. ollama pull llama3.1).
1. Launch the app — double-click loki-1.command (macOS) or loki-1.bat (Windows), or run npm run dev.
2. Open /launch and choose Host a server (this machine holds everything) or Connect to a host (join one on your network — clients are thin).
3. Create your bootstrap admin account.
4. The first-run wizard scans your local Ollama and lets you pick a model from a dropdown. That's it — you're answering in about 15 minutes.
Data lives under data/app-data/ on the host, encrypted at rest with a key held in your OS keychain.
Import knowledge from a local folder, an SMB share, Google Drive, or SharePoint / OneDrive. Supported formats: PDF, Word, Excel, HTML, CSV, EPUB, and images — with OCR for scanned PDFs and images.
Define roles in the Host Console. Each role has its own knowledge scope, persona, and starter prompts, plus optional sub-profiles for narrower contexts. Retrieval is hybrid keyword + vector search with cross-encoder re-ranking, and every answer is grounded in cited source passages you can open in the in-app viewer.
Improve it over time: thumbs up/down on answers actively re-weights the vector database, so retrieval gets sharper the more your team uses it.
Local-only by default — answers run against your local model and documents never leave the host. Cloud APIs (OpenAI-compatible / Anthropic) are opt-in only, gated behind an explicit "this turns off Local Only" disclaimer. A network-isolation mode is available.
Accounts: a bootstrap admin, then invites, accounts, and groups — with TOTP two-factor authentication. The Host Console manages users, roles, profiles, the model runtime, knowledge, analytics, and voice. Per-session cost analytics show your local-vs-cloud savings.
Pulse runs on-prem as a Python service or a native macOS app.
From source:
python -m venv .venv
.venv/bin/pip install -e ".[dev]"
For the fully-local path, install Ollama and pull the models:
ollama pull nomic-embed-text (embeddings)
ollama pull gemma4:e4b (agent drafting)
Prefer a real app? The macOS .dmg gives you a native windowed app; on first launch use System Settings → Privacy & Security → Open Anyway (it's ad-hoc signed). The first launch also creates your single super-admin account, then the signup path locks.
1. Connect a source. Pulse captures usage through transparent proxies — your app talks to Pulse, Pulse forwards upstream and streams the real answer straight back, then keeps a copy. Connectors: proxy (any OpenAI-compatible API), anthropic (Claude), ollama (fully local), plus imports from the Ollama desktop app and OpenWebUI.
2. Cluster & review. Hit Re-analyze — Pulse embeds prompts locally and groups them into patterns, classified as recurring questions, repetitive tasks, or coverage gaps, then tagged by department and ranked by volume in the recommendations queue.
3. Accept → draft → self-test. Accepting a recommendation drafts a complete agent definition on your local model, then the model self-tests it (scored 1–5) so a weak draft is flagged before deploy.
4. Deploy & measure. Deploy, pause, edit, or test-chat the agent — it answers on your local model. Pulse then measures whether question volume actually dropped after it went live.
Hamr is the companion built into Loki-1 — part assistant, part mascot. He lives in the app and hangs around this website too. Rumor has it he hides things here… (the FAQ below is a good place to look.)
No. Loki-1 is local-only by design — answers run against a local model via Ollama, and your prompts, documents, and indexes never leave the host machine. Cloud APIs are opt-in only, gated behind an explicit "this turns off Local Only" disclaimer. Network egress: 0 bytes by default.
Anything you can run through Ollama (e.g. Llama 3.1). The first-run wizard scans your local Ollama and lets you pick a model from a dropdown. Optional cloud providers (OpenAI-compatible / Anthropic) are available behind the Local-Only disclaimer.
PDF, Word, Excel, HTML, CSV, EPUB, and images — with OCR for scanned PDFs and images. You can import knowledge from a local folder, an SMB share, Google Drive, or SharePoint / OneDrive. Retrieval is hybrid keyword + vector search with cross-encoder re-ranking, and answers cite their source passages.
Yes. Launch the app and choose Host a server to run it on one machine, or Connect to a host to join one over your LAN/VPN — the host holds everything and clients are thin. Accounts, invites, groups, and TOTP two-factor auth are built in, managed from the Host Console.
Yes — local speech-to-text and text-to-speech (Kokoro), with a distinct voice per role, all running on-device.
Pulse sits in front of your AI as a transparent proxy: your app talks to Pulse, it forwards upstream and streams the real answer straight back, then keeps a copy. Capture, embeddings, clustering, and drafting all run on-prem. PII and secrets are redacted at capture, identities are hashed, and Pulse works on aggregate patterns — never individual profiling. Workflow capture detects repeated desktop tasks from semantic events only — never keystrokes or screenshots.
It classifies clustered usage into three opportunity types: recurring questions (→ a dedicated answering agent), repetitive tasks (→ a task-running agent), and coverage gaps where the AI showed uncertainty (→ improve your knowledge base). Each recommendation is drafted on your local model, self-tested (scored 1–5) before deploy, and its impact measured afterward — e.g. "▼ 67% fewer matching queries since the agent went live."
Hamr is the companion built into Loki-1 — a small shapeshifting spirit of liquid glass-light, bound to one machine. He grows on what you feed your local store, and he can wear other shapes: a fox, a stone, a puddle, once a raven. His eyes and his heart-light never change. (In Old Norse a hamr is the shape a shapeshifter wears — which is the sort of company Loki keeps.) Rumor has it he hides things around this website.
MEMBER PORTAL · PERSONAL
you@email.com
Personal editions install straight from your platform's app store — no license key to manage. Same local-first privacy as our enterprise builds; your data stays on your device.
Install from your platform's store above, open the app, and follow the first-run prompts. On desktop, Loki-1 Personal runs against a local model — the app will help you install one on first launch. Everything runs on your machine.
Personal editions are local-first, just like our business builds: your documents and chats stay on your device. Nothing is uploaded unless you explicitly turn on an optional cloud model.
Hamr is the companion built into the app — part assistant, part mascot. He lives on your device and grows on what you feed your local store.
Personal editions install from public app stores with no license key and are built for individuals. Business is enterprise-licensed — per-seat keys, team accounts, secure private downloads, and admin controls. You can upgrade to Business anytime from Contact.
Use the store for your platform above — the Apple App Store (Mac & iOS), Google Play (Android), or the Microsoft Store (Windows). Updates arrive automatically through the store.
Yes. It's local-first by design — your data stays on your device, and optional cloud models are opt-in only.
Reach us at hello@greyspan.ai — we respond within 1–3 business days.
Yes — get in touch and we'll help you migrate from Personal to a licensed Business deployment.
04 // MEET THE FOUNDERS
We build privacy-first AI agents that keep your data local and your operations resilient.
CO-FOUNDER · CEO
Brayden oversees engineering and architecture at Greyspan AI, with a focus on secure, scalable systems. He leads the design of Greyspan’s applied intelligence platform, ensuring agents integrate cleanly with tools like M365, Google Workspace, and AWS. Previously, he built infrastructure and automation stacks for high-growth startups and MSPs. When he’s not deep in code, Brayden enjoys "VR" trail running and experimenting with new hardware.
// CERTIFICATIONS
CO-FOUNDER · CTO
Jeff leads technical experience and partnerships at Greyspan AI, bridging the technical and business sides of the company. He focuses on building reliable automation agents for IT teams and MSPs, and is driving Greyspan's push into privacy-first infrastructure tooling. Before Greyspan, Jeff worked in systems engineering and space robotics, bringing a hands-on technical foundation to how Greyspan builds and deploys its agents. Outside of work, he's usually exploring new cities or tinkering with side projects.
// CERTIFICATIONS
PROJECT MANAGER
Brandon leads the engineering behind Greyspan's most demanding deployments. He designs AI infrastructure for enterprises and MSPs that has to stay secure and online around the clock, and he's the reason our autonomous agents hold up under real-world pressure.
DEVOPS ENGINEER
Mark builds the pipelines that keep Greyspan's AI running. He automates deployment and infrastructure so our agents stay reliable at scale, with NVIDIA AI certifications behind the work.
SOCIAL MEDIA MANAGER
Chris owns Greyspan's presence across social and is the main creative force behind its design. He turns complex AI topics into content that lands and shapes the brand's whole look and feel — from the website to the product's visual identity — making sure everything Greyspan puts out is clear, polished, and unmistakably its own.
NETWORK ENGINEER
Nick oversees Greyspan deployments for IT teams and MSPs, partnering with clients from first pilot to enterprise-wide rollout. AIIO-certified through NVIDIA, he makes adoption seamless and every implementation reliable at scale.
05 // CONTACT US
Tell us where the manual work is piling up. Whether it's IT teams answering the same questions over and over, support tickets stacking up, or onboarding that takes too long, Greyspan is built to turn that repetitive overhead into automation that actually works — and we'd like to show you how.
Our platform pairs two products that get smarter the more your team uses them. Loki-1 gives your organization a single source of truth, synthesizing your documentation, tickets, and runbooks into consistent, verified answers. Pulse, our Agent Discovery Engine, watches how that AI is used, surfaces the patterns worth automating, and turns them into purpose-built agents — so your roadmap is driven by real usage, not guesswork. And all of it runs on infrastructure you already own, with your data never leaving your environment.
Let's map out what that could look like for your organization. Reach out to our team and we'll walk you through how Greyspan can automate your day-to-day operations — making them faster, more resilient, and fully under your control, from first pilot to scaled deployment.
06 // CAREERS
We are building the next generation of autonomous intelligence. Express your interest below and let's transform the industry together.
We're not filling a specific role at the moment, but we're always on the lookout for exceptional people in applied AI and automation. Greyspan is a small, high-caliber team building privacy-first AI infrastructure that organizations actually trust — and the people who thrive here tend to be the ones who don't wait for a job posting to reach out.
If that sounds like you, we'd love to know you exist. Share your LinkedIn, tell us what you do best, and we'll keep you front of mind as we grow.
LEGAL
EFFECTIVE DATE: JUNE 3, 2026 · LAST UPDATED: JUNE 3, 2026
This Privacy Policy explains how greyspan.ai ("Greyspan," "we," "us," or "our") collects, uses, and protects information when you visit greyspan.ai (the "Site") or use our products and services (the "Services"). By using the Site or Services, you agree to the practices described in this policy.
Privacy is core to how Greyspan is built. Our products are designed to run inside your own environment, and we collect as little personal information as possible to operate our business and provide our Services. We do not sell your personal information. This is described in more detail in Section 8.
We collect information in the following ways:
Some Greyspan products may support optional or customer-configured integrations with third-party platforms, including cloud AI providers, identity providers, ticketing systems, documentation platforms, telephony providers, and other business systems. When those integrations are enabled, data may be transmitted to those third-party services as necessary to perform the requested function and subject to the applicable provider's terms and privacy policy.
We may collect limited account, billing, implementation, diagnostic, logging, and support information necessary to provide, maintain, troubleshoot, secure, and improve the Services. We do not knowingly collect more information than is necessary for the purposes described in this policy.
We use the information we collect to:
If you are located in the European Economic Area or the United Kingdom, we process your personal data on the following legal bases: your consent; the performance of a contract with you; our legitimate interests in operating and improving our business, where those interests are not overridden by your rights; and compliance with legal obligations.
We do not sell your personal information. We share it only in the following limited circumstances:
We may use third-party tools and service providers to operate the Site and Services, including website hosting and management tools, analytics tools, email and communication providers, customer support tools, billing and payment processors, cloud infrastructure providers, AI/model providers, telephony providers, speech-to-text providers, text-to-speech providers, identity providers, ticketing systems, documentation platforms, and other integrations requested or enabled by customers.
Examples of third-party services that may be used by Greyspan or integrated with Greyspan products include, as applicable, Google Workspace, Microsoft 365 / Entra ID, AWS, Anthropic Claude, OpenAI, Google Gemini, Twilio, Deepgram, Cartesia, ElevenLabs, Vapi, LiveKit, ConnectWise, HaloPSA, Datto Autotask, Kaseya, ServiceNow, IT Glue, SharePoint, Confluence, Google Drive, and payment, tax, analytics, hosting, or email providers used to operate our business. These providers may collect information as described in their own privacy policies. We encourage you to review them.
We use cookies and similar technologies to operate the Site, remember your preferences, secure the Site, and analyze traffic. You can control cookies through your browser settings; disabling them may affect some functionality. If the Site uses a cookie consent banner or cookie preference tool, you may use that tool to manage your cookie preferences where available.
Greyspan does not sell your personal information, and we do not share it for cross-context behavioral advertising. We have not sold personal information in the preceding twelve (12) months and have no plans to do so. If this ever changes, we will update this policy and provide any rights and notices required by applicable law before doing so.
Greyspan does not use customer data, customer documents, customer prompts, or customer operational data to train shared or public AI models unless expressly agreed in writing.
Depending on where you live, you may have rights regarding your personal information, including the right to:
To exercise any of these rights, contact us at legal@greyspan.ai. We will respond within the timeframe required by applicable law. We may need to verify your identity before fulfilling your request.
We retain personal information only as long as necessary to fulfill the purposes described in this policy, comply with our legal obligations, resolve disputes, and enforce our agreements. When information is no longer needed, we delete or anonymize it.
We use reasonable administrative, technical, and organizational measures designed to protect your information, including encryption in transit where appropriate, access controls, role-based access controls, audit logging, tenant isolation, credential protection, rate limiting, and other security measures appropriate to the nature of the Services. No method of transmission or storage is completely secure, however, and we cannot guarantee absolute security.
If you access the Site or Services from outside the United States, your information may be processed in the United States, including North America. Where required, we use appropriate safeguards, such as Standard Contractual Clauses, for international transfers.
The Site and Services are not directed to children under 13, and we do not knowingly collect personal information from them. If you believe a child has provided us information, please contact us and we will delete it.
We may update this Privacy Policy from time to time. When we do, we will revise the "Last updated" date above and, where appropriate, provide additional notice. Your continued use of the Site or Services after changes take effect constitutes acceptance of the updated policy.
If you have questions about this Privacy Policy or our data practices, contact us at:
GREYSPAN.AI LLC
110 MORAINE POINTE PLAZA #1001
BUTLER, PA 16001
USA
EMAIL: LEGAL@GREYSPAN.AI
Patterns worth automating, grouped by department
🔒 Admin-only · patterns are aggregated across interactions — no individual is profiled
LIVE CAPTURE updates live
Create a dedicated agent to answer this automatically.
✓ Agent deployed — self-test 4.6/5 · ▼ 62% fewer matching queries since go-live
0/10 FOUND
FIND ALL TEN TO OPEN HAMR'S HOARD.
You found all ten secrets. Hamr is genuinely impressed — and he doesn't impress easily.
EASTER EGG CODE · HAMSKIPTI-2026
(Prototype — wire this to a real reward later: swag, license discount, hall of fame…)