Resell, don’t just run
Unlimited end users, no per-seat tax. Turn agents into a revenue line for your own clients — a store assistant, client-facing BI, a branded support agent.
White-label agent platform · runs in your cloud Your agents. Your cloud.
Your agents. Your cloud.
Your models. Your brand.
Pengui is a turn-key platform to build, govern, and resell trustworthy AI agents — installed in your own cloud or on-prem, powered by the LLMs you already trust. Run it on your infrastructure, under your brand, and deliver branded agents to your own clients. Your data and the memory your agents build live in your own database — yours to keep.
- Deploy enterprise-grade agents in your own cloud or on-prem — your data never leaves your walls.
- Bring your own models through open protocols (MCP) — you control cost, residency, and model choice.
- White-label everything: your brand, your domain, your customers, unlimited end users.
- Memory-native and model-agnostic — answers you can trust, not just sentences that sound right.
Pengui · console tenant: acme-eu
Agents ▣ runs in your-cloud · eu-west-1
Client analytics agent gpt-class Live $0.012
Support copilot open-weight Live $0.004
Onboarding assistant self-hosted Draft —
your VPC Pengui your LLM
No data egress by default
A better way to ship agents — owned, governed, and ready to resell.
Agents + memory, on your infrastructure
No black boxes. The engine is yours — model, memory, and all.
Most tools rent you a chat window on top of someone else's model and someone else's analytics. Pengui inverts that. The agent runtime, the memory layer, and the conversational data interface are the product — model- and warehouse-agnostic — installed where your data already lives. Nothing leaves unless you say so.
Own-cloud AWS · GCP · Azure · your VPC
On-prem Your data center, air-gapped if needed
No egress Data & model traffic stay inside your walls
The difference
Delivery shells rent you AI. Pengui hands you the engine.
Dimension Bolt-on AI — delivery shell Pengui — value engine
Where it runs ✕Vendor's cloud; your data leaves ✓Your cloud or on-prem; data stays
The model ✕Locked to one provider ✓Bring your own, any provider, via MCP
The intelligence ✕An extra bolted onto someone else's BI ✓Agents, memory & data access are the platform
Memory ✕Stateless or opaque ✓Governed, inspectable, owned by you
Your customers ✕The vendor's, ultimately ✓Yours — white-label and resell
Turn-key deployment
Go live in your own cloud in hours, not months, with a guided installer. Skip the platform-building; start shipping agents.
Multi-tenant from day one
Isolated, branded environments for every client or business unit, from one console.
White-label, end to end
Completely yours, end to end.
Apply your brand to the console, the agents, and the chat experience. Your colors, your logo, your domain — deployed on your URL in minutes. Co-brand per client when you want to.
Pick a preset — the live console preview re-skins from one token group.
Pengui · console tenant: acme-eu
Agents ▣ runs in your-cloud · eu-west-1
Client analytics agent gpt-class Live $0.012
Support copilot open-weight Live $0.004
Onboarding assistant self-hosted Draft —
your VPC Pengui your LLM
How it works
From zero to a branded agent, in four steps.
- 01
Deploy
Install Pengui into your own cloud or data center with the guided installer.
- 02
Connect
Wire in your models and your data through MCP. Your keys, your endpoints, your residency.
- 03
Build & govern
Compose agents, give them memory and skills, and watch every action from the console.
- 04
Brand & resell
Put your name on it and deliver to your own clients as branded, multi-tenant experiences.
Capabilities
What's inside the engine.
Outcome first, architecture second. Each piece is separable — adopt one, adopt all.
See and govern every agent from one place.
Build, configure, observe, and control all your agents and tenants in a single pane — who did what, with which data, at what cost.
Agents that plan, act, and use your tools — reliably.
Compose multi-step agents that reason, call tools in parallel, and follow guardrails, instead of brittle single-prompt bots.
Agents that remember what your organization means.
A governed, inspectable memory layer so your agents learn your terms, your context, and your history over time — and get sharper with use.
This is the moat: the trusted context beneath the chat box.
Plain-language questions, answers you can defend.
Let business users query their own data in plain language, with validation and governance so the answer is right — not just fluent. Model-agnostic; uses the LLM you bring.
The right context, every time.
High-quality, multi-stage retrieval feeds your agents the relevant facts, so responses are grounded instead of guessed.
Give agents new powers — safely.
Equip agents with packaged, permissioned skills and connect any tool or system through MCP, with governance over exactly what each agent can touch.
Connect anything you run.
Wire Pengui to your warehouses, SaaS tools, and internal systems — off-the-shelf or custom-built (by you or by us; see Professional Services).
Your model choice, as a first-class feature.
Point Pengui at any model through MCP — commercial, open-weight, or self-hosted — and switch whenever cost, residency, or quality says so.
A reference agent experience, ready to brand.
Our opinionated, cost-efficient chat product shows how a good agent should feel — and it’s fully white-labelable as the front door for your clients.
Mission
Give every organization the power to build, govern, and own trustworthy AI agents — in their own environment, on their own terms.
Adopting AI agents today usually means renting someone else's chat box, bolted onto someone else's model and someone else's data warehouse — with your information leaving your walls and your costs tied to a vendor's meter. That isn't ownership. It's dependency with a friendlier interface.
Pengui exists to invert that. We put the agent engine — orchestration, memory, retrieval, governed skills, and conversational access to your own data — inside your cloud or data center, wired to the models you already trust. We measure success not by how dependent you become on us, but by how completely your agents are yours.
Vision
A world where deploying a trustworthy AI agent is as ordinary — and as sovereign — as spinning up a server.
The last platform shift gave organizations control over their compute. The agent shift, so far, has taken control away: intelligence locked behind APIs, context trapped inside vendors, memory you can't inspect, answers you can't trust.
In that world, the value of an agent isn't the cleverness of one answer — it's the accumulated, governed memory of what an organization knows and means, owned by that organization. The chat box is a commodity; the trusted context beneath it is the asset. Pengui is building the platform for that world.
Values
What we hold true.
01
Sovereignty by default
Your cloud, your models, your data, your brand. We treat any default that moves data or control out of your hands as a bug to fix, not a tier to upsell. Deployment in your own environment isn’t the premium plan — it’s the premise.
02
Trust over fluency
A confident wrong answer is worse than no answer. We build for governed, validated, repeatable responses — memory with provenance, queries with validation, agents you can audit — because in work that matters, being right twice beats sounding good once.
03
Model-agnostic, on principle
Models keep changing; your platform shouldn’t have to. We connect to the LLMs you choose through open protocols, so model choice stays a decision you make for cost, residency, and quality — never a lock-in we impose.
04
Composable, not monolithic
Agents, memory, retrieval, skills, and data access are separable parts. Adopt one, adopt all, extend any of them without asking permission. Open interfaces over closed suites — build on us the way you’d build on a good library.
05
Outcomes, not architecture
Nobody buys orchestration; they buy answers, hours saved, a new revenue line. We describe what the software does in the language of the people who use it and keep the cleverness under the hood, where it belongs.
06
Yours, not locked in
Your data and the memory your agents accumulate live in your own database, in your own cloud — readable, exportable, and portable, with or without us. What you license is the engine that makes that data intelligent; what you own is the data and the context it builds. No vendor holding your information hostage, because it was never on our side to hold.
Yours to control
Enterprise-grade, deployed inside your perimeter.
Pengui runs where your data already lives. Bring your own identity provider, keep your own keys, and govern every agent action.
- Own-cloud or on-prem deployment — no data egress by default.
- Bring your identity: SSO, SAML, OIDC — Okta, Entra, Auth0 and more through one provider seam.
- Role-based access, full audit trails, and governed skills that constrain what agents can do.
- Data residency you control, because the whole platform sits in your environment.
What’s yours, what’s licensed
Your data is yours. The engine is licensed while you subscribe.
A clear line, because deploying in your own cloud makes it matter:
Yours, in your own database
Your data, your agents’ configurations, and the memory your agents accumulate — readable, exportable, and portable at any time, with or without an active subscription.
Licensed while you subscribe
The Pengui engine — the orchestration logic, connectors, and protocols that turn that data into intelligent agents. Your right to run the engine continues while your subscription does.
Model-agnostic by design
Works with your stack — not the other way around.
Pengui is built on open protocols (MCP) so it connects to the models, warehouses, and tools you already run. Swap a model without re-platforming. No vendor lock, no forced migration, no waiting on a roadmap.
Any LLM via MCPYour warehouseYour identity providerYour cloudOpen-weight or hostedYour SaaS tools
Capabilities, not endorsements — bring the providers you already trust.
What you can buy
Three ways to work with Pengui.
Not three businesses — one platform as the spine, with white-label and services in orbit.
Deploy and build, in your own cloud.
License the platform on a tenant/capacity basis, install it in your environment, and start building agents. This is the core product.
Turn agents into your own product line.
Brand the platform as yours and deliver multi-tenant agents to your clients. Unlimited end users, no per-seat tax — a revenue line, not a cost center. Partner program for resellers.
Need agents, MCP servers, or integrations built? We’ll do it.
Our team designs and ships custom agents, MCP servers, apps, and integrations on the platform — for your first deployment or your hardest problems. Scoped engagements, not an open-ended retainer.
Pricing
Simple, capacity-based, resale-friendly.
Priced by tenant and capacity with unlimited end users, so resale economics work. You're buying the engine, not a portal.
Studio
$399 /mo
per tenant · billed monthly Teams getting started
1 tenant · Console · agent orchestration · BYO-LLM · deploy in your cloud · community support
Deploy free →Growth
$899 /mo
per tenant · billed monthly Teams scaling agents
Everything in Studio + 3 tenants · memory-as-a-service · ask-your-data (NLQ) · multi-IdP SSO · standard support
Deploy free →Scale
$1,900 /mo
per tenant · billed monthly Resellers & providers
Everything in Growth + 20 tenants · white-label & resale · skill packs · usage analytics · premium support
Deploy free →Enterprise+
$3,500+
custom · talk to us Regulated & on-prem
Everything in Scale + 50 tenants · on-prem option · dedicated environment · SLAs · services credits · elite support
Contact salesAdd-ons · à la carte
Additional tenant · additional console seat · premium support upgrade · dedicated infrastructure · extra SSO providers · custom MCP server build · usage above included capacity (metered on agent runs / memory volume). Consumption is metered on top of the plan, on our substrate — not per end user.
Figures are illustrative for this preview — the pricing structure (capacity + consumption, unlimited users) is the decision; exact numbers live in one data file.
Built to carry your name — these slots fill in as you grow.
Your logo hereYour logo hereYour logo hereYour logo here
Customer stories
Customer stories coming soon. We'd rather show you a deployment than a stock quote.
Where does Pengui run?
In your own cloud or data center. The platform installs into your environment; your data and your model traffic stay inside your perimeter unless you choose otherwise.
Which AI models can I use?
Any model you connect through MCP — commercial, open-weight, or self-hosted. You control the choice and can switch at any time for cost, residency, or quality.
What does “white-label” actually mean here?
Your brand, your domain, your customers. You can deliver Pengui-built agents to your own clients as if they were your product — with unlimited end users and no per-seat fee.
How do you keep answers trustworthy if I bring a weaker model?
Reliability comes from the platform, not just the model: governed memory, validated data queries, retrieval, and guardrails buffer model quality. Stronger models help, but the substrate is what makes answers defensible.
If we stop subscribing, do we lose our data?
No. Your data and your agents’ memory live in your own database, in your own cloud — you can read and export them at any time. A subscription licenses the engine that makes that data intelligent; it never holds the data itself. (Specific post-termination terms are set out in the agreement.)
Can you build the agents for us?
Yes. Professional Services can design and ship custom agents, MCP servers, and integrations — a fast path to your first deployment.
How long until we’re live?
The guided installer brings the platform up in your cloud in hours. First useful agent in days, not months.
Is there a free trial?
Yes — deploy in your own environment and build free for 30 days. No credit card, no data egress.
Deploy free · 30 days
Deploy agents you actually own.
Stand up Pengui in your own cloud and build your first agent free for 30 days. No data leaves your environment.