AI-native applications · Managed Azure + Cloudflare

Castle
your cloud.

Red Rook builds AI-native applications and platforms, then runs them on your own Azure and Cloudflare. Your repository, Slack channels and cloud environment are live the day you start. You watch every step, through to tested, secured, managed software.

One fixed monthly price. Every $2,000 a month includes $100,000 a year of development credit.

credit-ledger
Opening credit · 2 Rooks × 2 yrs$400,000.00
#142Tenant row-level securityAI24.0h−$6,600.00
#143Partner API + keysAI18.0h−$4,950.00
#144Review and releaseENG2.0h−$550.00
#145Right-size SQL tierOPSincl.$0.00
Remaining credit$387,900.00
rook › next task
$100kcredit a year for every $2k a month
$275per engineer-hour, human or AI
24/7AI operations on Azure and Cloudflare
100%of the code and tenant are yours

What you’ll notice

Outcomes, not features.

You don’t buy hours and hope. You get a way of working where every decision, change and dollar is visible to you as it happens.

  1. 01

    You see everything, as it happens.

    Every story, estimate, decision and pull request lives in your own GitHub project: the whole agile lifecycle, documented in real time. It’s yours, not ours.

  2. 02

    You hear about every release.

    The moment a deployment finishes, your Slack channel gets a note: what changed, how it was tested and the hours it used.

  3. 03

    Your feedback becomes a decision the same day.

    Say it in Slack. It’s captured as an issue, answered with a design decision and recorded, so nobody has to argue it twice.

  4. 04

    You always know what you’ve spent.

    Every task is estimated and approved before work starts, and posts to a live credit ledger when it ships. No end-of-month surprises.

  5. 05

    You’re never locked in.

    Your tenant, your repositories, your accounts, your documentation. Leave whenever you like and take all of it with you.

  6. 06

    Your auditor finds the work already done.

    The way we work produces SOC 2 evidence as a side effect: reviewed changes, access reviews, deploy logs, restore tests.

Day one

Day one, not month one.

Most projects spend their first month on setup. Ours is done the day you start, so the first real work begins the next morning.

Every item is yours from the start: your GitHub organization, your Slack workspace, your Azure tenant, your Cloudflare account.

day-one.logkickoff → done
  • GitHub repository builtIn your organization, with branch protection, CI and code scanning on.
  • Project board liveThe first stories from your kickoff call, estimated and ready to approve.
  • Private Slack channels builtOne to work in together, one for release notes, one for alerts.
  • Azure environment deployedDevelopment and staging, defined as code, with your team’s access set up.
  • Cloudflare configuredDNS, TLS, firewall and DDoS protection in front of everything.
  • Domains linked or transfers startedLinked the same day. Registrar transfers start the same day and finish on the registrar’s clock.
  • Security baseline onMulti-factor sign-in, Key Vault for secrets, least-privilege roles.
  • Monitoring and alerts onHealth checks, error tracking and cost alerts, reporting to Slack.
  • Documentation site startedArchitecture, runbook and decisions, published and kept current.
  • Credit ledger openedYour balance, live, from the first approved task.

Stages

From idea to live software, one proven stage at a time.

Every project moves through the stages the industry already trusts. Each ends at a gate you sign off, so you never pay to build on an unproven step.

  1. 01

    Discovery

    What are we building, and why?

    Users, outcomes, risks and an architecture outline, priced in engineer-hours.

    GateA plan and an estimate you approve

  2. 02

    Proof of concept

    Can the hardest part work?

    We prove the riskiest piece first with your real data, such as the AI reading your actual documents accurately.

    GateEvidence, and a go or no-go

  3. 03

    MVP

    Can people use it?

    The smallest complete version, in staging, with sign-in, core workflows and tests.

    GateYour team completes the core workflows

  4. 04

    Pilot

    Does it hold up in the real world?

    Real users and real data on a limited rollout, with daily feedback in Slack.

    GateSuccess measures met, no open critical issues

  5. 05

    Production

    Is it ready for everyone?

    Security review, load and restore tests, runbook and go-live.

    GateLive: fully tested, managed and secured

  6. 06

    Run and evolve

    What’s next?

    Managed operations, monthly reports, and a roadmap paid for with your credit.

    GateOngoing

Case study · Mojo Payments

A payments platform in seven days.

Mojo Payments needed a platform for merchants, resellers and point-of-sale partners: onboarding, payments, pricing and support in one place. One engineer directed the AI agents. Seven days later it was live in production.

Payments · Multi-tenant SaaS

from first commit to production
7days
of human engineering
72hours
of senior-engineer work, delivered by AI
2,015hours
output for every human hour
28×

What shipped

  • Merchants apply online and are reviewed, approved and onboarded in the portal
  • Resellers get branded sign-up pages on their own domains, drafted by AI from their website
  • A cloud point-of-sale gateway, priced and switched on location by location
  • Invoices paid by link with saved cards, split payments and business-card data
  • An AI agent reads a merchant’s processing statements and prices a competing offer
  • Register diagnostics that file reports and open issues for slow payments on their own
  • A public API, a command-line tool and a 42-page knowledge base

At $275 an engineer-hour, that is about $574,000 of engineering. One senior engineer working by hand would have needed 15 to 16 months.

Mojo Payments is affiliated with Red Rook’s founder. Figures are measured from the project’s repositories.

The model

Two moves. One price.

Most companies hire one firm to run the cloud and another to write the software. Red Rook does both, with AI doing the work and engineers accountable for it.

What we build

Not template websites or off-the-shelf stores. AI-native software with real business logic, real data and real money moving through it.

AI-native applications

Products where agents do the core work: reading, deciding within your rules, acting and handing exceptions to people.

Agents for your operations

The back-office work your team does by hand, run by agents with an audit trail and a human in the loop.

AI inside your product

Document reading, pricing, triage and assistants wired into your own data, with cost limits and evaluation tests.

Multi-tenant SaaS platforms

Organizations, roles, tenant isolation, billing and an admin portal, built the way your product actually works.

Payments and fintech

Processor connections, tokenization, settlement, statements and the compliance work around them.

Partner portals and public APIs

Self-service onboarding, API keys, webhooks and developer documentation for the people you sell through.

Move 1 · Run

Azure and Cloudflare, managed by AI

  • Tenant built to Microsoft’s landing-zone pattern, or taken over once an Azure partner has migrated it
  • Identity, policy and least-privilege access in Entra ID
  • Monitoring, patching, backups and recovery drills, around the clock
  • Cloudflare in front: DNS, TLS, firewall, DDoS protection and edge hosting
  • Email domain authentication (SPF, DKIM, DMARC) kept correct
  • Cost watching, right-sizing and a monthly report you can read
  • Security posture tracked in Defender for Cloud

Included in your monthly fee.

Move 2 · Build

Software, built by AI

  • AI-native applications and platforms designed around your business, not a template
  • Led by Claude Code agents, overseen and reviewed by a certified engineer
  • Every task estimated in engineer-hours before work starts
  • Code in your repositories, deployed through your pipelines
  • Paid for with development credit you already own

Paid from your development credit.

  1. 01

    Ask

    Describe what you need in plain language: a ticket, an email or a call.

  2. 02

    Estimate

    We price it in engineer-hours. You approve it, or change the scope.

  3. 03

    Build

    AI agents write the code and tests in a branch of your repository.

  4. 04

    Ship

    An engineer reviews and releases it. The hours post to your ledger.

Commitments

AI does the work. People keep the promises.

These are written into every agreement, at every plan.

Testing

Nothing reaches production on trust.

  • Every change ships with automated tests. Nothing merges until the full suite passes.
  • Unit, integration and end-to-end tests run on every pull request.
  • AI features have evaluation suites with pass marks. A model or prompt change that lowers quality is blocked.
  • Staging mirrors production. Every deployment is smoke-tested and can be rolled back in one step.
  • Backups are restored on a schedule to prove they work.

Security

Secure by default, not by request.

  • A named engineer approves every AI-written change before it reaches production.
  • Multi-factor sign-in for everyone with access. Least privilege, reviewed every quarter.
  • Secrets live in Azure Key Vault, never in code. Pipelines deploy without stored passwords.
  • Encrypted in transit (TLS 1.2 or later) and at rest.
  • Dependency and code scanning on every pull request, with patches applied on a schedule.
  • Security incidents reported to you within 24 hours.

How we use Azure

Your tenant, built so anyone could rebuild it.

  • Your tenant, your billing and your choice of region.
  • Everything is defined as code in your repository, so any environment can be rebuilt from scratch.
  • Separate development, staging and production environments.
  • Container Apps and Functions to run code, Azure SQL for data, Key Vault for secrets.
  • Entra ID for your staff, External ID for your customers.
  • Azure Monitor, Application Insights, Defender for Cloud and budget alerts.
  • Claude and other models through Microsoft Foundry or directly, on keys you own.

How we use Cloudflare

Everything public goes through the edge first.

  • Authoritative DNS and managed TLS certificates.
  • Web application firewall, DDoS protection, bot management and rate limiting.
  • Zero Trust Access in front of admin tools, so no VPN.
  • Turnstile on public forms instead of CAPTCHAs.
  • Workers and Pages for marketing sites, documentation and edge logic.
  • Email routing and DMARC monitoring for your domains.

SOC 2

SOC 2 ready

Audit-ready from the first commit.

  • Controls mapped to the AICPA Trust Services Criteria from day one.
  • Evidence gathered as we work: reviewed changes, access reviews, deploy logs, incident records, restore tests.
  • A full report pack ready for your auditor: system description, control matrix, policies and evidence index.
  • SOC 2 reports are issued by independent CPA firms. We get you ready; your auditor attests.

Development credit

Your subscription comes back as code.

A Rook is $2,000 a month. Each Rook includes $100,000 of development credit for every year of your agreement: about 364 hours of senior engineering, 4.2 times what the Rook costs you, with the managed cloud included.

Agreement term

2 Rooks × 2 yrs × $100,000 = $400,000

Development credit
$400,000
Engineer-hours at $275
1,455
Monthly price
$4,000
Total you pay over the term
$96,000

Engineer time

Hours our engineers spend on your work are logged and charged at $275 an hour.

AI time, converted

AI work is priced in engineer-hours: what a senior engineer would need to do it by hand. You approve that estimate first, and it never grows after you approve it.

A ledger, not a black box

Every task posts to a ledger you can see, with the estimate, who did the work and your remaining balance.

Plans

Pick your Rooks.

Fixed monthly prices on 1, 2 or 3-year agreements. Add Rooks whenever you need more room on the board.

Just starting out?

New Business Bundle

Everything a new business needs to look established on day one, and room to build what comes next.

$1,000/month

Half a Rook · + $50,000 credit / yr
≈ 182 engineer-hours

  • A marketing website designed, launched and hosted on Cloudflare
  • A 24/7 AI chat agent that answers questions and sends you the leads
  • Microsoft 365: email on your domain, Teams, OneDrive and calendars
  • Get paid from day one: card payments and payment links through Mojo Payments

Also included

  • Online invoicingCardACH
  • Social profiles set up
  • Deposits to your bank
  • Lead inbox (HubSpot CRM)
  • Online booking
  • SEO enrollment
  • Ready for AI search
  • Listed on every map
  • Email authentication
  • Monthly business report

Microsoft 365 licenses are billed per user at Microsoft’s price. Grow into a full Rook whenever you’re ready.

One Rook

A single production workload, run properly.

$2,000/month

+ $100,000 credit / yr
≈ 364 engineer-hours

  • Everything on the day-one list, live the day you start
  • One managed Azure subscription
  • Cloudflare DNS, edge security and hosting for your domains
  • Private Slack channels and a live project board
  • AI monitoring, patching and backups
  • SOC 2 evidence collected from the first commit
  • Business-hours escalation to an engineer

Four Rooks

The whole board, run for you.

$8,000/month

+ $400,000 credit / yr
≈ 1,455 engineer-hours

  • Unlimited subscriptions in your tenant
  • Everything in Two Rooks
  • Quarterly architecture review
  • SOC 2 report pack prepared for your auditor

In every plan

A dedicated, Claude Certified engineer leads your project

Claude leads most of the work. One named human engineer, certified by Anthropic, oversees it, settles open questions, approves every release and answers for your production systems.

  • AI spend guardrailsModel keys you own, usage caps and an AI cost line in every monthly report.
  • Evaluation tests for AI featuresA fixed test set re-run on every release, so a model or prompt change can’t quietly make answers worse.
  • Secrets and dependencies kept currentSecrets in Key Vault with scheduled rotation. Automated dependency updates and security patches.
  • Restore tests, not just backupsA restore drill every month, with the result in your report.
  • Uptime monitoring and a status pageHealth checks from around the world, and a status page your customers can see.
  • Living documentationArchitecture, runbook and decisions kept current by AI and published on your own docs site.
  • Exit readinessA handover document kept current, so you could take everything elsewhere on a week’s notice.
  • Identity baselineMulti-factor sign-in, emergency admin accounts and a quarterly access review.

Each Rook includes $200 a month of Azure, Cloudflare and AI tooling. Usage above that is billed at the provider’s price with no markup. Card payments are processed by Mojo Payments. Need more than four Rooks? We’ll build the board with you.

Ways to start

Start where you are.

An idea you’re still shaping, a plan that needs an engineer’s eye, or a project ready to go. Each has its own way in.

  1. 01Free

    Think it through with our Discovery agent

    Talk your idea over with an AI agent that asks the questions an architect would: who it’s for, what it must do, what could go wrong. You keep the summary it writes.

    Coming soon
  2. 02One-time fee

    Not enough? Build the spec with an engineer

    Collaboration sessions with a certified engineer to shape and document your idea. You get a full specification. If you like what you read, it holds everything we need to build your project.

    Book spec sessions →
  3. 03Today

    Ready? Start now

    Pick a plan and pay by card. Your account is created the moment payment clears, and your day-one setup starts right away.

    Choose a plan →

Built on

We build on the platforms you already trust.

Our AI engineers work these platforms directly through their command-line tools, MCP servers and APIs. That’s why the work is fast and the result is standard.

  • Microsoft AzureCloud tenant, data and AI servicesCLI · MCP
  • CloudflareDNS, edge security, Workers and PagesCLI · MCP
  • Anthropic ClaudeThe models and agents inside your productAPI · SDK
  • GitHubYour repositories, reviews and pipelinesCLI · MCP
  • iPaaS.comIntegrations between your platform and the systems you runAPI · CLI · SDK · MCP
  • SlackYour project channels: releases, approvals and feedbackAPI · MCP
  • Mojo PaymentsCard processing, payment pages and POS connectionsAPI · CLI
  • ShopifyCustom apps and storefront extensionsCLI · MCP
  • MintlifyDocumentation and developer portalsCLI
  • Microsoft 365 & EntraSign-in, identity and Microsoft GraphCLI · API
  • HubSpotCRM, forms and customer lifecycleCLI · MCP
  • TwilioText, voice and verificationCLI · API
  • Committed to something else?AWS, Google Cloud, Teams, GitLab, Jira or your own systems: tell us on the first call. If it has a command-line tool or an API, we can build on it.Tell us your stack →

Product names are trademarks of their owners. Listing a platform does not imply its endorsement.

FAQ

Fair questions.

What is an “AI engineer-hour”?

Before any work starts, we estimate the task in engineer-hours: how long a senior engineer would take to do it by hand. You approve that estimate. Work our AI agents deliver is charged against your credit at that estimate, whether the agents finish in ten minutes or ten hours. Time our people spend is charged as it is logged. Both show up line by line in your credit ledger.

When can I use my development credit?

From day one. The full credit for your term is available as soon as your agreement starts, and you can spend it at whatever pace suits you during the term.

Whose tenant is it?

Yours. We work inside a Microsoft Azure tenant you own, under delegated, auditable access. If you leave, you keep the tenant, the code and the documentation. Nothing is held hostage.

Who owns the code?

You do. Everything we build with your credit is assigned to you on delivery and lives in your repositories.

Do people actually look at the code?

Yes. Claude leads most of the work, including most day-to-day design decisions, within the architecture we agree with you. Your dedicated engineer oversees it, settles the open questions, reviews every change before it ships and answers for production. Each pull request records which parts were AI-written and who approved it.

What if I run out of development credit?

Two ways, your choice. Extend your agreement or add Rooks, and the new credit is available instantly: $100,000 for every Rook-year. Or keep your plan and pay for extra hours as you use them, at $275 an engineer-hour.

Can I start without talking to anyone?

Yes. Choose a plan and pay by card. Your account on our platform is created as soon as payment clears, and your day-one setup begins: repository, Slack channels, Azure and Cloudflare. Payments are processed by Mojo Payments.

What if I’m not sure what to build yet?

Start with our free Discovery agent: it talks your idea through and writes a summary you keep. If you need more, book spec sessions with a certified engineer for a one-time fee. You get a full specification, ready to build from.

What makes an application “AI-native”?

The AI is part of how the product works, not a chat box added later. Agents read documents, make decisions within rules you set, act through your systems and hand exceptions to people. We build in model cost limits, evaluation tests and logs, so you can see what the AI did and why.

Why Cloudflare as well as Azure?

Cloudflare sits in front of your applications: DNS, TLS, firewall, DDoS protection and fast global hosting. We manage it alongside your Azure tenant, so your domains, email authentication and edge security are covered by the same fee.

Will working with you make us SOC 2 compliant?

It makes you ready. The controls are in place and the evidence is collected from day one, and we prepare the report pack your auditor asks for. The SOC 2 report itself is issued by an independent CPA firm that you engage.

We’re committed to a different platform. Can you work with it?

Usually, yes. If you run on AWS or Google Cloud, use Teams instead of Slack, or GitLab and Jira instead of GitHub, tell us on the first call. Our AI engineers work with almost any platform that has a command-line tool or an API.

Do you migrate existing systems?

Yes, with help. Moving existing systems into Azure is done with a certified Microsoft Azure partner, and separate fees apply. The partner makes sure everything is migrated, secured and ready, then we take over and run it. We can introduce you to a partner and coordinate the handover.

Your move

Book a strategy call.

Thirty minutes. Tell us what you run and what you want built, and we’ll show you how many Rooks it takes.

Or email hello@theredrook.ai