ProgrammableMarketing/radiant.co (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 6
  • This quarter · 9
  • Big bets · 3
  • Wildcards · 2
  • How we can help
Generated Oct 26, 2026
  1. Programmable
  2. Playbooks
  3. Radiant
Radiant logo

growth playbook · Oct 26, 2026

radiant.co (opens in a new tab)

Radiant growth playbook

Diagnosis

Radiant combines Brookfield capital and powered land with Ori's cloud software, and its hiring shows a marketing and developer push: a technical product marketer, a social and communications director, DevRel, a technical writer, a Head of Design and product managers for MLOps and the platform.

The website tells the AI factory story well but leaves the on-demand cloud without a front door: old Ori pricing and product links land on a homepage copy, there is no pricing page, the docs use different product names, and only one case study is shown.

This playbook builds the builder path first (prices, service pages, quickstarts, clear calls to action), then adds proof, a speaking engine, a capacity form, POCs and sovereign briefings, with a cost-per-token model, a telco program and a FlightDeck tour as bigger bets.

Growth Score

7/10

Biggest Opportunity

Run two clear paths from one site: a self-serve builder path that keeps the Ori cloud growing, and an evidence-rich path for hyperscale and sovereign capacity.

Biggest Risk

CoreWeave, Nebius, Lambda, Nscale and Crusoe compete for the same customers; Nebius and Lambda publish on-demand prices, so builders who cannot find Radiant's prices or products go elsewhere.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Give Ori's on-demand cloud a front door: ori.co/pricing lands on a copy of the homepage

    List the top 20 old ori.co URLs by traffic and where each now lands.

    High impact First result in 1 month
  2. 2

    One product vocabulary across the website, docs and console

    List every product name used on the site, docs and blog.

    High impact First result in 1 month
  3. 3

    A current quickstart: first workload on GB300 or H200 in ten minutes

    Time today's path from sign-up to a running model.

    High impact First result in 1 month

02

Your idea portfolio

20 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 7 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 2Big bets · 8Fill-ins · 6Reconsider · 4
Effort →

Showing 20 of 20 ideas

This week 6

Quick wins with a clear first action.

  • 01

    Radiant said it would keep growing the Ori Global AI Cloud for customers who need on-demand capacity, and the docs describe pay-as-you-go VMs, Serverless Kubernetes, Inference Endpoints, Fine-Tuning and a Model Registry. But ori.co/pricing and ori.co/gpu-cloud now redirect to radiant.co/ori, a copy of the homepage with no prices, and radiant.co has no pricing page and no pages for inference, fine-tuning or the model registry. A developer who searched for Ori's GPU prices lands on AI factory messaging. Radiant is hiring DevRel to cut time-to-first-workload, a technical writer and a Head of Design who owns the website and console.

    High impactMedium effort First result in 1 month Senior Developer Relations

    First action

    List the top 20 old ori.co URLs by traffic and where each now lands.

    How to measure it

    Builder signups

    Now
    –
    Target
    +50%
    By
    2 quarters

    Steps

    1. 1

      Map redirects

      Old ori.co URLs and their best new page.

      Web team 2 days

    2. 2

      Decide prices

      On-demand or "from" prices per GPU.

      Senior/Principal Product Manager - MLOps 1 week

    3. 3

      Write the page

      Services, prices, quickstart, sign-up.

      Director, Technical Product Marketing 1 week

    4. 4

      Add service pages

      Inference, fine-tuning, model registry.

      Director, Technical Product Marketing 2 weeks

    5. 5

      Design

      Builder path in the site and console style.

      Head of Design 2 weeks

    6. 6

      Fix redirects

      Pricing and product URLs to new pages.

      Web team 1 day

    7. 7

      Measure

      Signups and first workloads by source.

      Senior Developer Relations weekly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Senior Developer Relations

    Smallest test

    Redirect ori.co/pricing to a simple page with "from" prices for two weeks and compare signups.

    Why it works

    Build a "Radiant AI Cloud" page for builders with on-demand GPU prices (or "from" prices), the self-serve services, a quickstart and the sign-up button. Point ori.co/pricing and the old product URLs to the matching new pages instead of /ori, and add Inference Endpoints, Fine-Tuning and Model Registry pages that link to their docs. Keep "Talk to an architect" for long-term capacity.

    Developers choose a cloud by price and time to first workload, and rivals such as Nebius and Lambda publish both. A clear builder path keeps the on-demand business growing while the enterprise and sovereign sales run on their own track.

    Ready to use

    ChecklistBuilder front door launch list

    [ ] Top 20 ori.co URLs mapped to new pages
    [ ] /cloud page with on-demand "from" prices
    [ ] Inference Endpoints, Fine-Tuning, Model Registry pages
    [ ] Quickstart linked from each page
    [ ] Sign-up and "Talk to an architect" split by buyer
    [ ] New pages in the sitemap
    [ ] Signup source tracking

    Risks and how to handle them

    • Prices change often

      Show "from" prices with a date and link to the console for live rates.

    • Mixed message with AI factories

      Keep builder and enterprise paths visibly separate.

    Evidence

    • Redirect checkori.co, ori.co/pricing and ori.co/gpu-cloud all redirect to radiant.co/ori, while Ori blog posts redirect one to one (for example /blog/nvidia-h200-vs-h100). radiant.co/ori is the Radiant homepage with a merger note and an FAQ ("the Ori Global AI Cloud continues to operate for customers as the Radiant AI Cloud"); it has no prices and is not in the sitemap.Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
    • Your websiteSitemap (132 URLs, read October 23, 2026): product pages for bare metal, virtual machines, managed Kubernetes, Slurm, storage, networks, FlightDeck, GPUs, datacenters, hyperscale and sovereign solutions; no pricing page and no pages for Inference Endpoints, Fine-Tuning or Model Registry. The 78 blog posts include "Introducing Radiant Inference Endpoints", "Introducing Serverless Kubernetes", how-to guides for running open models on cloud GPUs, and posts about "Radiant AI Catalyst". The homepage shows one case study (Baseten).Source
    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source
    • HiringSenior Developer Relations (posted October 13, 2026): "help developers discover, understand, adopt, and get value from Radiant's cloud platform" through "tutorials, guides, reference architectures, demos, sample applications", and "Help improve documentation, onboarding, developer experience, and time-to-first-workload".Source
    • Competitor researchRivals' own pages: Nebius publishes "On-demand prices: NVIDIA GB300, B300, GB200, B200, H200, H100, RTX PRO 6000, L40S"; Lambda offers "transparent pricing for on-demand GPUs, 1-Click Clusters, and reserved capacity"; CoreWeave, "The Essential Cloud for AI"; Nscale, a "full-stack AI cloud platform"; Crusoe, "The AI factory company" with "an energy-first approach".Source
  • 02

    The website sells Bare Metal, Managed Kubernetes and FlightDeck. The docs call the same things Supercomputers or "Radiant Metal" and Serverless Kubernetes, under "Radiant AI Cloud (RAIC)", and blog posts still describe "Radiant AI Catalyst". A buyer moving from a product page to the docs cannot tell whether they are reading about the same product.

    High impactLow effort First result in 1 month Senior Technical Writer

    First action

    List every product name used on the site, docs and blog.

    How to measure it

    Pages updated

    Now
    –
    Target
    100% of product and docs pages
    By
    1 quarter

    Steps

    1. 1

      Inventory names

      Site, docs, console, blog.

      Senior Technical Writer 2 days

    2. 2

      Agree names

      With product and marketing.

      Chief Marketing Officer 1 week

    3. 3

      Write glossary

      With old names.

      Senior Technical Writer 2 days

    4. 4

      Update docs

      Navigation and headings.

      Senior Technical Writer 2 weeks

    5. 5

      Update site

      Product pages.

      Web team 1 week

    6. 6

      Update console

      Labels.

      Head of Design 1 month

    First result
    1 month
    Cost
    Staff time only
    Owner
    Senior Technical Writer

    Smallest test

    Publish the glossary page.

    Why it works

    Agree one name for each product, write a short glossary with the old names, and update the website, docs navigation, console labels and the top blog posts. Add "formerly" notes where an old name still gets searched.

    Consistent names cut confusion in evaluations and help search and AI assistants connect the pages.

    Risks and how to handle them

    • Customers know old names

      Keep "formerly" notes for a year.

    Evidence

    • Your websiteSitemap (132 URLs, read October 23, 2026): product pages for bare metal, virtual machines, managed Kubernetes, Slurm, storage, networks, FlightDeck, GPUs, datacenters, hyperscale and sovereign solutions; no pricing page and no pages for Inference Endpoints, Fine-Tuning or Model Registry. The 78 blog posts include "Introducing Radiant Inference Endpoints", "Introducing Serverless Kubernetes", how-to guides for running open models on cloud GPUs, and posts about "Radiant AI Catalyst". The homepage shows one case study (Baseten).Source
    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source
    • HiringSenior Technical Writer: "own and evolve Radiant's developer and product documentation, including guides, tutorials, API documentation", "getting-started guides", working with Product, Engineering, SRE, Support and Solutions. Head of Design (US, posted October 16, 2026): owns "how Radiant looks, feels, and communicates across our cloud platform, developer experience, website, brand", including "console, documentation".Source
    • Redirect checkori.co, ori.co/pricing and ori.co/gpu-cloud all redirect to radiant.co/ori, while Ori blog posts redirect one to one (for example /blog/nvidia-h200-vs-h100). radiant.co/ori is the Radiant homepage with a merger note and an FAQ ("the Ori Global AI Cloud continues to operate for customers as the Radiant AI Cloud"); it has no prices and is not in the sitemap.Source
  • 03

    The docs getting-started list ends with running a Stable Diffusion model on Serverless Kubernetes, while the site leads with GB300, H200 and Rubin. DevRel is hired to improve time-to-first-workload.

    High impactLow effort First result in 1 month Senior Developer Relations

    First action

    Time today's path from sign-up to a running model.

    How to measure it

    Median time to first workload

    Now
    –
    Target
    under 15 minutes
    By
    1 quarter

    Steps

    1. 1

      Time today

      Sign-up to running model.

      Senior Developer Relations 1 day

    2. 2

      Write VM quickstart

      H200 and vLLM.

      Senior Developer Relations 1 week

    3. 3

      Write endpoint quickstart

      Autoscaling endpoint.

      Senior Developer Relations 1 week

    4. 4

      Edit

      Docs style.

      Senior Technical Writer 3 days

    5. 5

      Fix friction

      Console steps.

      Senior/Principal Product Manager - MLOps 1 month

    6. 6

      Measure

      Funnel in the console.

      Senior Developer Relations weekly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Senior Developer Relations

    Smallest test

    Publish the VM quickstart.

    Why it works

    Write two quickstarts: launch an H200 VM and serve an open model with vLLM, and deploy an inference endpoint with autoscaling. Time each from sign-up, put them on the docs home and the new builder page, and fix every step that slows them.

    A fast first success is the strongest predictor that a developer stays, and timing it shows where the console needs work.

    Risks and how to handle them

    • GPU availability varies

      Offer two GPU options in each quickstart.

    Evidence

    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source
    • HiringSenior Developer Relations (posted October 13, 2026): "help developers discover, understand, adopt, and get value from Radiant's cloud platform" through "tutorials, guides, reference architectures, demos, sample applications", and "Help improve documentation, onboarding, developer experience, and time-to-first-workload".Source
    • Your websiteSign-up page: "Provision GPU clusters, fine-tune models, and run high-performance inference in just moments", with account creation on the page. The GPUs page (GB300 NVL72, H200, Vera Rubin NVL72) ends in "Contact us" and "Speak to a GPU Specialist", with no prices or on-demand path; the footer reads "© 2025".Source
    • HiringSenior Technical Writer: "own and evolve Radiant's developer and product documentation, including guides, tutorials, API documentation", "getting-started guides", working with Product, Engineering, SRE, Support and Solutions. Head of Design (US, posted October 16, 2026): owns "how Radiant looks, feels, and communicates across our cloud platform, developer experience, website, brand", including "console, documentation".Source
  • 04

    The homepage puts "Talk to an architect" next to "Sign up", but the GPUs page ends only in "Contact us" and "Speak to a GPU Specialist". A developer ready to launch an H200 and a government planning a campus see the same button.

    Medium impactLow effort First result in 2 weeks Head of Design

    First action

    Add a "Launch on demand" button to the GPUs page.

    How to measure it

    Product-page signups

    Now
    –
    Target
    +30%
    By
    60 days

    Steps

    1. 1

      Define paths

      Builder and capacity.

      Head of Design 1 day

    2. 2

      Design buttons

      Two per page.

      Head of Design 2 days

    3. 3

      Update pages

      GPUs and compute pages.

      Web team 2 days

    4. 4

      Tag forms

      Path in HubSpot.

      Marketing Operations 1 day

    5. 5

      Route

      Capacity to architects.

      Sales 1 day

    6. 6

      Measure

      Clicks and outcomes.

      Marketing Operations monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Head of Design

    Smallest test

    Change the GPUs page this week.

    Why it works

    On each product page add two clear paths: "Launch on demand" (sign-up and quickstart) and "Plan dedicated capacity" (architect form with market, load and RFS). Track each path.

    Each buyer gets the next step they want, and sales stops sorting small requests by hand.

    Risks and how to handle them

    • Enterprise buyers click sign-up

      Ask expected GPU volume at sign-up and route large ones.

    Evidence

    • Your websiteSign-up page: "Provision GPU clusters, fine-tune models, and run high-performance inference in just moments", with account creation on the page. The GPUs page (GB300 NVL72, H200, Vera Rubin NVL72) ends in "Contact us" and "Speak to a GPU Specialist", with no prices or on-demand path; the footer reads "© 2025".Source
    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
  • 05

    On mobile the homepage scores 39 and its largest element, the intro paragraph, paints at 10.5 seconds in the lab; real visitors' load time is 2.7 seconds and getting slower. Text should be the fastest thing on a page, so fonts and scripts are holding it back.

    Medium impactLow effort First result in 2 weeks Web team

    First action

    Defer the largest unused script.

    How to measure it

    Mobile score

    Now
    –
    Target
    70+
    By
    30 days

    Steps

    1. 1

      Find blockers

      Render-blocking files.

      Web team 1 day

    2. 2

      Fix fonts

      Preload or fallback.

      Web team 1 day

    3. 3

      Defer scripts

      Non-critical JS.

      Web team 2 days

    4. 4

      Delay animations

      After first paint.

      Web team 2 days

    5. 5

      Retest

      Homepage and sign-up.

      Web team 1 hour

    6. 6

      Watch field data

      After 28 days.

      Web team 1 month

    First result
    2 weeks
    Cost
    $0
    Owner
    Web team

    Smallest test

    Fix the font loading and retest.

    Why it works

    Preload the main font or use a system fallback, defer non-critical scripts (1.1 MB unused), and move animations after first paint. Retest the homepage and the sign-up page.

    A fast first view matters to technical buyers who judge infrastructure companies by their own engineering.

    Risks and how to handle them

    • Design changes

      Agree with the Head of Design.

    Evidence

    • Google PageSpeedHomepage on mobile: score 39, largest contentful paint 10.5s in the lab (the largest element is the intro paragraph text), 1.0s of blocking time; desktop 55 with 1.5s of blocking time. Real visitors: LCP 2.7s ("needs improvement", 19% slower than before), INP 95 ms and CLS 0. Biggest savings: render-blocking requests (2.7s) and 1.1 MB of unused JavaScript.Source
  • 06

    radiant.co/ori answers the main question Ori customers have, but it is a copy of the homepage and is not in the sitemap, and the footer still says © 2025. People searching "Ori GPU cloud" or "Ori pricing" need a clear answer about what changed.

    Medium impactLow effort First result in 2 weeks Web team

    First action

    Draft five questions Ori customers ask.

    How to measure it

    Ori brand search clicks

    Now
    –
    Target
    held at pre-merger level
    By
    2 quarters

    Steps

    1. 1

      Draft questions

      From support tickets.

      Cloud Infrastructure Support Engineer 1 day

    2. 2

      Write page

      Changes and links.

      Director, Technical Product Marketing 1 day

    3. 3

      Publish

      Own layout.

      Web team 1 day

    4. 4

      Add to sitemap

      And footer year.

      Web team 1 hour

    5. 5

      Email customers

      Link to the page.

      Customer Success 1 day

    6. 6

      Measure

      Search clicks.

      Web team monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Web team

    Smallest test

    Add /ori to the sitemap today.

    Why it works

    Turn /ori into a short page of its own: what changed, what stayed the same (console, docs, support, billing), the new names, and links to the builder page and docs. Add it to the sitemap and update the footer year.

    A clear merger page keeps Ori's brand searches and reassures existing customers.

    Risks and how to handle them

    • Ori name confusion lasts

      Keep the page live for at least a year.

    Evidence

    • Redirect checkori.co, ori.co/pricing and ori.co/gpu-cloud all redirect to radiant.co/ori, while Ori blog posts redirect one to one (for example /blog/nvidia-h200-vs-h100). radiant.co/ori is the Radiant homepage with a merger note and an FAQ ("the Ori Global AI Cloud continues to operate for customers as the Radiant AI Cloud"); it has no prices and is not in the sitemap.Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
    • Your websiteSign-up page: "Provision GPU clusters, fine-tune models, and run high-performance inference in just moments", with account creation on the page. The GPUs page (GB300 NVL72, H200, Vera Rubin NVL72) ends in "Contact us" and "Speak to a GPU Specialist", with no prices or on-demand path; the footer reads "© 2025".Source

This quarter 9

The core plays for the next three months.

  • 07

    The Technical PMM is asked to build Radiant's library of customer, partner and product proof. The homepage shows one case study (Baseten). Hyperscaler, sovereign and telco buyers need several examples with architecture, scale and outcomes.

    High impactMedium effort First result in 1 quarter Director

    First action

    List customers who would agree to a story.

    How to measure it

    Customer stories

    Now
    –
    Target
    6
    By
    4 quarters

    Steps

    1. 1

      List candidates

      Across segments.

      Director, Technical Product Marketing 2 weeks

    2. 2

      Get approvals

      Legal and customer.

      Director, Technical Product Marketing 1 month

    3. 3

      Interview

      Problem, deployment, results.

      Director, Technical Product Marketing monthly

    4. 4

      Write

      Story and post.

      Director, Technical Product Marketing monthly

    5. 5

      Design

      Templates.

      Head of Design 2 weeks

    6. 6

      Distribute

      Sales and social.

      Director, Social Media & Technical Communications ongoing

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Director, Technical Product Marketing

    Smallest test

    Publish one new builder story.

    Why it works

    Publish six stories in a year: two builder stories from the on-demand cloud, two enterprise or telco stories, and two infrastructure stories (time to energize, utilization, reliability). Each has a one-page PDF and a short technical post.

    Proof is what technical buyers ask for first, and it feeds sales, social and speaking at once.

    Risks and how to handle them

    • Confidential customers

      Describe by sector and scale.

    Evidence

    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source
    • Your websiteSitemap (132 URLs, read October 23, 2026): product pages for bare metal, virtual machines, managed Kubernetes, Slurm, storage, networks, FlightDeck, GPUs, datacenters, hyperscale and sovereign solutions; no pricing page and no pages for Inference Endpoints, Fine-Tuning or Model Registry. The 78 blog posts include "Introducing Radiant Inference Endpoints", "Introducing Serverless Kubernetes", how-to guides for running open models on cloud GPUs, and posts about "Radiant AI Catalyst". The homepage shows one case study (Baseten).Source
    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
  • 08

    The new social and communications director must run a weekly pipeline of talks, podcasts and executive posts. Radiant already has 78 posts, many deep technical pieces on NVIDIA platforms, multi-tenancy and sovereign AI.

    High impactMedium effort First result in 1 quarter Director

    First action

    Pick the five strongest posts for abstracts.

    How to measure it

    Speaking slots

    Now
    –
    Target
    10
    By
    2 quarters

    Steps

    1. 1

      Build calendar

      Deadlines.

      Director, Social Media & Technical Communications 2 weeks

    2. 2

      Pick posts

      Five strongest.

      Director, Social Media & Technical Communications 1 week

    3. 3

      Draft abstracts

      Two a week.

      Director, Social Media & Technical Communications weekly

    4. 4

      Executive posts

      Weekly.

      Director, Social Media & Technical Communications weekly

    5. 5

      Clips

      From talks.

      Director, Social Media & Technical Communications monthly

    6. 6

      Measure

      Bookings and reach.

      Director, Social Media & Technical Communications monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Director, Social Media & Technical Communications

    Smallest test

    Submit two abstracts this week.

    Why it works

    Each week, turn one deep dive into an executive LinkedIn post, a short clip and a talk abstract. Keep a calendar of conference deadlines and submit at least two abstracts a week.

    Existing content gives every channel material, and steady submissions produce talks over a quarter.

    Risks and how to handle them

    • Executives short on time

      Ghostwrite and approve in one pass.

    Evidence

    • HiringDirector, Social Media & Technical Communications (London, posted October 2, 2026): will "own the company's social presence across executive and corporate channels, with a particular focus on LinkedIn", build "the public voices of the company's executives and technical leaders", and run "a proactive speaking and visibility engine" with "a consistent weekly motion: sourcing opportunities, drafting abstracts, preparing submissions, pitching panels, finding podcasts".Source
    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • Your websiteSitemap (132 URLs, read October 23, 2026): product pages for bare metal, virtual machines, managed Kubernetes, Slurm, storage, networks, FlightDeck, GPUs, datacenters, hyperscale and sovereign solutions; no pricing page and no pages for Inference Endpoints, Fine-Tuning or Model Registry. The 78 blog posts include "Introducing Radiant Inference Endpoints", "Introducing Serverless Kubernetes", how-to guides for running open models on cloud GPUs, and posts about "Radiant AI Catalyst". The homepage shows one case study (Baseten).Source
  • 09

    Radiant's blog benchmarks schedulers (Kueue, Volcano, Slinky) and inference endpoints. DevRel is hired to build reference implementations and sample applications.

    Medium impactMedium effort First result in 1 quarter Senior Developer Relations

    First action

    Choose the first benchmark to publish as code.

    How to measure it

    Repo forks

    Now
    –
    Target
    200
    By
    2 quarters

    Steps

    1. 1

      Pick benchmarks

      Two to start.

      Senior Developer Relations 1 week

    2. 2

      Clean code

      Scripts and configs.

      Senior Developer Relations 2 weeks

    3. 3

      Publish

      Public repo.

      Senior Developer Relations 1 day

    4. 4

      Link

      Posts and docs.

      Senior Technical Writer 1 day

    5. 5

      Promote

      Developer channels.

      Director, Social Media & Technical Communications 1 week

    6. 6

      Track

      Signups from repo.

      Senior Developer Relations monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Senior Developer Relations

    Smallest test

    Publish one benchmark repo.

    Why it works

    Put each benchmark and how-to in a public repository with scripts that run on Radiant, and link from the post and docs. Invite developers to reproduce results and share them.

    Reproducible results earn trust with engineers and turn readers into users.

    Risks and how to handle them

    • Results challenged

      Publish method and hardware details.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • HiringSenior Developer Relations (posted October 13, 2026): "help developers discover, understand, adopt, and get value from Radiant's cloud platform" through "tutorials, guides, reference architectures, demos, sample applications", and "Help improve documentation, onboarding, developer experience, and time-to-first-workload".Source
    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source
  • 10

    The hyperscaler page says capacity is ready in six months on powered sites and asks buyers to send market, load and RFS, because availability is not published. A structured form and a standard reply make that conversation faster.

    High impactMedium effort First result in 1 quarter Sales

    First action

    Draft the form fields.

    How to measure it

    Qualified requests

    Now
    –
    Target
    +40%
    By
    2 quarters

    Steps

    1. 1

      Define fields

      Market, load, RFS.

      Sales 2 days

    2. 2

      Build form

      HubSpot.

      Marketing Operations 1 week

    3. 3

      Set routing

      By size.

      Sales 2 days

    4. 4

      Write replies

      Readiness stages.

      Director, Technical Product Marketing 1 week

    5. 5

      Set response time

      Two business days.

      Sales 1 day

    6. 6

      Report

      Demand by market.

      Marketing Operations monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Sales

    Smallest test

    Use the fields on the hyperscaler page.

    Why it works

    Replace the generic contact form with fields for market, critical IT load, RFS date and architecture, route by size, and reply within a set time with readiness stages that fit (powered site, expansion, greenfield).

    Large buyers move faster when the first reply is specific, and sales sees demand by market.

    Risks and how to handle them

    • Buyers hide details

      Allow ranges.

    Evidence

    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
  • 11

    Radiant targets sovereign governments and has a sovereign guide and a post on UK sovereign AI. Each country asks different questions about data residency, control planes and local operations.

    High impactMedium effort First result in 1 quarter Director

    First action

    Pick three priority countries.

    How to measure it

    Sovereign pipeline opportunities

    Now
    –
    Target
    5
    By
    3 quarters

    Steps

    1. 1

      Pick countries

      Three.

      Chief Marketing Officer 1 week

    2. 2

      Research

      Rules and needs.

      Director, Technical Product Marketing 1 month

    3. 3

      Write briefings

      One per country.

      Director, Technical Product Marketing 1 month

    4. 4

      Review

      Legal and policy.

      Legal 2 weeks

    5. 5

      Host round tables

      With partners.

      Director, Social Media & Technical Communications 1 quarter

    6. 6

      Follow up

      Account plans.

      Sales ongoing

    First result
    1 quarter
    Cost
    Event costs
    Owner
    Director, Technical Product Marketing

    Smallest test

    Write the UK briefing.

    Why it works

    Write short briefings for three priority countries: rules, options, how Radiant would deliver, and timelines. Present them at private round tables with officials and partners.

    Specific, local briefings show seriousness and open long-term contract talks.

    Risks and how to handle them

    • Policy changes

      Date and update each briefing.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source
  • 12

    Radiant publishes a reference architecture and whitepapers on sovereign AI, telcos and tenancy. People who download them are researching a decision.

    Medium impactLow effort First result in 1 month Marketing Operations

    First action

    Check what happens after a download today.

    How to measure it

    Download-sourced meetings

    Now
    –
    Target
    10 a month
    By
    2 quarters

    Steps

    1. 1

      Audit

      Current flow.

      Marketing Operations 1 day

    2. 2

      Add fields

      Role and org type.

      Marketing Operations 1 day

    3. 3

      Write sequences

      One per paper.

      Director, Technical Product Marketing 2 weeks

    4. 4

      Set alerts

      Target accounts.

      Marketing Operations 1 week

    5. 5

      Route

      To sales.

      Sales ongoing

    6. 6

      Measure

      Meetings.

      Marketing Operations monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Marketing Operations

    Smallest test

    Write the sequence for the reference architecture.

    Why it works

    Ask role and organization type on download, send a short sequence matched to the paper, and alert sales when someone from a target organization downloads two papers.

    Downloads are early intent; matched follow-up turns them into conversations.

    Risks and how to handle them

    • Forms reduce downloads

      Keep two fields only.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source
  • 13

    Radiant has a trust center and data center certification docs, and is hiring security leaders. Sovereign and enterprise buyers check compliance early.

    Medium impactLow effort First result in 1 month Head of Security Operations

    First action

    List current certifications.

    How to measure it

    Days per security review

    Now
    –
    Target
    -20%
    By
    2 quarters

    Steps

    1. 1

      List certifications

      Current and planned.

      Head of Security Operations 1 day

    2. 2

      Write block

      Short and factual.

      Director, Technical Product Marketing 1 day

    3. 3

      Add to pages

      Three pages.

      Web team 1 day

    4. 4

      Link trust center

      Every page.

      Web team 1 hour

    5. 5

      Review

      Quarterly.

      GRC Analyst quarterly

    6. 6

      Measure

      Review time.

      Sales quarterly

    First result
    1 month
    Cost
    $0
    Owner
    Head of Security Operations

    Smallest test

    Add the block to the sovereign page.

    Why it works

    Add a short compliance block to the sovereign, hyperscaler and builder pages: certifications held, data residency options and a link to the trust center. Keep it current.

    Answering security questions on the page shortens reviews.

    Risks and how to handle them

    • Out-of-date claims

      Quarterly review.

    Evidence

    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source
    • HiringProduct hiring in October 2026: Senior/Principal Product Managers for MLOps (training, "fine-tuning, model management, deployment, inference" and "self-service"), Infrastructure, Storage, Cluster Platform, Observability and Security; a Senior Field Product Engineer to "work directly with customers on technical discovery, onboarding" and "Support proofs of concept, technical evaluations"; a Cloud Infrastructure Support Engineer for "24/7, 365" "follow the sun" support. Every post says Radiant "was established by Brookfield", with "over US$1 trillion of assets under management".Source
    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
  • 14

    The Field Product Engineer will support proofs of concept and technical evaluations. Buyers choosing a GPU cloud want to test on their own workload before committing.

    High impactMedium effort First result in 1 quarter Senior Field Product Engineer

    First action

    Write the success-criteria template.

    How to measure it

    POC conversion

    Now
    –
    Target
    50%
    By
    3 quarters

    Steps

    1. 1

      Define offer

      Duration and scope.

      Senior Field Product Engineer 1 week

    2. 2

      Template

      Success criteria and report.

      Senior Field Product Engineer 1 week

    3. 3

      Publish page

      With form.

      Director, Technical Product Marketing 1 week

    4. 4

      Capacity plan

      Reserve POC nodes.

      Infrastructure Operations 2 weeks

    5. 5

      Run POCs

      With reports.

      Senior Field Product Engineer ongoing

    6. 6

      Review

      Win rate.

      Sales monthly

    First result
    1 quarter
    Cost
    GPU capacity for POCs
    Owner
    Senior Field Product Engineer

    Smallest test

    Run one POC with the new template.

    Why it works

    Publish a POC offer: two weeks on a defined cluster, a kickoff with a field engineer, agreed success criteria and a final report. Ask for the workload and success criteria in the form.

    A clear, time-boxed POC lowers risk for the buyer and makes the sales cycle predictable.

    Risks and how to handle them

    • Capacity tied up

      Limit concurrent POCs.

    Evidence

    • HiringProduct hiring in October 2026: Senior/Principal Product Managers for MLOps (training, "fine-tuning, model management, deployment, inference" and "self-service"), Infrastructure, Storage, Cluster Platform, Observability and Security; a Senior Field Product Engineer to "work directly with customers on technical discovery, onboarding" and "Support proofs of concept, technical evaluations"; a Cloud Infrastructure Support Engineer for "24/7, 365" "follow the sun" support. Every post says Radiant "was established by Brookfield", with "over US$1 trillion of assets under management".Source
    • Your websiteSign-up page: "Provision GPU clusters, fine-tune models, and run high-performance inference in just moments", with account creation on the page. The GPUs page (GB300 NVL72, H200, Vera Rubin NVL72) ends in "Contact us" and "Speak to a GPU Specialist", with no prices or on-demand path; the footer reads "© 2025".Source
    • Competitor researchRivals' own pages: Nebius publishes "On-demand prices: NVIDIA GB300, B300, GB200, B200, H200, H100, RTX PRO 6000, L40S"; Lambda offers "transparent pricing for on-demand GPUs, 1-Click Clusters, and reserved capacity"; CoreWeave, "The Essential Cloud for AI"; Nscale, a "full-stack AI cloud platform"; Crusoe, "The AI factory company" with "an energy-first approach".Source
  • 15

    Radiant will keep growing the on-demand cloud for builders, and DevRel wants adoption. Startups and university labs choose a cloud early and grow with it.

    Medium impactMedium effort First result in 1 quarter Senior Developer Relations

    First action

    Set the credit amount and eligibility.

    How to measure it

    Credit-to-paid conversion

    Now
    –
    Target
    25%
    By
    3 quarters

    Steps

    1. 1

      Set rules

      Amount and eligibility.

      Senior Developer Relations 1 week

    2. 2

      Partner

      Two accelerators, two universities.

      Senior Developer Relations 1 month

    3. 3

      Build page

      Application form.

      Web team 1 week

    4. 4

      Onboard

      Quickstarts and Discord.

      Senior Developer Relations ongoing

    5. 5

      Check usage

      Monthly.

      Senior Developer Relations monthly

    6. 6

      Convert

      Offers at credit end.

      Sales ongoing

    First result
    1 quarter
    Cost
    Credit costs
    Owner
    Senior Developer Relations

    Smallest test

    Offer credits to ten startups.

    Why it works

    Offer GPU credits to vetted startups and labs through accelerators and universities, with the quickstarts and a Discord channel. Track which accounts grow into paid usage.

    Credits build a base of builders who become paying customers and case studies.

    Risks and how to handle them

    • Credits abused

      Vet applicants and cap usage.

    Evidence

    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
    • HiringSenior Developer Relations (posted October 13, 2026): "help developers discover, understand, adopt, and get value from Radiant's cloud platform" through "tutorials, guides, reference architectures, demos, sample applications", and "Help improve documentation, onboarding, developer experience, and time-to-first-workload".Source
    • Documentationdocs.radiant.co ("Radiant AI Cloud (RAIC)"): Virtual Machines, Serverless Kubernetes, Supercomputers ("our Radiant Metal service"), Inference Endpoints, Fine-Tuning, Model Registry and Object Storage, each with a Billing page; on-demand services are "Pay-as-You-Go" through Stripe. The getting-started list ends with "Run a Stable Diffusion model on a Serverless Kubernetes cluster"; support is a support desk and Discord, and the support section includes a "Data Center Certifications" page; the site links a trust center.Source

Big bets 3

Larger plays that can change the trajectory.

  • 16

    Radiant's argument is economic: "inexpensive capital makes inexpensive intelligence", with below-market power from Brookfield's portfolio. Buyers compare on cost per token and cost per GPU hour; few providers show how they get there.

    High impactHigh effort First result in 3 quarters Director

    First action

    Agree the assumptions with finance.

    How to measure it

    Meetings from the model

    Now
    –
    Target
    20 a quarter
    By
    3 quarters

    Steps

    1. 1

      Agree assumptions

      Finance and infra.

      Director, Technical Product Marketing 1 month

    2. 2

      Build model

      Spreadsheet first.

      Director, Technical Product Marketing 1 month

    3. 3

      Design tool

      Web version.

      Head of Design 1 month

    4. 4

      Legal review

      Claims.

      Legal 2 weeks

    5. 5

      Launch

      With a talk.

      Director, Social Media & Technical Communications 1 month

    6. 6

      Use in sales

      Every proposal.

      Sales ongoing

    First result
    3 quarters
    Cost
    Design and engineering time
    Owner
    Director, Technical Product Marketing

    Smallest test

    Share the spreadsheet with three prospects.

    Why it works

    Build an interactive model: GPU type, utilization, power price, financing term and workload produce cost per GPU hour and per million tokens, with Radiant's assumptions shown. Use it in sales and talks.

    Showing the math makes the main differentiator concrete and gives sales a tool every buyer will use.

    Risks and how to handle them

    • Seen as a price promise

      Label it an estimate with stated assumptions.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
    • Competitor researchRivals' own pages: Nebius publishes "On-demand prices: NVIDIA GB300, B300, GB200, B200, H200, H100, RTX PRO 6000, L40S"; Lambda offers "transparent pricing for on-demand GPUs, 1-Click Clusters, and reserved capacity"; CoreWeave, "The Essential Cloud for AI"; Nscale, a "full-stack AI cloud platform"; Crusoe, "The AI factory company" with "an energy-first approach".Source
  • 17

    Radiant serves telecommunication providers and argues telcos hold the key to sovereign AI. Telcos have national networks, government relationships and enterprise sales teams.

    High impactHigh effort First result in 4 quarters Chief Marketing Officer

    First action

    Shortlist three telcos.

    How to measure it

    Telco partners

    Now
    –
    Target
    2
    By
    4 quarters

    Steps

    1. 1

      Shortlist telcos

      Three.

      Chief Marketing Officer 1 month

    2. 2

      Design program

      Terms and support.

      Partnerships 1 quarter

    3. 3

      Sign first

      One telco.

      Partnerships 2 quarters

    4. 4

      Train teams

      Partner sales.

      Senior Field Product Engineer 1 month

    5. 5

      Co-market

      Joint launch.

      Director, Social Media & Technical Communications 1 month

    6. 6

      Review

      Pipeline.

      Chief Marketing Officer quarterly

    First result
    4 quarters
    Cost
    Partner team costs
    Owner
    Chief Marketing Officer

    Smallest test

    Run one partner workshop.

    Why it works

    Design a program where telcos offer Radiant capacity under their own brand or as a joint service, with training, co-marketing and shared revenue. Start with one telco in a priority country.

    Partners extend reach into countries and enterprises Radiant cannot cover directly.

    Risks and how to handle them

    • Channel conflict

      Define which accounts each side owns.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • Press releaseFebruary 24, 2026: Radiant merged with Ori Industries; Radiant will offer AI compute "for sovereign governments, select global enterprises and telecommunication providers under long term contracts" and "will continue to grow and operate the Ori Global AI Cloud for customers who need on-demand capacity and rapid deployment". The cloud "presently offers pre-configured bare metal offerings, GPU instances, as well as a variety of AI as-a-service offerings including Inference, Kubernetes, Fine-Tuning, Model Registry, and Storage".Source
    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source
  • 18

    FlightDeck is described as "one control surface across the platform", and the Head of Design owns the console and website. Enterprise and sovereign buyers rarely get console access before contract.

    Medium impactHigh effort First result in 3 quarters Head of Design

    First action

    Pick the five screens that matter most to buyers.

    How to measure it

    Tour completions

    Now
    –
    Target
    500
    By
    3 quarters

    Steps

    1. 1

      Pick screens

      Five.

      Head of Design 2 weeks

    2. 2

      Sample data

      Realistic and safe.

      Product team 1 month

    3. 3

      Build tour

      Clickable.

      Head of Design 1 quarter

    4. 4

      Embed

      FlightDeck page.

      Web team 1 week

    5. 5

      Use in demos

      Sales.

      Sales ongoing

    6. 6

      Measure

      Completions to meetings.

      Marketing Operations monthly

    First result
    3 quarters
    Cost
    Design time
    Owner
    Head of Design

    Smallest test

    Record a two-minute video tour first.

    Why it works

    Build a guided, clickable tour of FlightDeck with sample data: fleet health, placement, tenancy and audit. Embed it on the FlightDeck page and use it in demos.

    Seeing the control plane makes an abstract platform claim tangible.

    Risks and how to handle them

    • Shows too much

      Use sample data only and review with security.

    Evidence

    • HiringSenior Technical Writer: "own and evolve Radiant's developer and product documentation, including guides, tutorials, API documentation", "getting-started guides", working with Product, Engineering, SRE, Support and Solutions. Head of Design (US, posted October 16, 2026): owns "how Radiant looks, feels, and communicates across our cloud platform, developer experience, website, brand", including "console, documentation".Source
    • Your websiteHomepage: "Trusted by technology giants. Adopted by builders. Partnered with nations.", with "Talk to an architect" and "Sign up" side by side; Brookfield brings "more than $100 billion in deployable capital"; "NVIDIA Cloud Partner"; a cloud "scaling from 10K to 100K+ GPUs". Hyperscaler page: "Capacity is ready in 6 months on powered sites and 9-12 months on greenfield sites" and "Availability is matched to your requirement, not published as a static list. Send us your market, load and RFS".Source
    • HiringDirector, Technical Product Marketing (reports to the CMO): focused on "the end-to-end datacenters we build" for "hyperscaler customers"; writes "product and solution pages, whitepapers, reference architecture content, case studies, customer stories"; will "Build Radiant's library of customer, partner, and product proof"; success means "The website explains the product more clearly".Source

Wildcards 2

Unexpected experiments, each with a cheap way to test it.

  • 19

    Radiant argues that utilization is the only AI metric that matters, and it runs 24/7 support and FlightDeck health tracking. A monthly public report of fleet uptime and utilization backs the claim with data.

    Medium impactMedium effort First result in 2 quarters Senior/Principal Product Mana…

    First action

    Choose three metrics that are safe to publish.

    How to measure it

    Monthly report readers

    Now
    –
    Target
    2,000
    By
    2 quarters

    Steps

    1. 1

      Pick metrics

      Three.

      Senior/Principal Product Manager - Observability 2 weeks

    2. 2

      Build pipeline

      Aggregates.

      Infrastructure Operations 1 month

    3. 3

      Review

      Security and legal.

      GRC Analyst 2 weeks

    4. 4

      Publish

      Monthly page.

      Director, Technical Product Marketing monthly

    5. 5

      Share

      Social.

      Director, Social Media & Technical Communications monthly

    6. 6

      Improve

      Add metrics.

      Senior/Principal Product Manager - Observability quarterly

    First result
    2 quarters
    Cost
    Analyst time
    Owner
    Senior/Principal Product Manager Observability

    Smallest test

    Publish one month internally first.

    Why it works

    Publish a monthly summary of aggregate GPU availability, node recovery times and utilization bands, with a short note on any incidents and fixes.

    Measured reliability is rare in GPU clouds and builds trust with operators.

    Risks and how to handle them

    • Bad months are public

      Explain fixes; consistency builds trust.

    Evidence

    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • HiringProduct hiring in October 2026: Senior/Principal Product Managers for MLOps (training, "fine-tuning, model management, deployment, inference" and "self-service"), Infrastructure, Storage, Cluster Platform, Observability and Security; a Senior Field Product Engineer to "work directly with customers on technical discovery, onboarding" and "Support proofs of concept, technical evaluations"; a Cloud Infrastructure Support Engineer for "24/7, 365" "follow the sun" support. Every post says Radiant "was established by Brookfield", with "over US$1 trillion of assets under management".Source
  • 20

    The blog has deep dives on GPUs, networking, storage, scheduling and how-to guides, but they are a single list. Developers and platform engineers learn in sequences.

    Medium impactLow effort First result in 1 quarter Senior Technical Writer

    First action

    Draft the first path from existing posts.

    How to measure it

    Subscribers

    Now
    –
    Target
    3,000
    By
    3 quarters

    Steps

    1. 1

      Group posts

      Three paths.

      Senior Technical Writer 1 week

    2. 2

      Write intros

      Each path.

      Senior Technical Writer 1 week

    3. 3

      Publish

      Path pages.

      Web team 1 week

    4. 4

      Newsletter

      Monthly.

      Director, Social Media & Technical Communications monthly

    5. 5

      Update

      New posts.

      Senior Technical Writer monthly

    6. 6

      Measure

      Signups from paths.

      Senior Developer Relations monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Senior Technical Writer

    Smallest test

    Publish the inference path.

    Why it works

    Group posts into three paths (choosing GPUs, running inference, building a multi-tenant cluster), add a short intro and next steps, and send a monthly newsletter with one path step and one new post.

    Paths keep readers longer and bring them back, which grows the signup pool.

    Risks and how to handle them

    • Old posts outdated

      Refresh before including.

    Evidence

    • Your websiteSitemap (132 URLs, read October 23, 2026): product pages for bare metal, virtual machines, managed Kubernetes, Slurm, storage, networks, FlightDeck, GPUs, datacenters, hyperscale and sovereign solutions; no pricing page and no pages for Inference Endpoints, Fine-Tuning or Model Registry. The 78 blog posts include "Introducing Radiant Inference Endpoints", "Introducing Serverless Kubernetes", how-to guides for running open models on cloud GPUs, and posts about "Radiant AI Catalyst". The homepage shows one case study (Baseten).Source
    • Your websiteResources and blog: a reference architecture, a sovereign guide, a telco whitepaper and a tenancy whitepaper; posts such as "Powering UK Sovereign AI", "Why Telcos Hold the Key to Sovereign AI", "The Math of Radiant: Inexpensive Capital Makes Inexpensive Intelligence", "Why Utilization Is the Only AI Metric That Matters" and benchmarks of Kueue, Volcano and Slinky.Source
    • HiringSenior Developer Relations (posted October 13, 2026): "help developers discover, understand, adopt, and get value from Radiant's cloud platform" through "tutorials, guides, reference architectures, demos, sample applications", and "Help improve documentation, onboarding, developer experience, and time-to-first-workload".Source

03

How Programmable can help

Where an idea in this playbook is something Programmable builds or runs, it's listed here with what we'd do. Everything else is yours to run.

ProgrammableDigital

Show it, don't just describe it

These ideas need something to show: a demo, a video, an animation, graphics or a deck. Digital turns screenshots or the site into all of them, in the company's own brand.

Make a demo (opens programmable.digital in a new tab)
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Parameter growth strategyAI agents that pentest web apps, APIs, cloud and code continuously: autonomous pentesting, Sentinel pull-request review, cloud security…23 growth ideas
  • Maven AGI growth strategyAn enterprise platform for AI agents that resolve customer service conversations on chat, email and voice, connecting to helpdesks, CRMs…21 growth ideas
  • Authentic Brands Group growth strategyA global brand and entertainment platform that owns iconic brands and celebrity and estate IP (Reebok, Champion, Brooks Brothers, Ted…22 growth ideas
  • Second Front Systems growth strategyA public-benefit software company whose 2F Suite (Game Warden, Workshop and Frontier) lets commercial software companies build, authorize…22 growth ideas
  • Zenor growth strategyA daily performance app: short morning and evening check-ins, one right-sized action a day and a weekly review that surfaces personal…22 growth ideas
  • Eloquent AI growth strategyAI Operators for financial services: software that learns a team's procedures (ESOPs), is tested on historical cases, is rolled out in…24 growth ideas
  • OP Labs growth strategyThe team behind Optimism's technology: it develops the OP Stack, ships upgrades every OP Stack chain inherits, and delivers OP Enterprise…20 growth ideas
  • Skin Analytics growth strategyA London health tech company behind DERM, a Class III CE-marked AI medical device for skin cancer triage used in NHS pathways, and DERM…23 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on October 26, 2026, and cites them throughout. Radiant did not commission, review or endorse it, and Programmable is not affiliated with Radiant. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Radiant runs its business.

Something wrong, or do you represent Radiant? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 306 ideas

    The builder front door and redirects, one product vocabulary, current quickstarts, split calls to action, mobile speed and the Ori page.

  2. Days 31 to 6011 ideas

    Customer proof, the speaking engine, benchmark code, the capacity form, sovereign briefings, download follow-up, compliance blocks, POCs and credits; utilization reports and learning paths.

  3. Days 61 to 903 ideas

    Start the big bets: the cost-per-token model, a telco partner program and a FlightDeck tour.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value access to current NVIDIA platforms. The most cited problem: GPU scarcity and long lead times.

  • Managed infrastructure from datacenter to workload
  • Power and site constraints
  • In their words: “time to token”

Market

Competitive landscape

5 competitors compared: CoreWeave, Nebius, Lambda, Nscale and 1 more.

  • Gap to exploit: Shown economics of capital and power
  • CoreWeave: Less vertical control of power and land
  • Nebius: Less sovereign focus

Hiring signals

Radiant is building its go-to-market: a Director of Technical Product Marketing to make the website explain the product and build customer proof (ideas 1, 7, 11, 16), a Director of Social Media & Technical…

  • Director, Technical Product Marketing: Website clarity and customer proof. Ideas 1, 7, 16.
  • Director, Social Media & Technical Communications: LinkedIn, executive voice, speaking pipeline. Idea 8.
  • Senior Developer Relations: Time-to-first-workload. Ideas 1, 3, 9, 15.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks first (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence