> For the complete documentation index, see [llms.txt](https://docs.reo.dev/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.reo.dev/agents/tech-projects/gpu-compute-intensity.md). # GPU Compute Intensity > *New here?* This page assumes you already know [what a Tech Project and a Tech Indicator are](https://docs.reo.dev/agents/tech-projects). If not, we'd recommend [you start there](https://docs.reo.dev/agents/tech-projects). It's a 2-minute read and everything below will make a lot more sense. GPU Compute Intensity is one of Reo's Tech Indicators, the signal that tells you a company is actually investing in a Tech Project *(for example - optimizing GPU spend)*. **This tech indicator helps you answer: how much your target accounts are actually investing in GPU compute** {% hint style="info" %} Reo runs on a library of agents. **Tech Indicators is one of Reo's agents. It rates how a company actually builds and runs its technology.** {% endhint %} If you sell GPU capacity, inference optimization, AI cost management, or AI governance and security tooling, **this tells you which accounts have the workloads that make your product necessary in the first place.** Companies spending heavily on GPUs aren't just interested in AI, they're already committed to the infrastructure that makes a product like yours relevant. Reo scores each account into one of three tiers. Here’s what the output looks like:

Tier	What it means + our recommended course of action
High	• Actively operating GPU infrastructure • Think self-hosted model training or large-scale inference • Treat it as a strong account; lead with scale and cost
Medium	• Meaningful AI investment • But hybrid, or still growing toward self-hosted compute • Your best window to get in early
Low	• Mostly an API consumer or an early-stage with minimal compute footprint • AI compute is likely running through someone else’s infra • Your product might not be a priority yet, but if you sell cost management, API routing, or observability tools, it’s worth pursuing

*** ### How to use it 1. From the left navigation menu, select **Agents** 2. Under **Chat agents >** select the **Tech Indicators card** > click **Chat and Build**

3. Choose the **list or segment you want scored** {% hint style="info" %} This runs on accounts you've already built, so start from whichever list you're working today. If you've just discovered new accounts through Reo Agents chat, add them to a list or segment first, then come back here to score them. {% endhint %}

4. Choose the technology signal you want to enrich your accounts with: **GPU Compute Intensity** or **AI Nativeness** {% hint style="info" %} That choice sets the prompt Reo runs for this pass. Want both on the same list? Run the enrichment twice, once for each. {% endhint %}

5. Click on **Preview sample** to check the preview of the first 10 accounts - each with a tier, its reasoning, and a short summary of what Reo found

6. Reo checks 5 sources and each source is circumstantial on its own: 1. the **Product agent** reads homepages, docs, changelogs, and engineering blogs 2. the **Hiring agent** parses live job descriptions, titles, and required stacks 3. the **GitHub agent** analyzes repos, contributors, and dependency graphs 4. the **LinkedIn agent** tracks headcount snapshots, team composition, and growth 5. the **Firmographic agent** analyzes funding, stage, revenue scale, and industry

7. **Reo agent runs against a prompt pertaining to the tech signal you’ve chosen.** The agent’s output returns the following for each account: 1. domain and company name 2. the tier or score for each account 3. a plain-English summary explaining the reasoning 8. Click **View Reasoning** next to any row. This opens the specific evidence behind that one account's answer, the exact facts Reo found and the logic connecting them, so you can check the result before you act on it.

9. At the end of the table, **Reo also synthesizes the pattern across the whole preview:** 1. which accounts are the strongest signal within a tier 2. where accounts split on a specific factor 3. which ones are outliers worth a second look 10. Click **Enrich & Add column**

11. The ‘Enrich and Add’ popup will open. Configure the following details before you proceed: 1. **Column name**: what the score appears as on your account list 2. **Re-run this enrichment**: pick a refresh cadence, monthly for example, and check the box to automatically score new accounts as they're added to this segment. You set this up once and it keeps on auto-pilot. 3. **Agent details**: name and tag this agent - this will let you add this column to other segments as well, so you can enrich your segment much faster

12. Once you’ve provided these details, click **Enrich & Add column** to run the enrichment

13. **You're taken back to your account list, and the score lands as a new column on every matching account:** 1. GPU Compute Intensity as High, Medium, or Low 2. AI Nativeness as AI-Native, AI-Powered, or AI-Enabled

14. Click **any account to open the right side panel:** the full prompt that ran, the detailed evidence and reasoning, and when it was last updated. {% hint style="info" %} **This persists - o**nce a list is enriched, the reasoning stays attached to every account in it. You don't lose it, and you don't have to re-run anything just to look at it again later. {% endhint %}

*** ### Other ways of adding tech indicators You don't need to go through the Tech Indicators card to do this. It works just as well straight from your Accounts list: {% stepper %} {% step %} ### From Agents * Select **Agents** from the left navigation menu * Select **Enrich existing accounts** tab > from the dropdown, **choose the Segment / List you want to enrich** * From the top-left corner of this chat box, toggle between **Reo data** or **Public data** * In the chat box, type your question, or pick one of the suggested prompts * For GPU Compute Intensity, try 'Find companies running large-scale GPU infrastructure' * For AI Nativeness, try 'Find companies training their own AI, not just using an API' * It runs on the same reasoning engine behind the standard tiers, just tuned to your exact question

{% endstep %} {% step %} ### From Accounts * From left navigation menu, select Accounts and click **Add Column** * Alternatively, from Segments or Lists, open the segment or list you want to enrich > click **Add Column**

* Toggle between **Reo data** or **Public data**, then type your question, or pick one of the suggested prompts * For GPU Compute Intensity, try 'Find companies spending heavily on GPU infrastructure' * For AI Nativeness, try 'Find companies training their own AI, not just using an API'

* Reo returns a preview: each account gets its tier and a **View reasoning** link that expands right there in the row, showing the specific evidence behind that output.

* Expand **Agent details** to see the exact prompt Reo is running, plus an agent name and tag it generates for you, for example 'GPU Infrastructure Scout' tagged 'Heavy Gpu Infrastructure'

* If the read isn't quite right, click **Chat with agent** to refine it in conversation before you commit to anything * Set **Re-run this enrichment** to a cadence like monthly, and check **Enrich new accounts added to this segment** so accounts added later get scored automatically * Click **Add column** > provide the **column name and Agent tag** to proceed

{% endstep %} {% endstepper %} *** ### How this can actually help you

If you sell…	Look for	What this tells you
GPU capacity or inference optimization	High GPU Compute Intensity	• You're looking at a company already running training or inference workloads + spending money on the problem you solve for • If it's not your infrastructure yet, this is your opening to become their next vendor
LLM cost management or API routing	Low GPU Compute Intensity + AI-Native	• You're looking at a company paying full price per token, with no infrastructure of its own to negotiate leverage • Consider this your Tier 1 ICP account
Foundation models or model API access	Low or Medium GPU Compute Intensity + AI-Native or AI-Powered	• You're looking at a company that's AI-first but hasn't built its own infrastructure yet • They're already paying someone for model access. If it's not you, this is your chance before they build in-house
Vector databases or RAG infrastructure	AI-Native or AI-Powered, especially with Medium or High GPU Compute Intensity	• You're looking at a company whose product depends on retrieving and searching data to feed a model • That's exactly the architecture vector databases are built for
AI agent frameworks or developer platforms	AI-Native with Low or Medium GPU Compute Intensity	• You're looking at a company building its core product around AI agents or automated workflows • They often run mostly on model APIs today, but their roadmap depends on tools like yours
Fine-tuning or model training infrastructure	High GPU Compute Intensity + AI-Native or AI-Powered	• You're looking at a company training or fine-tuning models themselves, not just calling an API • That's exactly the workload your infrastructure is built for
AI governance, security, or red-teaming	High or Medium GPU Compute Intensity + AI-Native or AI-Powered	• You're looking at a company running AI in production at real scale • That's exactly where hallucinations, prompt injection, and compliance risk start to matter
AI observability or LLMOps tooling	High or Medium GPU Compute Intensity + AI-Native or AI-Powered	• You're looking at a company running enough models in production to need real visibility into performance and failures • That's a need a single API call doesn't create
AI enablement or developer tooling for teams just starting out	AI-Enabled with growing (Medium) GPU Compute Intensity	• You're looking at a company in the early stages of building real AI capability • They haven't settled on an architecture or a vendor yet, that's your window

--- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.reo.dev/agents/tech-projects/gpu-compute-intensity.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.