GTM Desk

Find AI companies to investigate as prospects

Compare funding, team growth and technical hiring, then inspect the people and records behind each company.

What this demo does

For a compute provider: build an evidence-backed account shortlist, then qualify each company before approaching it.

The data

Thirty-five saved companies across generative AI, computer vision and speech/voice. Funding, growth, active indexed jobs, news and recorded people come from Dealroom.

What it does not tell you

Hiring matches are job-title keyword matches, not proof of compute demand. The demo does not know GPU spend, cloud vendor, purchase intent or buying authority. No emails or phone numbers are inferred.

Build your own

Copy this into a coding assistant that can build apps. Adapt the brief before you start. You will need your own Dealroom API access for live data; never paste a secret into the prompt.

Read the build prompt

Build GTM Desk for an AI compute provider researching potential customers. Start with a bounded sample of operational, non-exited AI companies with 10-1,500 employees and $1M-$1B total funding, across generative AI, computer vision and speech/voice. Exclude chip manufacturers and clearly disclose the selection.

Make an interactive prospect map: x-axis latest dated funding month, y-axis annual employee growth, bubble area proportional to the latest round amount in USD. Keep companies with missing plot values in the list. Do not replace a recent undisclosed round with an older disclosed round. Label any compressed growth scale.

Add segment, country, team-size, recent-funding and technical-hiring filters. Clicking a company opens funding sources, active indexed roles, news, and recorded technical leaders or founders with profile links. Explain the job-title keyword rules. Allow browser-local saved accounts and safe CSV export. Never infer emails, phone numbers, GPU spend, cloud provider, buying authority or intent. Do not send outreach.

First read the Dealroom Premium API documentation at https://docs.dealroom.co/docs/premium-api. Discover supported filters and resolve taxonomy IDs rather than guessing them. Check current data availability before committing to a feature.

Use my own Dealroom Programmatic (M2M) credentials through a server-only environment file excluded from Git. Never ask me to paste a secret into chat or embed credentials in frontend code. Implement the documented OAuth2 client-credentials flow, cache tokens, refresh once on expiry, and send Authorization and X-Client-Id. Respect access limits, pagination and Retry-After. Do not silently treat a failed or restricted response as an empty dataset.

Build a working responsive web app, not a mockup. Start with a bounded cohort and cache responses with retrieval dates. Show currency, sources, missing values and partial coverage. Keep a clearly labelled saved-data mode for a public demo. Include loading, empty and error states, accessible keyboard alternatives, tests for the data joins and a README explaining setup. Do not publish, send messages or write to Dealroom without asking me.

Reference demo: https://dealroom.co/api/showcase/gtm-desk/
Build guide: https://dealroom.co/cookbooks/gtm-desk/

How it works

These are the data joins behind the app. Confirm current routes, filters and response shapes in the API documentation before implementing them.

Build a bounded account cohort

Resolve tags, discover supported filters and document the selection. Deduplicate companies that match several AI segments.

GET /reference/filters?scope=companies
GET /reference/filters/tag_id/values
GET /data/companies

Join evidence by company UUID

Retrieve dated rounds, team members, active job records and news. Retain source links and mark capped or incomplete collections.

GET /data/companies/{id}/funding-rounds
GET /data/companies/{id}/team
GET /data/jobs
GET /data/news
GET /data/entities/{id}

Plot without inventing values

Use month-level dates, growth and funding amount. Bubble area, not radius, represents amount. Keep missing data in the table and make chart selection open the same account record.

Make qualification possible

Show recorded roles and profile links with the underlying evidence. A saved prospect is a research choice, not a confirmed buyer or a contact with verified purchasing authority.

Connect safely

Use Programmatic (M2M) credentials in a server-only environment file excluded from Git. The documented OAuth flow returns a token; requests need both Authorization and X-Client-Id. Cache responses and tokens, handle pagination and rate limits, and display retrieval dates and incomplete coverage.

A public version can serve a reviewed saved snapshot. A live shared workspace also needs authentication, storage and controls for who can see personal notes or export records.

Read the API documentation →