# How Do I Reach AI Teams the Week They Ship a Model?

> A GTM play for Revnu, for data labeling, evaluation and AI infrastructure companies. A team that just shipped a model is already planning the next one and knows where it fell short. Revnu finds new model releases, reads the release notes and benchmarks, and reaches the research or data lead by email and LinkedIn.

Canonical: https://revnu.com/docs/play-ai-model-launches
Last updated: 2026-11-08

The week a team ships a model, it publishes what the model can do, where it falls short and which benchmarks it chased. That release note is a map of their next training run. Revnu finds new model releases in your market, reads what each team published, and reaches the research or data lead with an offer built on their own results.

## Why it works

A shipped model means the team trains its own models, has compute budget and has a roadmap. Its release notes name the weak spots, such as long-context reasoning, a language or a domain, and every weak spot needs data. An email that says "your model card shows X, here is a pilot set for X" is about their work, not your company.

## How to run it

1. **Name the launches you care about.** Open-weight releases, new versions from AI companies, fine-tuned domain models, or models in one field such as code, legal or medical.
2. **Find the releases.** Revnu finds new releases on model hubs, company blogs, research papers and launch posts, each with its date.
3. **Read the release.** Revnu notes what the model is for, the benchmarks reported and any gaps or next steps the team names.
4. **Find the buyer.** The head of data, research lead or post-training lead, or the founder or CTO at a smaller company, with a verified work email.
5. **Write by email and LinkedIn.** The email congratulates them on the release with one specific result, names the gap and offers a pilot batch for it. A LinkedIn connection request follows, then a follow-up with a sample.
6. **Run it weekly.** New models ship every week.

## The prompt

> Find AI companies and labs that released a new model in the last 30 days, in [domain or kind of model]. Read each release note or model card and note the benchmarks they report and any gaps or next steps they name. Find the head of data, research lead or CTO and their work email. Write an email that mentions one result from the release, names the gap and offers a pilot [data type] batch for it, then connect on LinkedIn three days later.

[Try this in Revnu](https://revnu.com/sign-up?prompt=Find%20AI%20companies%20and%20labs%20that%20released%20a%20new%20model%20in%20the%20last%2030%20days%2C%20in%20%5Bdomain%20or%20kind%20of%20model%5D.%20Read%20each%20release%20note%20or%20model%20card%20and%20note%20the%20benchmarks%20they%20report%20and%20any%20gaps%20or%20next%20steps%20they%20name.%20Find%20the%20head%20of%20data%2C%20research%20lead%20or%20CTO%20and%20their%20work%20email.%20Write%20an%20email%20that%20mentions%20one%20result%20from%20the%20release%2C%20names%20the%20gap%20and%20offers%20a%20pilot%20%5Bdata%20type%5D%20batch%20for%20it%2C%20then%20connect%20on%20LinkedIn%20three%20days%20later.)

Each lead keeps the release on its **Signals** tab with its date and link. Your first sequence waits for your approval in **Review**.

## Run it every week

> Every Monday, find models released in the last week in [domain], read the release notes, and add the teams that fit our buyer to the Model Launches campaign.

Set it up once as a [workflow](https://revnu.com/docs/recurring-work.md).

## Related

[Eval engineer hiring play](https://revnu.com/docs/play-eval-engineer-hiring.md), [fresh funding play](https://revnu.com/docs/play-fresh-funding.md), [find leads by buying signal](https://revnu.com/docs/uc-buying-signals.md), [Revnu for data labeling](https://revnu.com/for/data-labeling) and [all GTM plays](https://revnu.com/docs/play-overview.md).

## Frequently asked questions

### Which releases are worth reaching out about?

Teams that train or fine-tune their own models: open-weight releases, new versions from AI companies and domain models. A wrapper app built on someone else's model rarely buys training data.

### Who should the email go to?

The person who owns the data for the next run: head of data, post-training lead or research lead at a lab, and the CTO or founder at a smaller company.

### Isn't the week of a launch too busy?

It's the week they're reading feedback and planning the next run. Keep the email short and about their result, and Revnu follows up a week later for anyone who didn't reply.

### Can I combine this with funding?

Yes. "Shipped a model and raised in the last six months" gives the shortest, warmest list. Say both in one message.