Launching Nimble models: #1 decision models outperforming Jev. Read more.

Data Acquisition

Turn your codebase into AI training data, and get [paid] for it.

Frontier AI labs need more than finished code. They need to see how it was actually built: the Slack threads, Notion docs, Jira and Linear tickets, and the decisions in between. We package your codebase and its full operating history into structured training data, and turn it into a payout for your startup.

Get in touch See if you qualify

What We Do

In plain terms: we turn how you built your product into training data for AI models.

You get paid for engineering work you already did. Your product keeps running exactly as it does today, and nothing moves without a scoping conversation first.

  1. 01

    Source and scope

    We identify the assets together: codebase, Slack history, Jira/Linear tickets, Notion docs. You decide what's in scope before anything changes hands.

  2. 02

    Package and structure

    We turn it into structured training environments frontier labs can actually use, without extra engineering work on your side.

  3. 03

    Payout and handoff

    A single payout for assets you already created, as a by-product of simply building your product.

What We're Looking For

Three-plus years in. Real product. Real complexity. We want to talk.

We're not chasing every startup. We're looking for technical teams who've spent real time building something with actual depth: multiple features working together, real edge cases, real decisions behind the architecture. If that's you, your operating history is probably worth more than you think.

[Code Data]

[Contextual Data]

Your codebase itself: the commits, branches, and pull requests behind what shipped.

Everything that shaped it along the way: Slack, Jira, Linear, Figma, and more.

What We Work From

One connected operating record, not a single file drop.

Pick a category to see exactly what we mean by it, and what's always off the table.

In scope

Codebase & commit history

The full repository, not just a snapshot: every commit, branch, and pull request that shows how the product actually evolved.

  • GitHub
  • GitLab
  • Bitbucket
  • Full commit history
  • Pull request history

In scope

Team communication

The engineering channels where the real decisions happened: debates, trade-offs, and the reasoning that never made it into a commit message.

  • Slack
  • Microsoft Teams
  • Engineering channels only

In scope

Docs & knowledge bases

Specs, postmortems, onboarding docs, and the institutional knowledge that explains why the system looks the way it does.

  • Notion
  • Confluence
  • Google Docs
  • Internal wikis

In scope

Project & ticket tracking

Roadmaps, sprints, and the tickets that got reprioritized twice before anything shipped: the working history behind the backlog.

  • Jira
  • Linear
  • Asana

In scope

Architecture & design records

RFCs, ADRs, and design docs that capture the trade-offs behind major technical decisions, plus longer-range technical roadmaps.

  • RFCs
  • ADRs
  • Design docs
  • Technical roadmaps

In scope

Design files

The design decisions behind the interface: flows, component libraries, and the iterations that got cut before launch.

  • Figma
  • Design system files

Always excluded

Customer & financial data

This never enters scope, regardless of company or domain. See the full data & privacy policy below for the complete list.

  • Customer PII
  • Financial accounts
  • Credentials & secrets
  • HR & payroll records

Data & Privacy

What we do, and don't, touch.

Acquisition scope focuses on engineering and product materials. Do not include personal data, customer records, employee-sensitive information or credentials in materials you propose to share. Business contact details used to handle inquiries are separate from acquisition materials. Any later transfer requires separate scoping and an agreement.

In scope

  • Source code and commit history
  • Engineering Slack/Teams channels
  • Notion, Confluence, and product docs
  • Jira/Linear tickets and roadmaps
  • Figma and other design files
  • Architecture and design decisions

Excluded from acquisition materials

  • Personally identifiable information (PII) of any kind
  • Customer data and customer records
  • Employee sensitive information (HR files, payroll, performance reviews)
  • Credentials, secrets, and financial account details
  • Anything you flag as sensitive during scoping

VCs, Lawyers & Advisors

Referring a founder is easy.

If you work with founders, as an investor, counsel, or consultant, and think a portfolio company or client is a fit, we'd welcome a warm introduction. The process is confidential, moves at the founder's pace, and we're happy to keep you looped in as much or as little as you'd like.

Refer a founder

Questions & Answers

The things people ask first.

What kind of data are you looking for?

Your codebase and commit history, plus the operating record around it: Slack or Teams channels, Notion/Confluence docs, Jira/Linear tickets, and architecture or design decisions. See "What We Work From" above for the full breakdown.

Do I need to be shutting down or selling my company?

No. This is an acquisition of your codebase and operating record as a standalone data asset, not an acquisition of your company. Your product keeps running exactly as it does today, on your infrastructure, under your control.

Is my source code required, or just the surrounding context?

We're interested in both, but the surrounding context (the Slack threads, tickets, and docs) is what makes a submission stand out. A codebase on its own is useful; a codebase with its full operating history is significantly more valuable.

How is the payout determined?

It comes down to a few things: how long you've been operating, how complex the product is, and how complete your contextual artifacts are (Slack, tickets, docs, and the rest). We weigh all of it together and come back with a single valuation and one contract after scoping, not a recurring arrangement.

How do you handle sensitive or personal data?

PII, customer data, credentials, and financial or HR records are always out of scope, full stop. You define exactly what's in and out before anything moves, and anything you flag as sensitive during scoping stays excluded. See "Data & Privacy" above for the complete list.

How long does the process take?

It starts with a short scoping conversation, followed by a review of a representative sample of your artifacts, then a single valuation and contract. Most engagements move in a matter of weeks, not months.

What if my company doesn't check every box?

Apply anyway. The rough bar is a real tech company, three-plus years operating, and a product with genuine complexity behind it, but we'd rather hear about your specific situation than have you self-select out.

Can VCs, lawyers, or consultants refer a founder?

Yes. If you work with founders and think a portfolio company or client is a fit, we welcome a warm introduction. The process is confidential and moves at the founder’s pace.

Get in Touch

Tell us about your company, or the one you're referring.

Online applications are currently unavailable. Email our team to arrange an introductory conversation.

Email company@bespokelabs.ai