Integrates with

GreenhouseAshbyWorkableLeverRecruiterFlow

Work samples for any toolset, backed by explanations.

Screenshares collected by context-aware AI that only listens, inquires and records.
Humans judge the result.

JournalistSubstack, Google Docs, Otter.ai, Notion
Financial AnalystExcel, Bloomberg, Python, Tableau
Software DeveloperClaude Code, Cursor, GitHub, Vercel
Marketing ManagerBuffer, HubSpot, Google Analytics, Mailchimp
Graphic DesignerCanva, Figma, Adobe CC, Midjourney
Product ManagerLinear, Jira, Amplitude, Miro
Data ScientistJupyter, Snowflake, dbt, SQL

Next-Level Takehomes

Candidate-first terms to retain the best.

Paid

Automatic honorariums

Every completed work sample is paid out automatically via Stripe, and candidates get their own dashboard to track earnings, payout status, and history in real time.

Asynchronous

Work when it's convenient

Candidates set their own deadline and work on their own schedule, turning a self-imposed deadline into a chance to show off their reliability and follow-through.

Respectful

No surveillance

No keystroke logging, webcam monitoring, or background tracking, just a retrospective screenshare session with improv explanations.

Native Tools

Bring your own toolset

Candidates work in whatever they already know, like Claude Code, Notion, GitHub, or Figma, with no unfamiliar sandboxed tools to learn first.

Ownership

Keep what you make

Candidates retain full ownership of every work sample they produce; nothing is claimed by the hiring team, whether or not they get the job.

Constructive Assessment

Realistic take-homes, collected by AI that asks hard questions.

Candidate

As you can see here in my mockup, [uh] I made the settings intuitive, with [um] tooltips, these awesome sliders and descriptions.

Pudding Agent

I see that. It looks great! I see you chose that framework. Now, how did you decide to use this framework for this audience? Please explain your thought process.

Retrospectives

Explained work

Pudding agents watch the screenshare live and ask sharp inline questions about the decisions behind the work. Real-time probing of reasoning, not a scripted checklist.

CONTEXT AWARE

An agent that knows your stack

Frontier model AI that knows exactly what to ask in realtime.

On Demand

Start with one click

No scheduling back-and-forth. Candidates decide exactly when they're ready and open a live screenshare session with the AI agent at the click of a button.

Every engagement is scoped by the hiring team, never pulled from a generic vendor library.

Deliverable + screenshare

Submit an asset, then explain it live.

Screenshare only

The live walkthrough is the deliverable.

Anonymize deliverables

Job relevant

Hirer-defined engagements

No generic prompts pulled from a vendor library: hiring teams write the brief, choose deliverable-plus-explanation or explanation-only, and can anonymize submissions before review.

What everyone is saying

How modern hirers approach take-homes.

Modernize your hiring approach today and engage your best candidates with paid, realistic, explained take-homes.