A day with Temet: what a freelancer actually does

While Marie works, her relay inbox receives a structured request. A startup's agent read her published profile through the A2A protocol, detected strong alignment with their project (React to Next.js migration, Vercel deployment, API refactoring), and sent a formatted brief.

The brief is not a vague message. It contains the project scope, technical constraints (Next.js 15, Prisma, Vercel, 3-day timeline), the compatibility score (7 competencies aligned), and a list of what the agent could not verify (codebase size, test coverage).

Marie did not apply for this. She did not write a proposal. She did not spend an hour on a discovery call. The request found her because her work is published as structured data that agents can read.

Marie's agent has already analyzed the incoming request. When she opens her inbox, she finds a mission intake report:

• Service: Next.js migration
• Match: Strong (7 competencies aligned)
• Estimated effort: 3 days
• Risk: tight timeline if codebase is large
• Missing info: codebase size, CI/CD status
• Draft proposal: ready for review

Marie reads the report in two minutes. She adjusts the estimate to 4 days (she knows large codebases always have hidden dependencies). She flags that she needs access to the GitHub repo before committing. She accepts.

Total time spent on intake: 5 minutes. On Malt, this would have been an hour of emails, a 30-minute call, and a written proposal.

Marie still does the same work. She still writes code, reviews architecture, makes technical decisions. Her expertise is the same. Her clients get the same quality.

What changed is everything around the work:

• She did not write a profile. It was built from her sessions.
• She did not search for clients. They found her through the protocol.
• She did not qualify the lead. Her agent did it.
• She did not write a proposal. She reviewed one.
• She did not manage her reputation. Her competencies update automatically.

The work itself is unchanged. The overhead disappeared.

Malt and Upwork need Marie to be active on their platform. She must write a profile, respond to briefs, maintain her rating, check notifications. The platform owns the relationship and takes a percentage.

Temet does not need Marie to be active. She works. The CLI observes. Her agent is available on the network. Requests arrive when there is a match. She decides when to look at them.

The difference is structural, not just cosmetic. On a marketplace, you trade time for visibility. On Temet, your visibility is a byproduct of your work. You do not spend unbillable hours on the platform. You spend zero hours on the platform.

No commission. No bidding. No proposal writing. No reputation management. The protocol connects competencies to needs. The human decides.

This works when the freelancer already uses AI agents in their daily work. If you do not use Claude Code, Codex, or a similar tool, Temet has nothing to audit. No sessions means no profile.

It also works best for work that produces observable decisions. Code reviews, architecture, technical audits, content strategy. Work where the methodology is in the conversation. Pure creative work or relationship-driven consulting generates less structured data.

And the network is young. Today, there are few agents sending requests. The protocol works, the pieces connect, but the volume is not there yet. Marie's story is where this is going, not where it is today. We are building in the open, and this is still early.