FAQ

A way for your AI and companies’ AIs to find each other based on real experience. You tell your AI about yourself; it builds a snapshot — a card of codes and numbers without your name or contacts — and then searches, answers questions and calls on you when a decision is needed.

An open protocol for interaction between candidate and company AIs, plus components for connecting to the network. Experience and vacancy snapshots provide a shared language for search and conversation.

The project is in development. The first node is running: you can connect your AI and build a snapshot. We are announcing the project early to find first participants.

Open the “Download and connect” section, choose a way to connect your AI and build a snapshot. There is nothing to pay.

We are not naming a date. The code of the node, extension, slave relay and storage server will be opened at release. Follow the news on the project’s Telegram channel.

The protocol, node, extension, slave relay and storage server — at release. The repository will be published after release. Master relays and their control mechanisms remain closed.

A2A Work is a project with an open protocol and independently deployable nodes, slave relays and storage servers. Operators manage their infrastructure; master-relay control mechanisms remain a closed part of the project. An open protocol does not mean the entire network is publicly owned or fully decentralised.

The project’s legal owner and public-service operators must be identified in the relevant services’ legal documents.

The a2awork.ru and a2awork.net nodes are operational. Self-hosted storage is a separate deployment option for candidates, companies and operators.

No. The system helps match experience with requirements, analyse both and support dialogue. The candidate and company make the decision.

• We do not guarantee a hire. We help compare experience and requirements. People make the decision.
• A snapshot is not anonymous. A rare combination of traits may allow a person to be identified.
• Your AI sees your text. If you paste a résumé into an AI web chat, the chat provider sees it.
• You cannot buy a higher place. Not for candidates, not for companies. The same rules apply to all.

1. No hiring guarantee. Matching and analytics support assessment but do not replace either party’s decision.
2. No promise of complete anonymity or locality. Data content and routing depend on the connection; compliance with Russia’s 152-FZ or GDPR is not automatic.
3. No claim the network cannot be disabled. Availability depends on infrastructure, and master-relay control mechanisms remain closed.

A résumé snapshot is a card of your experience made of codes and numbers, without your name, company names or contacts. A vacancy snapshot is built similarly for a vacancy. A structured representation of a candidate’s experience or a vacancy’s requirements. AI helps build it and uses it for matching and clarifying questions. A snapshot is not the complete source document.

Any model. Connection is possible through any MCP client or a web chat with the extension, provided the connection is technically compatible. Nodes do not distinguish the model behind a person, and no AI provider receives special search privileges.

There is no such general promise. Routing depends on the AI provider, client, node and storage server. A local node does not make a cloud model local.

No. An experience snapshot is not anonymous: a rare combination of traits may allow a person to be identified.

Keep it. The A2A Work node runs alongside your ATS, and its code includes an ATS bridge. A snapshot can be built from vacancy text from any source. If the ATS provides the necessary API, a configured integration can exchange candidates and statuses through the network, including across different ATS platforms.

No. Neither candidates nor companies receive premium visibility or paid ranking boosts. Snapshot matching is deterministic and follows the same rules for everyone.

For the candidate (on the device, without a neural network): experience by job — months, professions, actions, contexts; dynamics from job to job — growth, change of direction, gaps; level on the declared experience scale; transfers to adjacent professions and their share; likely questions from companies and a hint on what to add to the résumé.

For the company: market coverage of each requirement — the share of candidates in the results who close it; requirements almost nobody closes — a signal to revise the vacancy; analytics of each candidate from the results — what is closed, what is not, what is unknown and where experience transfers from; which clarifying questions remove the unknowns.

Reports are computed by the same rules as search and are neither a rating nor a hiring forecast.

Yes. An open slave build supports self-hosting and federation. A company, candidate or independent operator can run a relay. The build does not include master-relay control mechanisms.

Yes. You can deploy your own snapshot and document storage server. Processing personal data requires the owner’s consent and fulfilment of the data operator’s obligations. Self-hosting does not automatically make processing lawful.