Find the right person.
People describe a real need in their own words. The search view translates that request into relevant people and services while keeping the social path behind each result available for inspection.
A Softmachines work · Trust networks
A good word, made practical.
GiraVoce gives the trust already living between people a way to travel without losing its origin.
The premise
Trust cannot be bought.
When someone needs a capable professional, the useful question is rarely who advertised best. It is who has already done good work for someone you know.
GiraVoce turns that fragile, spoken knowledge into a consultable network. Every recommendation retains the person and relationship behind it, making trust easier to exercise rather than pretending to automate it.
How it works
Relationships stay in the picture.
See where trust comes from.
The network makes a recommendation’s path visible. A person can assess the relationships behind a suggestion and decide with more context, not less.
Let help meet everyday life.
A search or recommendation can begin inside WhatsApp, where people already ask friends for help. The interface shows the assistant guiding the exchange without pretending to be the source of trust.
Turn an introduction into practical help.
Availability and appointments stay connected to the trusted introduction that began the exchange. The calendar removes coordination friction without flattening the relationship into an anonymous transaction.
AI in the network
Understand the need. Preserve the person behind the answer.
GiraVoce uses intelligence to interpret ordinary language, organize information, and keep the network available in conversation. It never uses a model to decide who deserves trust.
Turn free-form experience into a useful passport.
Professionals describe their work as people naturally do: in services, examples, qualifications, places, and availability. AI helps structure that material into a searchable professional passport without replacing the professional’s own account or inventing missing credentials.
Semantic matching then connects a request such as “I need a caregiver for my mother” with the profiles and services whose meaning actually aligns. The result remains anchored to visible recommendations and relationships, so relevance does not become an invisible trust score.
Let the professional define the agent’s role.
A professional can specify what their agent knows, how it introduces the work, which questions it may answer, and when the conversation should move back to a person. Conversational memory keeps an exchange coherent across requests without giving the agent authority to accept work or create a recommendation.
This is presence without impersonation. The agent handles repetitive first contact and scheduling while the professional is busy, but the choices that create obligation or trust remain unmistakably human.
Intelligence helps the good word travel.
GiraVoce uses semantic understanding to connect an ordinary sentence, such as a leaking tap or the need for a caregiver, with the right profiles, services, and offers. It reduces the friction of describing a need and organizing professional information that was never written for a database.
The system is deliberately unable to purchase credibility or manufacture a recommendation. Invitation, visible origin, and human choice keep the network accountable. AI makes the relationships more useful and available; it does not become the relationship.
Over time, that creates a different kind of marketplace: good work travels through communities because somebody can stand behind it, and people face important everyday needs with more context and less uncertainty.
AI can clarify and connect. It cannot invent trust.
Return to works