# Overview Source: https://docs.havilo.ai/concepts The most valuable fact in a network is rarely stored in one place. It may begin in a meeting, reappear in an email, matter because of a relationship, and become useful only when the right question is asked. Havilo is a system for understanding and acting through networks. It brings the people, organizations, relationships, knowledge, and ongoing work of a team into one connected system. Havilo uses that context to answer questions, identify opportunities, prepare work, and improve through every interaction. Sometimes that begins with a question. Sometimes Havilo notices the change first. ## A flexible data model Every team sees its network differently. Each workspace defines the kinds of records it needs, the fields that describe them, and the relationships that connect them. A record can represent a person, company, fund, program, project, or something specific to the way the team operates. Together, these provide the familiar structure of a customizable CRM without forcing the team into a fixed schema. Relationships carry their own context. A person can advise a company, work at a fund, mentor a founder, or participate in a program, with roles and dates that preserve how the connection changed over time. Lists, views, boards, and folders organize those records for different work without creating a separate database for each process. The [data model](https://docs.havilo.ai/concepts/entities.md) also extends beyond conventional CRM records. The same graph can hold private knowledge, imported data, public research, meeting history, ongoing work, and the outputs Havilo produces from them. ## Provenance and history [Provenance and history](https://docs.havilo.ai/concepts/provenance-history.md) keep the origin of information connected to the information itself. An imported value can retain its file and import history. A public fact can retain the page or passage that supports it. Mail, calendar events, and field memos remain available as versioned source material rather than being reduced to unattributed text. Grounded enrichment records how a value was derived, when it was observed, what source supported it, and whether it replaced an existing value. Identity decisions retain the searches, evidence, and reasoning used to reach them. Unsupported claims can be rejected without entering the record. New information does not have to erase what came before. Previous values and source revisions can be preserved for review. Automated identity changes retain enough history to restore the prior state; if someone has since edited the record, Havilo refuses to overwrite the newer work. ## Identity and enrichment [Identity and enrichment](https://docs.havilo.ai/concepts/identity-enrichment.md) may begin with an email address, a calendar attendee, a business card, or a row in an import. Havilo resolves those signals against the existing network so the same person does not become a new record every time they appear in a different system. When a record is incomplete, Havilo can use the context already available—such as a name, email domain, employer, school, program, or known relationship—to search for a public identity. Candidate matches are evaluated against cited evidence. If the evidence is inconclusive, Havilo can hold the decision and reconsider it when another signal arrives. Once an identity is known, enrichment can add and refresh information from authenticated LinkedIn profiles, Crunchbase, company websites, and other sources. It can connect people to their organizations and roles, create supporting records and relationships, and regenerate dossiers as the underlying evidence changes. Routine refreshes run in the background, while work triggered by a meeting or direct request can take priority. ## Reasoning across your network Consider a request that crosses the network: > Plan an event for our startups with Turkish founders, ideally founders who have had a previous exit. Identify the problems those startups share, find a few mentors who can help, use my Event Invite template to prepare invitations for everyone, and prepare the LP update for our Quarterly Founder Outreach metrics. Even the first clause carries more meaning than a normal filter can express. “Turkish founder” describes a person, not automatically the company they founded. “Ideally” makes a previous exit a preference rather than a requirement. Finding mentors depends on understanding the startups' common problems and the experience that would actually help. [Reasoning across your network](https://docs.havilo.ai/concepts/asking.md) can draw on structured facts, relationships, Notes, correspondence, meetings, public research, uploaded material, templates, and prior work, then carry the result into invitations and reporting. The person asking does not have to translate the request into filters or coordinate each part as a separate prompt. ## Shared context [Notes](https://docs.havilo.ai/concepts/notes.md) add firsthand context. Introductions and relationships record who became connected. Meetings add attendees, transcripts, decisions, and follow-up. Enrichment refreshes public evidence. Saved [artifacts](https://docs.havilo.ai/concepts/artifacts.md) preserve useful analysis and work product. Each one contributes to the same context. The next search, match, briefing, or piece of outreach can use what the team learned from the work before it. Havilo does not reset to the original import after each task. Its understanding changes as people use the system and as new evidence arrives. ## Proactive work Some work should not wait for another question. Havilo can notice relevant changes in connected mail, calendar events, meetings, and field memos; interpret them against what the team already knows; prepare follow-up; and surface work in Home when a decision or approval can move it forward. Each open commitment it finds this way becomes a [Situation](https://docs.havilo.ai/concepts/situations.md): a live assessment of what is outstanding and who owes whom what, filed by whether it needs you, is waiting on someone else, or has gone quiet. Situations are not tickets and there is no command that cancels or closes one. A matter moves when the underlying action happens in the mailbox, or when someone records testimony — what they know that the record does not, such as work they finished over the phone. Testimony is weighed against everything else the matter knows rather than obeyed, so a person is allowed to be wrong. ## Havilo applications Home, Mail, Meetings, Raise, Data, Entities, Lists, and Artifacts each offer a different way into Havilo. They read and write the same network. Someone discovered through a calendar invite remains the same person when they appear in a search, join a list, receive an introduction, or become part of a later briefing. The concepts below explain each part of that shared system. ### The network * [Data model](https://docs.havilo.ai/concepts/entities.md): Entities, entity types, fields, and relationships. * [Identity and enrichment](https://docs.havilo.ai/concepts/identity-enrichment.md): Signals become grounded, connected entities. * [Provenance and history](https://docs.havilo.ai/concepts/provenance-history.md): Sources, decisions, changes, and reversibility. * [Notes](https://docs.havilo.ai/concepts/notes.md): Written knowledge becomes searchable context. ### Work and access * [Reasoning across your network](https://docs.havilo.ai/concepts/asking.md): Complex objectives across the connected system. * [Situations](https://docs.havilo.ai/concepts/situations.md): Open commitments Havilo tracks, and how you close them. * [Artifacts](https://docs.havilo.ai/concepts/artifacts.md): Tables, documents, email drafts, and network graphs. * [Entity Lists](https://docs.havilo.ai/concepts/lists.md): Cohorts, pipelines, events, and other working sets. * [Workspaces and access](https://docs.havilo.ai/concepts/access.md): Boundaries, roles, permissions, and sharing.