AI answers, source attribution & the knowledge graph — explained

Kindora generates AI answers by reasoning over your organization's structured knowledge graph — donors, grants, programs, volunteers, and finances — built from the records already in your Kindora workspace. Rather than guessing from similarity, Kindora traces relationships across those records and returns an answer with the specific sources behind it.

You can ask operational questions like 'Which donors gave more than $10,000 in the last quarter?', 'Which grants are expiring this quarter?', or 'How many volunteer hours did each program log in Q2?'. Kindora answers across donor CRM, grants, case management, accounting, and volunteer data using multi-hop reasoning over the knowledge graph.

Kindora is designed so answers are checked by evidence, not asserted. Every answer is returned with verifiable citations and a reasoning path, so you can open the exact source records before acting. If the knowledge graph cannot find sufficient evidence, Kindora says so instead of guessing.

Source attribution means every AI answer in Kindora is returned with the exact records it used as evidence — the donor profile, grant file, program report, or expense entry that supports the result. You can jump straight from any answer to its underlying source to confirm the information.

Yes. Each AI answer links to the source records and shows a reasoning path, so your team can open the original donor, grant, program, or finance record that produced the answer. This turns AI results into auditable, defensible outputs for boards, funders, and compliance.

When every metric, narrative, and forecast in a donor report or board deck can be traced back to a source record, reporting is easier to review and approve. Source attribution gives your team confidence that AI-assisted reports are grounded in real organizational data, not assumptions.

VeritasGraph is an open-source GraphRAG knowledge-graph framework developed by Bibin Prathap. Kindora uses VeritasGraph to structure your records into a knowledge graph, reason across connected entities, and return every AI answer with a verifiable citation and reasoning path.

Yes. VeritasGraph is open source on GitHub at https://github.com/bibinprathap/VeritasGraph, with documentation at https://bibinprathap.github.io/VeritasGraph/index.html. Because it is open source, you can inspect exactly how indexing, graph construction, and source-attributed reasoning work.

A plain chatbot or basic retrieval approach (RAG) often pulls similar-looking chunks and can produce unsupported answers. VeritasGraph reasons over a structured knowledge graph of entities and relationships, performing multi-hop reasoning and returning a citation and reasoning path with every answer.

VeritasGraph runs on-prem or 100% local, so record indexing, graph construction, and reasoning happen without data egress. This keeps AI search sovereign for nonprofits managing sensitive donor, beneficiary, and compliance data — nothing is sent to third-party AI providers.

Kindora verifies answers by grounding them in the knowledge graph: the reasoning path must resolve to real, existing records. Each answer is delivered with citations to those records so a team member can confirm correctness against the source before relying on the result.

Every answer in Kindora includes an expandable reasoning path that shows the steps and the records used — the specific donors, grants, programs, or entries that led to the result. You can follow the path from question to evidence to answer.

If the knowledge graph cannot find sufficient supporting evidence, Kindora flags the answer as unsupported or asks for clarification instead of producing a confident guess. This prevents fabricated answers and keeps AI outputs trustworthy.

A knowledge graph organizes entities — donors, grants, programs, volunteers, and finances — as connected nodes rather than isolated records. Kindora uses a knowledge graph so AI can perform multi-hop reasoning across those relationships and explain the path it took to reach each answer.

Kindora indexes the records in your workspace — donor profiles, grant files, program outcomes, volunteer hours, and expense entries — into entities and relationships, then connects them into a structured knowledge graph managed by VeritasGraph. New records update the graph so answers stay current.

Keyword search finds pages that contain the words you typed. A knowledge graph understands meaning and relationships, so Kindora can answer complex questions like 'Which grants funded programs in more than three states?' by reasoning across connected entities rather than matching keywords.