Lead processing · Production

Kindora uses Claude to power AI-driven funder prospecting for small nonprofits

The problem

Small nonprofits lack affordable, accurate tools to identify which funders to approach; existing platforms are expensive and return thousands of unfiltered matches that overwhelm small teams with no capacity to evaluate them.

First attempt

A platform costing $4,000 returned 3,000 funder matches with no meaningful filtering, providing no actionable signal for a small nonprofit.

Workflow diagram · grounded in source
1
Nonprofit accesses prospecting via MCP
trigger
“Kindora's MCP connector lets nonprofits access its prospecting tools directly within Claude”
2
Batch intent classification and match scoring
ai_action
“Haiku 4.5 handles the behind-the-scenes batch processing where speed and cost matter: classifying user intent, scoring thousands of matches during discovery, and powering the free-tier public grant critique tool”
3
Agentic deep-search refines funder matches
ai_action
“drives the agentic deep-search feature that autonomously scores new funder matches”
4
Deep-research briefs generated
ai_action
“generates deep-research briefs on prospective funders”
5
Funder outreach written
output
“Claude Sonnet 4.6 powers the AI assistant, writes funder outreach”
6
Voice pitch practice tool
ai_action
“a working interactive voice tool that acts as a program officer, asks challenging questions, and then uses Claude to analyze the transcript against a nonprofit sales training curriculum”
Reported outcome

Kindora's AI prospecting eliminated 90% of 3,000 funder matches, narrowing the list to about 75 worth pursuing; applying to all of them landed eight grants totaling $100,000 in the first year.
The platform grew to 328 nonprofits with monthly signups roughly doubling, and a separate donor engagement pilot raised $101,000 in a single month.

Reported metrics
Funder matches eliminated as poor fits90%
Funders worth pursuing after filteringabout 75
Grants landed in first yeareight
Grant funding raised in first year$100,000
Show all 12 reported metrics
funder matches eliminated as poor fits90%
funders worth pursuing after filteringabout 75
grants landed in first yeareight
grant funding raised in first year$100,000
nonprofits signed up in first week of beta50
nonprofits joined by end of beta period200
total nonprofits on platform328
monthly signup growth rateroughly doubling from January through March 2026
development velocity increase with Claude Code10 times more code than before, with fewer errors
SAFE Note investment received from Camelback$50,000
donor campaign raised in single month$101,000
cost of failed prospecting platform$4,000
Reported stack
Claude Sonnet 4.6Haiku 4.5Opus 4.6Claude CodePythonMCP connector
Source
https://www.anthropic.com/customers/kindora
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Kindora's AI prospecting eliminated 90% of 3,000 funder matches, narrowing the list to about 75 worth pursuing; applying to all of them landed eight grants totaling $100,000 in the first year.

What tools did this team use?

Claude Sonnet 4.6, Haiku 4.5, Opus 4.6, Claude Code, Python, MCP connector.

What results were reported?

Funder matches eliminated as poor fits: 90%; Funders worth pursuing after filtering: about 75; Grants landed in first year: eight; Grant funding raised in first year: $100,000 (source-reported, not independently verified).

What failed first in this deployment?

A platform costing $4,000 returned 3,000 funder matches with no meaningful filtering, providing no actionable signal for a small nonprofit.

How is this lead processing AI workflow structured?

Nonprofit accesses prospecting via MCP → Batch intent classification and match scoring → Agentic deep-search refines funder matches → Deep-research briefs generated → Funder outreach written → Voice pitch practice tool.

Related lead processing cases