diff --git a/ai/claude-code/plans/nlnet_commons_fund_proposal.summary.md b/ai/claude-code/plans/nlnet_commons_fund_proposal.summary.md
new file mode 100644
index 00000000..73c11389
--- /dev/null
+++ b/ai/claude-code/plans/nlnet_commons_fund_proposal.summary.md
@@ -0,0 +1,27 @@
+Structured to match NLNet's form fields:
+
+- Proposal name: "piker: sovereign, structured-concurrent trading infrastructure for the commons"
+- Abstract: comprehensive description of what piker is today + 5 concrete deliverables for the grant
+- Relevant experience: your background with tractor/trio/structured concurrency ecosystem
+- Comparison: positions piker against proprietary platforms, permissively-licensed FOSS libraries, and
+institutional systems — hits NLNet's "market failure" framing hard
+- Technical challenges: 5 real engineering problems (distributed SC, zero-copy shm, broker heterogeneity, UI
+ perf, reproducible packaging)
+- Ecosystem engagement: upstream contributions, Matrix community, pikers.dev self-hosting, open data
+formats, standards work
+- Budget: €50k breakdown across 7 task areas at €75/h
+
+Key arguments aligned with NLNet Commons Fund mission:
+
+1. Financial infrastructure as captured commons — the "market failure" framing NLNet explicitly looks for
+2. AGPLv3+ as the license that actually prevents enclosure (not MIT/Apache which they've seen fail)
+3. "Honest, open, inclusive, robust" — their exact language mapped onto piker's properties
+4. Zero-web / federated / your-hardware-your-data as data sovereignty
+5. tractor as a standalone commons contribution beyond just trading
+
+Things you'll want to customize before submitting:
+- Your Matrix room URL
+- Budget amounts / hourly rate to match your actual needs
+- Contact info on the form itself
+- Whether to attach a demo screencast or architecture diagram
+- The AI disclosure section (they require it since we used AI to draft)