SONG DealFinder — Principal-Level Briefing | Grand Canal Capital Partners

§ 0 · Executive summary

A pre-screened forward-pipeline engine — built outside SONG, calibrated for SONG.

Built by Grand Canal Capital Partners over six months alongside Dave Murray's live origination work. Compounds rather than decays: 982 connected wiki pages, 13,742 source files, 29 sector packs, eight active geographies — every Tier A target scored against SONG's hard underwriting constraints before either side spends time on it. Eclairion (>€700m, France, close August 2026) is the first deal sourced, scored and advanced entirely through this framework.

— 01

An origination platform, not a research folder.

A second brain that compounds. Every new target inherits the scoring framework. Every new sector inherits the appetite map. Bi-weekly pipeline refresh, continuous research waves, Coverage Registry pre-flight gate — the system gets sharper every cycle, not staler.

— 02

Pre-filtered through SONG's box.

Every surfaced target has cleared the fixed-rate test, MA-eligibility check, size threshold and sector appetite tier. SONG team time spent on credit and structuring, not on screening. 60-point proprietary score, four output tiers, hard-constraint filter.

— 03

Live deal — Eclairion — validates the whole stack.

The first deal sourced, scored, and advanced entirely through the SONG DealFinder framework. >€700m senior secured term loan, HPC sponsor, two French data-centre sites, S&P rating in progress, fixed coupon c.5.72% / 295 bps, 10-year tenor. SONG's first deployment in France. The system held at every layer.

The system at a glance

Connected wiki entries

982

Curated nodes · 18 sub-vaults · fully back-linked

Source files indexed

13,742

5-LLM research engine · tier-tagged · primary-sourced

Sectors mapped · Geographies

29 · 8

DF-01 → DF-29 · 8 active European markets

Live deal · validating proof

€700m+

Eclairion · target close August 2026 · GCCP fee €2.6–2.8m

§ 1 · Origin

What we set out to build.

In November 2025, we set out to build the system that would surface every MA-eligible European real-asset deal opportunity six to twelve months before it landed on a banker's desk — and pre-screen it through SONG's hard constraints before either side spent time on it.

Origin

Nov 2025

Sector mapping

Dec – Jan

Scoring model

Feb 2026

Research waves

Mar – Apr

Pipeline build

Apr 2026

Knowledge graph

Apr – May

Today

May 2026

— 01

Calibrated, not generic.

Every target scored against SONG's actual underwriting constraints — size, fixed-rate, MA eligibility, counterparty quality, sector appetite. Built outside SONG, but for SONG.

— 02

Compounded, not cumulative.

982 connected wiki pages. Every insight back-links to every other. Six months of research that compounds — not a folder of PDFs that decay.

— 03

Live, not static.

Bi-weekly pipeline refresh. Continuous research waves. Coverage Registry pre-flight gate. The system gets sharper every cycle.

§ 2 · The system in one image

The visual universe of BPA market intelligence.

Every node in this graph is a unit of BPA market intelligence — a target company, an insurer, a regulatory event, a refi catalyst, a comparable transaction, a sponsor, an adviser. Every line is a connection. Built natively inside Claude Code, synthesised across five LLMs (Claude · Perplexity · ChatGPT · Grok · Gemini), grounded in primary sources (IR pages, annual reports, regulatory filings, Companies House, CRO, Land Registry, ISIN bond data). 982 connected nodes. 13,742 source files. One second brain, calibrated to SONG.

Every node a unit of BPA market intelligence. Every line a connection.

SONG DealFinder · BPA Market Intelligence Universe · 982 connected nodes

982 nodes· ~10,000 connections· 18 sub-vaults· Six months compounded

LAYER 01

Raw sources

13,742files ingested

5-LLM research engine cross-validated against IR pages, annual reports, regulatory filings, Companies House, CRO, Land Registry, ISIN bond data.

LAYER 02

Synthesis

4,868markdown pages

Each file tier-tagged ([PRIMARY] / [SECONDARY] / [SCREENING]). Anti-hallucination flags applied (N/A · DISPUTED · STALE · SINGLE-SOURCE).

LAYER 03

Wiki distillation

982wiki entries

One curated page per target · sector · capital provider · regulation · reference transaction · adviser · sponsor · regulatory event.

LAYER 04

Backlink network

~10K[[wikilinks]]

Bidirectional wikilinks across the graph. Open one target — see every sector, lender, transaction and regulation that touches it.

LAYER 05

The graph

1connected system

Obsidian renders the entire network as a navigable, queryable knowledge graph. Search across 13,742 files in milliseconds. Insight compounds — nothing decays.

Why a second brain, not a research folder. A single Tier A target — say, Eclairion — connects to its sector pack (DF-06 Data Centres), country pack (France), sponsor (HPC), comparable transactions, MA-eligibility regulatory overlay, four versions of deal sheet, fee agreement structure, meeting intel, pipeline tracker, and the active live-deal dashboard. Click any node, see all the others. Every fact carries a source. Every new insight makes every existing node sharper.

§ 3 · The scale & full inventory

What six months built — equivalent output of a four-person research team for a year.

Compressed into six months through an agentic build pipeline GCCP developed alongside the research itself. Headline numbers resolve into four research domains: regulatory landscape, geographic coverage, deal sectors, named companies — each researched as a standalone discipline and stitched together through the index.

13,742

source files indexed

Researched individually · tier-tagged · back-linked into the graph

4,868

intelligence pages

Markdown research outputs across all sub-vaults

982

connected wiki entries

Curated knowledge graph · 18 sub-vaults · fully back-linked

1,414

institutional docs

Word memos · partner briefings · IM-style outputs

1,171

HTML dashboards

v1.3 deal sheets · pipeline views · per-target dossiers

463

scoring databases

Excel + CSV target lists, scored cohorts, benchmark tables

258

standing rules

Codified discipline · anti-hallucination · H-rule constraints

29

sectors mapped

DF-01 through DF-29 · tiered by SONG appetite fit · 8 active geographies

29 Deal Sectors · Calibrated by SONG Appetite Tier

29DF codes · 30 sub-sectors

Tier 1

10 / 10 fit

BPA anchor

DF-01 Utilities DF-02 Wind DF-03 Solar DF-04A BESS DF-04B Hydrogen

Tier 2

9 / 10 fit

Strong fit

DF-05 Hospitals DF-07 Grocery DF-08 Pubs / Leisure DF-09 Care Homes DF-18 UK Ground Debt DF-22 Healthcare Ireland

Tier 3

8 / 10 fit

Proven

DF-11 Holiday Parks DF-14 Self-Storage DF-15 Industrial / Logistics DF-23 Irish Infrastructure

Tier 4

7 / 10 fit

Selective

DF-06 Data Centres DF-12 PRS / BTR DF-13 PBSA DF-17 Aviation Infra DF-20 France DF-21 Germany DF-24 PBSA Ireland DF-25 Iberian Solar DF-28 Italian Renewables

Tier 5

6 / 10 fit

Conditional

DF-10 Hotels DF-16 Fibre / Digital DF-19 Theme Parks DF-26 Nordic Renewables DF-27 Benelux Utilities DF-29 Emerging Sectors

Regulatory landscape

12+ 176 mon.

PRA CP8/26 — FundedRe

Cons. Jul 26

PRA MAIA — PS17/25

Live

UK PRA Dear CEO 2026

Live

Solvency II Review (UK)

Monitor

Irish CMP / ISEM MA

Live

NTMA Sovereign Wrap

Active

RTA Act 2026 — IE PBSA

In force

EU Green Taxonomy + CSRD

Monitor

Regulatory Calendar

Quarterly

Geographies researched

8markets

United Kingdom

280+

Ireland

210+

France

94

Germany

39

Iberia

22

Italy

14

Benelux

11

Nordics

8

Companies & assets

~670entities

~670Entities researched

391Master ranked universe

238Bonds & loans · ISIN-level

177Named target wikis

~30Tier A scored cohort

Why the numbers matter. The graph on the prior section isn't decorative. Every node is a researched object — a company, a regulation, a transaction, a sector pack. Each one ingested from primary sources (Perplexity Deep Research · IR filings · annual reports · regulatory papers · Companies House · CRO · Land Registry) and indexed under a discipline compounding for six months.

§ 4 · Coverage map

Mapped at the asset level across Europe.

Every active opportunity geocoded onto a sector-specific pipeline map. Solar across the continent. Infrastructure across France. The full target universe across Ireland. Granularity that SONG's internal team would have to hire eight country leads to maintain.

[ Solar Pipeline Map · Europe ]

DF-03 Solar · Pan-European Pipeline

Every solar developer / sponsor mapped across UK, Ireland, France, Germany, Iberia, Italy, Nordics.

● DF-03 Solar · Pan-European Pipeline

[ France ]

France Pipeline Map

Country command portal · full pipeline

● France · Full Pipeline

[ Ireland ]

Ireland Target Map

Tier A cohort · geocoded

● Ireland · A-Tier Cohort

Per-sector pipeline maps · 8 sectors mapped

Solar, wind, BESS, utilities, data centres, hotels, healthcare, transport — each with its own geocoded pipeline map. Every named developer, sponsor, sub-asset plotted with refi catalyst and SONG fit tag.

Per-country command portals · 6 countries

France, Ireland, UK, Germany, Iberia, Italy each have a dedicated command portal — pipeline, fee tracker, debt maturity wall, outreach calendar, competitive matrix.

Granularity SONG can't internally maintain

Replicating this level of asset-level geographic detail in-house would require eight country leads + a research function. The system maintains it through automation and bi-weekly refresh cycles.

Where the white space is

Nordics, CEE, Italy — mapped but lightly exploited. Each represents a structurally addressable extension if SONG signals appetite. The system is already prepped to surface what's there.

§ 5 · Methodology

How every target gets scored.

A proprietary 60-point institutional scoring framework, calibrated to SONG's hard constraints. Every target ranked, every output partner-ready, every recommendation traceable to a source-tier and a constraint check.

Six scoring factors

MA eligibility

Size fit

Counterparty quality

Operational status

Relationship access

Timing catalyst

→

60

Total possible points

Then filtered through hard constraints

Fixed-rate only · MA-eligibility verified · sector appetite tier · borrower-vs-tenant test · development checklist where applicable

→

Output tiers

Tier A · PURSUE · ≥ 45

Tier B · DEVELOP · 35–44

Tier C · MONITOR · 25–34

REJECT · < 25

Built-in discipline

Every claim source-tier tagged ([PRIMARY] · [SECONDARY] · [SCREENING]) · anti-hallucination protocol with N/A · DISPUTED · SINGLE-SOURCE · STALE flags · Coverage Registry pre-flight gate prevents duplicate research · jurisdiction-matched scoring perimeter (ring-fenced security only, not platform AUM) · platform-aggregation divestiture check on every Tier A score.

§ 6 · A second asset

The BPA School — institutional curriculum, ready for SONG's team.

Beyond the intelligence platform, we built a parallel asset: the BPA School Origination Academy. A 78-module learning platform that brings any new SONG team member up to speed on the BPA market, MA mechanics, Solvency UK regulation, the 29-sector taxonomy, and the GCCP method. Already live, brand-locked to the v1.3 institutional standard.

Bulk Purchase Annuity School · GCCP Origination Academy

From "what is an annuity?" to "Tier A scoring + H-rules" — in one curriculum.

A structured learning platform that takes a new analyst from zero BPA market knowledge to fluent practitioner. Adapts to any team member who needs to ramp on this lender class — without the months of unstructured onboarding cost.

78

Modules

13

Groups

v1.3

Brand-locked

01

Foundation Level

30 chapters

Pension · annuity · BPA · MA mechanics · MA-eligibility · ground debt · senior term · 60-point score · H-rules · [AI-DRAFT] discipline · rated note feeders. Zero to fluent.

02

SONG Capital-specific

4 modules

Overview · credit box · track record · decision process. Tailored to your underwriting parameters specifically.

03

BPA Fundamentals

8 modules

Basics · insurer economics · asset mechanics · rated note feeders · regulatory · originator landscape · pricing & spreads · advanced topics.

04

European Markets

5 modules

Insurer landscape · regulatory frameworks · country profiles · currency & structuring · competitor landscape. Pan-European context for cross-border deals.

05

Sectors

29 sector modules

One per DF code. Utilities · wind · solar · BESS · hospitals · data centres · holiday parks · grocery · pubs · care homes · self-storage · industrial · etc. The taxonomy you saw earlier, taught.

06

Playbooks & Glossary

7+ quick-refs

Partner curriculum · live pipeline · 90-second screen · originator playbook · glossary · syllabus · changelog. The on-the-job reference layer.

Why a second asset matters. SONG scales by hiring. Every new analyst, associate, and origination team member needs ramp time on the BPA market, MA eligibility, the regulatory landscape, and the 29-sector taxonomy. That curve normally takes months. The BPA School compresses it into a structured curriculum a new joiner can work through in days. As part of a structured partnership, the curriculum becomes a SONG team enablement asset — not just a research engine.

§ 7 · Output · Two Formats

One shell. Every artefact renders through it.

Left: Gas Networks Ireland — a target-level dashboard showing the two-lane proposition (€400–700m). Right: Ireland BPA Market Overview — the geography-level dashboard ranking the country's Tier A cohort. Same v1.3 shell, same brand discipline, scales from one target to one market to one sector to one continent.

Target-level · Gas Networks Ireland · €400–700m two-lane proposition

GNI Team Review Dashboard · target-level v1.3 sample

Market-level · Ireland BPA Overview · Tier A cohort ranked

Ireland Market Overview · market-level v1.3 sample

v1.3 Brand-Locked Shell· Target → Market → Sector → Continent· Partner-Ready

Locked structural shell

Every artefact renders through the same v1.3 template. Shell, register, and structure fixed. Content (charts, numbers, narrative) varies per target. Mechanical consistency = institutional credibility.

Deal-sheet matrix

Headline · structure · pricing benchmark · refi catalyst · capital-provider fit · scoring breakdown · H-rule check · reference transactions · open items. Standardised across all 320 dossiers.

Source-tier discipline

Every fact carries [PRIMARY] / [SECONDARY] / [SCREENING] tag. Anti-hallucination flags surface every gap. No fabricated rates, no fabricated DSCR, no fabricated terms.

[AI-DRAFT] discipline

All artefacts tagged [AI-DRAFT] until partner sign-off. Outputs not promoted to "partner-circulated" until a named principal has reviewed and approved. Built for partner-level signal protection.

§ 8 · Pipeline · The system works

Bi-weekly refreshed pipeline · Eclairion proves it.

A bi-weekly-refreshed pipeline showing every active opportunity by stage, geography, sector and risk-adjusted fee envelope. Forward funnel visible six to twelve months out. And one deal already moving toward close that validates the entire framework.

Live Deal · Live Proof

Eclairion — >€700m Senior Secured Term Loan, France

SONG-LED · BBB TARGET · CLOSE AUG 2026

The first deal sourced, scored, and advanced entirely through the SONG DealFinder framework. HPC Capital sponsor, two French data centre sites (Bruyère and Besse), S&P rating in progress, fixed coupon ~5.72% (295 bps), 10-year tenor. SONG's first deployment in France. First test of every layer of the system — sector pack, scoring, H-rule discipline, capital provider fit — and the system held.

>€700m

Facility size

€2.6–2.8m

GCCP arrangement fee

80%

Close probability

Aug 2026

Target close

Active forward funnel · 8 named opportunities

Every active deal stage-gated screen → develop → pre-clearance → mandate → close. Risk-adjusted fee envelope per row. Total active book: ~€1.9bn+ facility · €11–14m+ fee potential. Updated bi-weekly with archive trail.

Beyond the immediate · the maturity wall

2026–2028 view across European real-asset issuers. 238 individual bonds mapped at ISIN level. €6.2bn+ in 2026 maturities pre-screened for SONG fit. Forward catalysts surface before deals get banked.

§ 9 · Why it matters to SONG

SONG's growth case is capacity-constrained, not capital-constrained.

£3bn → £4bn AUM by year-end. £50bn/yr BPA market growing. Capital "relatively infinite" — the ceiling sits on origination throughput, not investor demand. The SONG DealFinder is engineered to break that exact constraint. Without forward pipeline at this scale and discipline, SONG's growth case caps where its in-house origination capacity does.

01

Forward visibility · 6–12 months out

Refi walls, maturity catalysts, sponsor exit windows, regulatory triggers — mapped before deals get banked. SONG sees flow that hasn't yet entered the broker funnel.

02

Pre-filtered through SONG's box

Every surfaced target has cleared the fixed-rate test, MA eligibility check, size threshold and sector appetite tier. SONG team time spent only on credit and structuring — not on screening.

03

Geographic reach SONG can't internally maintain

Ireland · UK · France · Germany · Iberia · Italy · Benelux · Nordics. Continuous coverage across eight markets with sector-level depth. Materially extends SONG's de facto origination perimeter without hiring eight country leads.

04

A second brain SONG doesn't have to build

Six months of compounding research, 5-LLM synthesis architecture, calibrated to your credit box. Replicating this internally = ~£1.5m and 12–18 months. Replicating the relationship network (Dave's CBRE + GCCP's 14k contacts) is not buyable.

The counter-scenario. Without a structured origination partnership of this scale, SONG's path to £4bn AUM relies on growing in-house origination team headcount, paying broker fees on more transactions, or accepting slower deployment velocity. We've already built the alternative. The conversation is whether SONG owns access to it through a structured commercial frame, or sees it deployed elsewhere.

§ 10 · The proposition

How we'd structure a deeper working relationship.

Three framings to open the conversation. None of them final — they are the shape of what we think makes this most valuable to both sides. The right structure is the one we agree at principals level.

A

Strategic Origination Partner

Retainer + Success

Annual retainer for platform access and continuous coverage, plus uplift on the standard arrangement fee for SONG-routed closes. Recognises the platform's value as distinct from per-deal economics. Locks the relationship beyond any single transaction.

B

Sector or Geography Exclusivity

First-Look Mandate

GCCP first-look in named sectors and geographies for a defined period — Irish ground debt, French infrastructure, BESS first-mover, UK refi wall. Lower headline cost for SONG; protected channel for GCCP. Proves the moat without crowding the rest of the market.

C

Embedded Research Function

Quarterly Principal Sessions + Live Access

Quarterly principal-level sessions, live access to dashboard layer for the SONG team, co-developed sector briefs. The closest analogue to having a four-person research desk inside SONG — without the cost, the management overhead, or the talent risk.

What we'd hold back from the working layer. The scoring methodology, the hard-constraint rule architecture, the operating model and the build pipeline are the platform IP — what makes the recurring value possible. The output layer (pipeline view, deal sheets, sector cohorts) is what the SONG working team engages with directly. We treat that boundary the same way any institutional manager treats its alpha engine.

Indicative economics · For principals-level discussion

Annual retainer

£250–400k

Platform access · continuous coverage · quarterly principals sessions

Standard arrangement fee

1.00%

GCCP's current standard on SONG-routed closes

Uplift on platform-sourced closes

+25 – 50 bps

Recognition that the deal was platform-surfaced, scored, and pre-cleared

On a typical £300m Tier A close, total GCCP economics would land in the £3.75–4.5m range per transaction, retainer-inclusive. Structure, magnitude, and exclusivity terms all subject to principals-level negotiation. The numbers above anchor the conversation — they're not a final offer.

SONG DealFinder. The system, the proof, the proposition.

A pre-screened forward-pipeline engine — built outside SONG, calibrated for SONG.

An origination platform, not a research folder.

Pre-filtered through SONG's box.

Live deal — Eclairion — validates the whole stack.

What we set out to build.

Calibrated, not generic.

Compounded, not cumulative.

Live, not static.

The visual universe of BPA market intelligence.

What six months built — equivalent output of a four-person research team for a year.

Mapped at the asset level across Europe.

How every target gets scored.

The BPA School — institutional curriculum, ready for SONG's team.

One shell. Every artefact renders through it.

Bi-weekly refreshed pipeline · Eclairion proves it.

SONG's growth case is capacity-constrained, not capital-constrained.

How we'd structure a deeper working relationship.

Bringing a SONG team member into the working layer is the right second step — but it works only once the commercial frame is set at principal level.