AI-native infrastructure for music publishing.
Built for the streaming-native creator economy traditional publishing CAC economics cannot justify.
Problem: The math doesn't math.
Music Publishing suffers from a long-standing problem: the CAC is asymmetric. Human-led acquisition only pays back at massive scale, or via the top 0.25% of writers (the Taylor Swifts of the world). The 200K to 1M monthly-listener segment sits in the gap, generating real revenue but not enough to justify a human writer-acquisition engine.
“I ran North America for Sentric with a five-person headcount. The CAC/LTV does not work. Publishers mask these asymmetrical economics with existing catalogue revenue and/or a fee-based model.”
Sector growth is significant, but the economics in the territories driving it do not support human-led writer acquisition.
11TH CONSECUTIVE YEAR OF GROWTH
HALO'S ADDRESSABLE MARKET
IN 2025
Writer revenue, by tier
Mega-stars
Halo's segment
Self-serve territory
The CAC asymmetry, hidden in plain sight
CAC is flat. Revenue is skewed. Economics close at two ends: mega-stars (UMPG / Sony / Warner) or volume with no service (Songtrust, DistroKid, TuneCore). The middle-tier writer sits in the gap.
Catalogue masks the rot. Incumbents amortise human CAC against a writer's lifetime value, but the math does not close on the mid-tier. The writers driving growth sit in exactly the segment human CAC cannot reach.
Fast-growth regions amplify it. MENA 97.5% streaming. LatAm 88.1%. Global 69.6%. Digital-first, dispersed, multilingual, low-ARPU. Human admin cannot reach them profitably.
Legacy publishers and platforms are caught in post-M&A integration holding patterns.
- Sentric / TuneCore Pub · acquired by Believe.
- Songtrust (Downtown) · acquired by UMG.
- Kobalt / Kosign · acquired by Primary Wave.
That affords Halo time to build what major publishers will need to buy next.
Incumbents could build this. The constraint is organisational, not technical. Companies in the middle of integrating an acquisition rarely take on a fundamental rework of how they acquire and service writers at the same time.
Validation needs time they do not have. 18 to 24 months of real signing data is the minimum to prove the model.
AI talent is misaligned with incumbent comp structure. Music industry pay does not flex to match OpenAI- or Anthropic-tier packages, and the AI-native founders who could build this are incentivised by venture exit upside, not legacy payroll.
Their roadmaps are full. Legacy tech orgs allocate engineering capacity years out, and post-acquisition integration absorbs the rest.
Their workflows depend on humans coordinating legal, registration, servicing and support. Rebuilding from inside is internal warfare.
Catalogue revenue still funds attractive margins. The pressure to re-architect is muted until that cushion thins.
They are not nimble. Compliance overhead, data-security reviews and risk frameworks turn what should be sprint iterations into multi-quarter approval cycles.
AI still feels experimental to a legacy tech organisation. Agent-led production systems sit outside the risk frameworks built to vet known-safe vendors.
The constraints are real, but the incumbents know they need to solve the math problem. Acquiring a validated platform is the rational medium-term move. The 24-month window is when Halo becomes that platform.
Agentic AI collapses the cost of the entire publishing workflow by an order of magnitude.
It is the ONLY model that closes the unit economics for the middle-tier writer.
Publishing administration is an unusually clean fit for agentic execution.
The deal lifecycle is structured. The data interfaces (Spotify, MLC, PROs) are well-defined. The revenue waterfall is auditable. The work breaks naturally into discrete agent functions with clear handoffs.
Agentic systems capable of running lead generation, conversion, and registration end to end have only become production-viable in the past 12 to 18 months.
The capability stack (tool-use, structured outputs, long-horizon planning) shipped through 2025 and into 2026. The writer-acquisition engine that closes the unit economics for the middle-tier writer could not have been built in 2024. It can be built now.
Halo already operates the workflow. The platform scales it.
Halo Music LLC has operated as a publishing business since 2024. The ten deals below were closed operator-led, using the human playbook the platform is now automating. This round scales what is already in operation.
(7 WRITERS · 3 PLATFORMS)
ALREADY IN THE POOL
ACROSS ALL DEALS
Direct writers
- Alan Vuong
- BAILE
- Bird Language · Backer & Hutchison
- Blvck Svm
- Jordan Dykstra
- Twen · Jones & Fitzsimmons
- Andrew Schur · signed April 2026
White-label platforms
- Xelon Digital · ~1,000 writer-equivalents across 350 labels
- Tone Tree Music · ~200 writer-equivalents
- Oracle Artists Music Publishing · ~20 writer-equivalents, signed January 2026
Infrastructure live
Sentric admin backbone active. MLC bulk-data feed and public API provisioned. Platform scope v2.2 fully specified (~150 pages).
Objective: validate writer lead gen × conversion module.
Core functions: Spotify × MLC cross-reference, prospect scoring, outreach orchestration, royalty opportunity surfacing.
Objective: reduce servicing overhead, reduce onboarding friction, automate repetitive workflow layers.
Core functions: onboarding workflows, legal review, registration orchestration, statement generation, DIY catalogue management.
Objective: increase expansion revenue, improve retention metrics, identify cross-sell opportunities.
Core functions: churn indicators, sync matching, growth scoring, rights intelligence, early investment signals.
“Simon and Matt possess a deep knowledge of the black box complexity of publishing. More importantly, they understand what makes writers tick, and that helps us sign even the most reticent writers.”
Platform architecture: four lead agents, four supporting agents.
Halo is vertically integrated across the administration lifecycle, agent-to-agent. Most platforms in this space cover one or two steps; the handoff between steps is where operational integrity fails. Halo is designed to run the full lifecycle with no manual reentry between stages.
ACROSS THE AGENT FLEET
Build status · The platform is in active build against the v2.2 scope (approximately 150 pages). The architecture above describes agent functions at completion. This pre-seed round funds phases 1 through 3.
The new economics of a vertically integrated publishing business.
Traditional publishing administration is a services business: cost scales with writer count. An agentic platform inverts that curve. The platform cost is fixed; per-writer marginal cost is small. Margin expands with scale rather than holding flat. This is the shift from services multiples (~2x revenue) to vertical-SaaS multiples (8 to 12x).
Cost-to-serve is decoupled from headcount. That is what makes the middle-tier writer profitable to serve for the first time.
Modelled steady state at approximately 3,320 writers. Per-writer streaming economics calibrated against Sentric's Q4 2025 statement data.
| Direct gross royalties (stream + sync) | ~$1.2M |
| Existing partner gross (Xelon + Tone Tree + Oracle) | ~$10.5M |
| New platform partner gross (two additional partners) | ~$17.5M |
| Total gross royalties | ~$29M |
| Less: Sentric admin skim (~18%) | (~$5.2M) |
| Net royalties · Halo commission base | ~$24M |
Cost base · Lean steady state
Customer-success FTE ($91K). Part-time accounting ($9K). Fractional legal ($60K). Platform opex ($20.5K). Platform amortised ($36.9K). Total approximately $217K per year. Founders defer compensation through the build phase to maximise runway; modest base activates at Stage 1 validation.
Platform payback
MVP build cost approximately $110K. Recouped from Halo revenue at the 5% commission rate in approximately one month, at 10% in approximately two weeks, at 15% in approximately eleven days.
Calibration: Sentric Q4 2025 statement covering seven Tone Tree platform writers. Sample-level Sentric skim (~16%) and per-stream rate (~$700 per million streams) align with the modelled inputs. Two writers in the sample carry ambient/sleep-music catalogues, which skews the sample mech/perf mix to 94/6; the modelled 73/27 mix reflects typical writer behaviour and is the more defensible projection for a broader roster.
Halo as farm system: four acquirer archetypes, four valuation lenses.
The asset's value to an acquirer is not the writer base or the commission line in isolation. It is forward visibility into which writers are likely to graduate to 1M+ monthly listeners, twelve to twenty-four months before the broader market identifies them. Different archetypes value that visibility for different reasons.
Rights-only admin assets trade at services multiples. Platform-only music-data assets (Chartmetric, BeatBread-style) trade as data plays. The combination of administered rights with an integrated automation platform has no clean public comp because no one has built it. That combination is the source of the strategic premium. Halo enters this comp territory once cohort data validates LTV/CAC.
$1.0M pre-seed for 18 months of runway to validate unit economics.
Case Study: Halo Direct signing.
The 200K to 1M monthly-listener segment, in one writer. Globally distributed, MLC-anchored, streaming-native. The thesis in a single statement.
Alan Vuong · 1.4M monthly listeners
this cycle
earning
territories
via MLC
Actual paid out to Alan: $7,585, his 33.3% share across three-way co-writes, after ~15% combined Halo + Sentric deductions. Annualised run-rate places Alan in the upper-middle of the segment the human-led CAC model cannot economically reach.
Modelled unit economics at Stage 1.
Comparative projection. Human-led admin figures are industry benchmarks; Halo figures are modelled from the Stage 1 cost base (~$217K against ~3,320 writers) and design-target acquisition economics. See Section 07 for current build status.
Source: human-led benchmarks reflect industry estimates for traditional admin and acquisition. Halo figures derived from the Stage 1 Sanity Check model; cost-to-serve is the modelled steady-state cost base divided by writer count, CAC is a design target based on the agentic acquisition mechanism. Halo is in active build, not in production.
Halo vs Kosign.
Kobalt's Kosign is the closest in-market analogue to Halo. The comparison is on the table because sophisticated investors will raise it.
| Kosign | Halo | |
|---|---|---|
| Positioning | Marketed above the line. Brand-led acquisition. | Lower-funnel, agentic. Targets writers with named uncollected tracks. |
| Prospect identification | Inbound and sales outreach. | Spotify × MLC cross-reference. Specific tracks named in the first message. |
| Deal-making | Sales reps, counsel, deal framers. | Agreement agent drafts, redlines, executes. |
| Registration | Manual. Ops staff file across PROs and MLC. | Catalog agent files automatically, agent-to-agent from execution. |
| Customer support · FAQ | Human CS and FAQ teams. | Support agent. Churn-risk indicators feed Intelligence. |
| Cross-sell | Sales-led, relationship-bound. | Data-led. Coordinated behaviour of Acquisition and Intelligence on the signed base. |
| Data layer | Operational records inside Kobalt. | Longitudinal cross-writer dataset designed to be portable to an acquirer. |
Across each row, Kosign deploys substantial teams: Kobalt's referral pipeline drives lead generation, dedicated counsel and deal framers handle deal flow, registration is staffed by ops humans, and customer support and FAQ desks handle service. The "manual" shorthand reflects a full human stack at every layer.
Kosign's product surface is the marketing layer. Halo's product surface is the operational lifecycle, with marketing as one input. Closure is the moat.