Direct lender access. Bespoke credit structures. Execution-driven outcomes. We treat origination as intelligence — public filings, cap-table adjacency, liquidity events, and affinity signals, stacked and routed into a pipeline most placement channels miss.
After 50+ engagements across CRE, fintech, energy, healthcare, and venture, one finding holds across every vertical we've run: the gap between a 1-in-100 outreach hit rate and a 1-in-20 outreach hit rate isn't message quality. It's segment selection, and segment selection is detectable.
Form D filings, 8-K transactions, 10b5-1 plans, AngelList syndicate participation, cap-table adjacency, partnership-class announcements — every capital event leaves a signature in the 90 days before the conversation is even possible. We catch those signatures at the source.
One signal is noise. Four signals on the same record — affinity + capacity + geography + network proximity — produces confidence one to two orders of magnitude higher than any individual signal. That math is what separates a 1-in-100 channel from a 1-in-20 channel.
Conventional placement engages partners only when they appear in market. By that point most of the cohort is allocated. The signal stack catches the wave at the earliest detectable point — within days of a liquidity event, weeks of a transition, or quarter of a distribution.
A real-time view of how public filings, cap-table data, LinkedIn deltas, and ecosystem rosters flow through enrichment and identity-resolution layers into a tier-segmented pipeline. Each packet is a single record moving through the system. The counters update as it operates.
A working snapshot of what our signal architecture is detecting right now, and how a stacked signal profile produces confidence one to two orders of magnitude beyond any single-signal channel. Adjust the stack — the numbers move.
Confidence is multiplicative across independent signal classes. The same individual presented with a single signal is noise; the same individual presented with four stacked signals is a referenceable Tier-1 candidate. This is the math that separates 1-in-100 outreach from 1-in-20.
Confidence multiplier when any two signal classes co-fire on the same candidate. Hover a cell.
Every Starbase engagement runs the same five-stage pipeline. Each stage filters and enriches the candidate population — the bar charts beneath show throughput in real time. Records that pass all five emerge as either Tier-1 referenceable or Tier-2 stacked-cold outputs.
The signal stack is universal architecture. The signals themselves are vertical-specific. A CRE equity raise responds to Form D and Reg A filings; a B2B SaaS Series A responds to ARR multiples and headcount velocity; a VC fund LP raise responds to liquidity-cycle and cap-table signals. Select a vertical to see its active signal set.
An engagement is not delivery work — it's a tuning sequence. The signal stack is calibrated to the mandate over the first 30 days, holds tight through the second 30, and produces referenceable pipeline by day 90. The four nodes below describe what we tune at each stage, and what gets delivered.
Mandate brief, vertical anchoring, signal-stack initial weighting. We map the counterparty universe against the raise structure, then pre-load the first 6 signal classes against the mandate profile.
First full detection cycle runs against the live signal feed. 4 — 6 candidate profiles surface within the first week. We hand-review each against mandate fit, retune the stack, and confirm the routing rules with the client.
Full stack live. Weekly delivery rhythm — qualified introductions, signal-class attribution per intro, calibration deltas. Secondary signal classes light up as the cohort develops; secondary anchors get added on signal weight, not opinion.
Feedback loop tightens — candidates that converted feed back into the stack, the stack gets sharper. Output by day 90 is a referenceable pipeline with attribution: which signals produced which mandates, and at what conversion rate.
The case below is a real engagement with verifiable numbers. The five anchor profiles beneath are anonymized representations of how the same architecture is currently tuned across active mandates — what we anchored against in each, and what the system surfaces as a result. Specific candidate identities are redacted.
A New Jersey commercial finance and capital advisory firm was generating roughly 60 qualified introductions per quarter through a strong referral network and a top-performing LinkedIn channel. Leadership wanted a controllable outbound layer that complemented those channels — and contractually had to beat them.
Starbase committed to a 1.5× performance floor against their existing LinkedIn benchmark. The engagement delivered 1.7×, converted into a $340M mandate pipeline, and closed $29M of funded transactions across solar, acquisition financing, fleet, and aviation deals.
Sell-side counterparties for a generalist commercial finance book: family offices with private-credit allocation, limited partners in opportunistic credit, direct investors in asset-backed deals, and institutional lenders with idle warehouse capacity. Segmented by sector (solar, acquisition, fleet, aviation) and reachability — not by AUM-tier alone.
Five active or recently active engagements. Each shows what we tuned the signal stack to detect — and the kind of candidate population that surfaced once detection ran for a cycle. Specific client identifiers, candidate names, and counterparty figures are anonymized.
We expected outbound. We got an intelligence layer.
Capital allocators and operators both face the same underlying problem in a different dialect. The architecture below is shaped by the mandate; the discipline beneath it does not change.
Funds in the $10M–$50M target band raising under 506(b) or 506(c). The institutional air gap is structural; LP origination is the bottleneck. We build the LP surface your mandate is too small for placement agents to bother with.
Series A/B/C founders, project sponsors, and operating-company principals who refuse to lead with a banker. We surface strategic capital partners whose existing portfolio and thesis match — before the broker channel touches the deal.
CRE equity sponsors, sponsor consortia, and family-office-backed deal sponsors expanding LP rosters past their referral ceiling. We extend the surface area you can't manually cover while filtering the population that has historically wasted outreach budget.
Three discrete service lines, one underlying origination architecture. Each is structured around the same signal-detection discipline that powers the engine above — calibrated to your specific mandate.
Operator-side origination and counterparty positioning for finance platforms, capital-raising sponsors, and emerging fund managers — the proof side of the engine.
Bespoke credit structures sourced through direct lender relationships. ABL, cash-flow, equipment, specialty, growth, bridge — placed against the right capital partner for your structure.
Complex structuring, execution oversight, and capital markets positioning for institutional sponsors and operating companies running multi-counterparty processes.
Four primary counterparty profiles. The signal stack adapts to each; the discipline beneath does not change. Click any to jump to the matching engagement profile.
Operating companies and tech-enabled founders raising Series A/B/C or refinancing growth — capital partners sourced before the banker channel.
See thesis →Principal-led sponsors raising 506(b)/(c) equity, expanding LP rosters past the warm-network ceiling. Anchor LPs and family-office capital surfaced.
See thesis →DSCR, bridge, hard-money, and equipment-finance lenders launching credit vehicles or warehouse facilities — LP and counterparty origination.
See thesis →Energy, infrastructure, industrial, and equipment-heavy operators — structured debt and project-finance placement against active-deployer counterparties.
See thesis →A representative sample. The featured engagement is verifiable end-to-end under NDA; the structured-finance examples are anonymized per counterparty confidentiality. Specific identifiers and counterparties are redacted.
102 new sell-side introductions across solar, acquisition, fleet financing, aviation. Outperformed their LinkedIn channel by 1.7×, generated $340M mandate pipeline, converted to $29M of funded transactions.
Complex multi-jurisdictional structuring for an industrial energy sponsor. Direct lender placement against asset-backed collateral with cross-border perfection.
506(b) syndication for a principal-led CRE sponsor expanding past warm-network LP ceiling. Mixed RIA-channel and accredited-individual placement.
Working papers, methodology notes, and field observations from active mandates. Reasoning we'd rather show than hide — the same analytical posture clients see during an engagement.
Single-channel origination collapses below 12% precision at scale. Stacking three to four calibrated signal classes compounds confidence into the 80 — 95 band. The paper walks the math, the calibration cadence, and the 102-introduction engagement that validated the framework end-to-end.
Read working paper →Market-structure analysis of the underserved mid-band: too small for top-tier placement agents, too large for direct-network outreach. The mechanics behind the gap and what fills it.
Walk-through of the three signal classes that dominate every other input in calibration runs. Why affinity matters most, where capacity overrides it, and how liquidity timing decays.
Fresh signals outperform large stacks of stale ones. We model 14-day, 30-day, and 90-day decay curves across affinity, capacity, and liquidity classes — and why most platforms get this wrong.
A 30-minute working session is enough to lock target segments, surface the first signal stack, and tell you whether this is the right fit for the close window you're working against. We respond inside one business day.
30-minute working session. We tune the engine to your mandate in front of you, surface the first qualified candidates, and tell you whether this is the right fit before either of us commits to anything.