The first in a series breaking down real prompts our members run on Octum’s Ora, and why the wording matters as much as the data underneath it.
Most insurance LP lists are built the wrong way around. You start with a universe of insurers, filter by AUM, maybe screen for anyone with a stated alternatives allocation, and end up with a spreadsheet of 300 names that every other private credit manager in the market is also working. The list is technically accurate and practically useless, because it tells you who could be a buyer without telling you who behaves like one.
The prompt below fixes that by starting from a structural preference instead of a sector label.
The prompt
I’m building our insurance company LP channel for our private credit platform and want to focus specifically on insurers that have awarded separate account mandates rather than commingled fund commitments — a structure we can accommodate and that tends to drive larger ticket sizes. Identify North American insurance companies with AUM over $10B that have awarded a separate account or SMA mandate to a private credit or direct lending manager in the last 24 months. For each, show total AUM, investment portfolio size, private credit allocation percentage, the specific mandate awarded with approximate size and strategy, and the Head of Fixed Income, Head of Private Assets, or CIO contact with direct email. Return a ranked prospect list of the top 20.
Run this on Ora and you get back a working prospect list, not a database export. Here is why each piece of it is doing work.
Why separate accounts are the right filter
Insurers are now one of the deepest pools of third-party capital in private credit, but they are not a homogenous buyer. A mid-sized life insurer building out private assets with a dedicated internal team behaves very differently from a P&C carrier writing a $25M check into a commingled fund alongside forty other LPs.
Separate accounts and SMAs signal a particular kind of buyer:
- They have the internal capability to negotiate. An SMA means bespoke investment guidelines, custom reporting, and often ratings and capital treatment considerations specific to that balance sheet. An insurer that has done this once has the staff, consultants, and board comfort to do it again.
- Ticket sizes are structurally larger. SMAs carry real setup cost on both sides. Nobody builds one for a small allocation.
- The relationship is stickier. Commingled commitments end when the fund ends. Separate accounts tend to persist and get topped up.
- It self-selects for managers who can service it. If your platform can accommodate the structure, the SMA filter is a moat. Most competitors screening the same insurer universe cannot deliver it, which means the list is not as crowded as the raw name count suggests.
The strategic move in this prompt is that it screens on observed behavior in the last 24 months, not on stated allocation targets. Stated targets are aspirational and often years stale. A mandate awarded eighteen months ago is evidence.
Anatomy: what each constraint is actually doing
“North American insurance companies with AUM over $10B.” Two filters in one clause. Geography narrows the regulatory regime so the mandates are comparable. The AUM floor is not arbitrary; below roughly $10B, insurers generally lack the internal staffing to run separate accounts and default to commingled or fund-of-funds routes. You are screening out the population that structurally cannot buy what you sell.
“Awarded a separate account or SMA mandate to a private credit or direct lending manager in the last 24 months.” This is the behavioral event. The 24-month window matters: long enough to capture a real sample, short enough that the decision-makers are likely still in seat and the allocation thesis is still current. Naming both “separate account” and “SMA” catches the terminology variance across public filings, consultant announcements, and trade press.
“Total AUM, investment portfolio size, private credit allocation percentage.” Three numbers that together tell you capacity. Total AUM is the headline. Investment portfolio size is the number that actually matters for an insurer, since general account assets are what get allocated. Allocation percentage tells you whether they are early in a build-out with room to grow, or already at target and unlikely to expand.
“The specific mandate awarded with approximate size and strategy.” This is the difference between a lead and a briefing. Knowing an insurer awarded a $400M senior direct lending SMA tells you the ticket range to anchor to, the risk appetite, and whether your strategy is complementary or directly competitive with an incumbent you would have to displace.
“Head of Fixed Income, Head of Private Assets, or CIO contact with direct email.” The ordering here is deliberate and worth copying. In an insurance general account, private credit frequently sits under fixed income rather than under an alternatives bucket, so leading with Head of Fixed Income catches the right owner more often than defaulting to CIO. The fallback chain means you get a usable name even where the org structure differs.
“Return a ranked prospect list of the top 20.” Forcing a ranking forces prioritization. Twenty is a number a two-person IR team can actually work in a quarter. Two hundred is a number that sits in a folder.
What you do with the output
The list is the starting point, not the deliverable. The sequence that works:
- Triage by fit before you touch the phone. Compare each flagged mandate against your own strategy. Where an insurer just awarded a senior direct lending SMA, you are probably not the second senior lender. You might be the opportunistic or asset-based sleeve.
- Verify the contact and the seat. Insurance investment teams turn over. Confirm the person is current before drafting anything.
- Write to the structure, not the strategy. The whole premise of this list is that these accounts prefer separate accounts. Your first paragraph should say you can accommodate one, not describe your fund.
- Track the consultant. Many insurance SMAs are intermediated. If a consultant ran the search, the consultant is the second call.
Variants worth running
The same skeleton adapts across the whole LP universe. Change the entity type, the structural preference, and the event window:
- Swap insurers for public pension plans and separate accounts for first-time fund commitments to build an emerging-manager channel.
- Swap North America for EMEA and private credit for infrastructure debt, where SMA structures are common for the same balance-sheet reasons.
- Swap the 24-month mandate window for CIO or Head of Private Assets appointments in the last 12 months, which surfaces accounts in the middle of a strategy reset.
- Keep everything and change only the AUM band to $2B–$10B to find the tier that is just now building the internal capability, before your competitors reach it.
Each of those is a different channel built from the same three ingredients: a size filter, an observed behavioral event, and an output schema you can act on.
Why this is a prompt and not a database query
The reason this cannot be done with a conventional screen is that no field in a traditional investor database is called “awarded a separate account mandate to a direct lending manager in the last 24 months.” That fact lives across consultant announcements, statutory filings, trade press, and manager press releases. Assembling it manually is a week of analyst time per refresh, which means it gets done once and then goes stale.
Ora reasons across those sources and returns the synthesis, with the reasoning shown so you can verify it. The prompt is the specification. The more precisely you write the constraints, the less time you spend cleaning the output.
One practical note: treat every contact detail as a lead to confirm, and make sure your outreach process is squared with your compliance team and the relevant data and marketing rules in each jurisdiction. The list gets you to the right twenty names. Everything after that is still your job.
This is the first post in Prompt Anatomy, where we take a real prompt run on Ora and break down why it is written the way it is. Have one you want dissected? Send it to support@octum.ai.
New to Octum? Ora is our research engine for institutional investing, built on GAR, our own cognitive architecture. Join in about three minutes.
