Fund Recs · working papers
What four hours of public research found
One analyst, one AI stack, no access to their systems. Everything here is from sources anyone can reach. Start with the findings; the working is behind them.
⚠ Pre-human
No human has read a word of this. It is machine output from four hours of research against public sources, published deliberately in its raw state so the reasoning can be attacked before anyone polishes it.
Judge the chain, not the copy. Every conclusion is shown as numbered steps with an evidence grade. If a step is wrong, say which number — the words downstream are cheap to rebuild.
It is a strawman. Built entirely from what Fund Recs and five competitors publish. With their own data — client count, contract values, funnel rates, win/loss — the commercial half stops being a guess.
Eight things you can say out loud
"You already have a named AI agent with better numbers than SmartStream's — and it lives only on Alan's personal LinkedIn."
Rex: 95% of conversations, 42% resolved with no human, 94.1% CSAT, n=699. Confirmed absent from all 301 crawled site pages. measured
"You own two category words outright. We found no competitor content on depositary or transfer agency at all."
Fund Recs publishes on depositary 142 times, transfer agency 28. See the corpus caveat — the ratio is inflated by crawl asymmetry, the direction isn't. measured
"Duco publishes $80,000 and 'live in a day' on AWS Marketplace. You publish nothing. In a $50–150k band that isn't discretion, it's being skipped."
Duco's full price card, verbatim, with the metered ladder to $240k. measured
"Duco isn't out-thinking you. They're out-spending you. 344 live ads to your zero."
LinkedIn Ad Library, both counts measured the same way. A rented position, not an owned one. measured
"Your best numbers are in a 2024 blog post nothing links to. 79% straight-through at Channel Capital. 5%→20% and 57%→97% on EMIR."
Their own case study and EMIR page — the only before/after table they own. measured
"Three AI models describe you wrongly. ChatGPT calls your product 'Apex Group's software'."
The OEM relationship with the brand stripped off. Fixable with one page. reported
"Four competitors were in Stockholm on 2–3 September. You have London and Dublin to yourselves, with Alan moderating."
PostTrade360 vs InvestOps 15–17 Sept and Irish Funds 1 Oct. measured
"Gresham just partnered with Finologee on automated depositary reconciliation in Luxembourg. Someone has finally noticed your ground."
First competitive move onto the uncontested territory. measured
Five things that are wrong
| Gap | Evidence |
|---|---|
| No third-party validation anywhere | No G2. No Capterra. No Gartner Peer Insights — their Reconciliation Solutions market page exists and Fund Recs isn't on it. No analyst coverage. Four of five competitors have at least one. This blocks procurement outright. |
| Nothing to try, nothing to price | A 93% proof-of-concept close rate, and no trial, no sandbox, no published price, no self-assessment. Every route is "Book a Demo". |
| Coverage | 60 staff, 10 in Business Development, 52 of 60 in Ireland, selling into 40+ countries. Four to six quota carriers against roughly 1,545 buying units. |
| The consequence is never stated | They publish the control for every failure mode and the cost of almost none. The GlobeOp US$43.5m NAV figure sits on their own blog, from 2015, on no product page. |
| A live arithmetic error | "The number of breaks has been reduced by over 2,000%" — on the same page as their best before/after data. |
Why they're in a strong position
The business works
$10m+ ARR
Growth accelerated 15% → 49% YoY. Cash-burning to ~10% net margin. Rule of 40 score of 59. 93% PoC-to-close. CSAT 97.1% against a stated 79% software average.
Self-published, LinkedIn, 3 May 2026
The ground is unclaimed
Personal liability
Their buyer signs for a NAV they didn't calculate. Duco's buyer doesn't — so Duco cannot credibly make the accountability argument even if they wanted to.
Corpus + domain
The assets exist
98 controls
Pre-built fund-native workflows — CLO, PCAP, TER, capital calls — against a competitor toolkit. Plus an owned conference that has raised €371,750.
Library crawl
The fixes are cheap
~16 PoCs
That's the whole year's new-logo requirement. Best proof is filed wrong, not missing. Four client videos sit untranscribed. No product change needed.
Derived — see Commercial
Six things we need from them
| Ask | What it unlocks |
|---|---|
| Client count and average contract value | Validates or kills the entire commercial model. One question. |
| Funnel conversion rates, and what counts as a PoC | The comms plan is sized on benchmark rates, not theirs. |
| Revenue by module | We're prioritising by evidence volume, not by what pays. |
| One named EMIR reference | Largest product line, zero named humans. Muzinich, Innocap, Three Rock or Maples. |
| Win/loss notes | Confirms who they actually lose to. We inferred it. |
| Permission to publish Rex's numbers | Best AI proof point in the category. |
Where everything is
| Page | What's on it |
|---|---|
| Deliverables | All five things they asked for, as strawmen, each with its derivation chain and how to break it. |
| Position | The argument, the ten ways their buyer loses the licence, and what's at stake in plain English. |
| Audience | Six accountable buyers, the named client story, and the partner channel. |
| Commercial | What a deal is worth, the market size, coverage, and the comms plan with a budget. |
| Comparison | When not to buy Fund Recs, and what the AI models say today. |
| The corpus | What we collected and how. The thing that made the rest possible. |
| The brief | What they originally asked for, and the four things that changed under research. |