Public finance
Mandates are won on relationship depth. Market updates, pool RFQs, transaction RFPs and refunding analyses are not, and every one is built before a fee is earned. Starch produces the set from your own precedent and prior submissions, so those hours go back to the bankers.
And the sources and analysis behind each one.
Selection is governed largely by things nobody controls: price, rotation policies, local preference. The return on all that bespoke work is low and unpredictable.
| What you use today | What it does well | Where it falls short |
|---|---|---|
| Horizontal AI platforms for finance | Firm-wide search and Q&A across your documents | One product sold to hundreds of firms. It answers questions about the official statement; it does not produce the RFP response, and every group bidding against you runs the same thing. |
| General AI assistants | Fast drafting, summarisation and Q&A on public information | Cannot touch MNPI, deal pipelines or your compliance perimeter, and hold no memory of the materials and deals your group has already produced. |
| Market data and analytics platforms | Levels, issuer data and comparables | A data layer, not a workflow layer. Bankers still assemble the pitch by hand. |
| An internal proposal database | Consolidates prior submissions in one place | Maintained by hand, and hard to query across. |
| Junior bankers | Analysis and asset creation | Capacity-bound, and turnover erases institutional memory. |
Nothing on the market automates the banker workflow itself.
Starch is built for this coverage group, on your own precedent.
We work with more than one firm and we would rather say so. What we will not do is hand another group your build. The software is built against your precedent, your templates and your approval standards, and those are not things a competitor can license.
It also does not stand still. Every mandate leaves precedent behind and the next response starts from it. A platform hundreds of firms share is as good on day one as it will ever be for you. Yours keeps pulling away.
The curve has a start date. A group that begins now is building something the group that waits will spend years catching up to.
Thirty minutes, on the work your group actually produces.