Giantfish helps private credit managers and direct lenders turn their templates and past work into shared AI workflows. Start with one recurring task, test it with the people who do the work, and train the team to review the output.
A COO may need a consistent process across the firm. An investment lead may need associates to screen deals in the same format. Both need a working version they can check against completed work, and someone inside the firm who owns it after handoff.
We start by reviewing the task, its source documents, and your current template. Together we choose a first workflow whose output has a clear reviewer. The scope names what the model drafts, which figures need separate checks, and where a person makes the decision.
Use the CIM and borrower financials to prepare a draft in your screening template. Keep source pages beside the figures, flag conflicting numbers, and leave the pursue-or-pass decision with your investment team.
Explore the screening memo workflow →Start with the questions your team already asks. Use the screening gaps and your previous request lists to draft a deal-specific list that an associate checks before sending to the sponsor or management team.
Scope a workflow around a recurring reporting package, your existing spreadsheet, and the prior period. Define the figures to extract and the exceptions a reviewer needs to see. Your credit team checks calculations and interprets covenant terms.
Each is a candidate to scope, rather than a package of prebuilt integrations. We agree the files, output format, access, and review requirements before building.
A U.S. private credit manager
We built a deal screening and diligence request workflow in the firm's own AI account. It used their Excel screening template. Their compliance team and counsel reviewed the access terms before work began.
The team tested the first working version within three days. Structured feedback shaped three more versions. Five live, recorded training sessions supported the handoff, and associates began proposing additions of their own.
Read the engagement details →The pilot is usually about five weeks. We build one workflow from your templates and completed examples, then revise it as your team tests working versions. Testing covers source accuracy, missing information, and the review effort the draft needs.
You receive the workflow, a playbook, a review checklist, and five recorded training sessions. The workflow runs in your firm's approved AI account. Your firm owns the materials and can run the process after the engagement.
Bring one workflow owner, one or two people who do the task today, someone for access and IT, and a handful of completed examples. Each stage has a fixed fee, quoted after the first call and scoped in writing.
Compare workshop, pilot, and ongoing support →That is our usual starting point. We check your approved plan, file handling, and access controls with the people responsible for them. Copilot projects depend on your licenses and setup, so we confirm what is practical during scoping.
Your firm approves the accounts, documents, and access before we start. We write that boundary into the statement of work, work under your policies, and agree retention and deletion terms. Your compliance team reviews the arrangement. We remove our access at handoff.
Investment decisions remain with your credit professionals. The first scope defines the draft and the checks. Numerical calculations need a verifiable calculation method and a reviewer; interpreting covenant definitions requires the credit team's judgment.
For the data-handling questions to ask internally, read our guide to using AI with a confidential CIM.
Tell us what arrives, what your team produces, and who reviews it. We can discuss the process before you share any confidential deal documents.
Book a 30-minute call