Financial services teams are still spending hours rewriting AI-generated outputs, even as the industry rushes to adopt the technology. Noah Faro, CTO and co-founder of Farsight, says the core problem is that most AI tools produce work that looks finished but collapses under the scrutiny of a live deal - and they have no understanding of how a particular firm makes decisions.
The gap between AI-generated and client-ready work
"A lot of the tools on the market today can generate something that looks complete. Ask for a pitch deck or a financial model and you'll get back an output that may pass a quick first impression," Faro said. "The challenge is whether that work will hold-up and pass the stress test in a real deal context - oftentimes, that isn't the case, and that is what causes the iteration."
That gap exists because a financial deliverable isn't just a collection of slides or numbers. It's a structured argument where every section must connect logically, drawn from deep research and shaped into a narrative that fits the firm's unique style. General-purpose AI models don't reflect those internal standards out of the box, so teams end up rewriting the output to match what their clients and senior bankers actually expect.
What institutional knowledge really means
Institutional knowledge is often reduced to indexing a firm's past documents so an AI can pull reference slides or historical formatting. Faro calls that useful but insufficient. "There are thousands of decisions that get made along the way to creating such a deliverable - and teaching an AI system why something was done a certain way helps educate the process for getting to that polished output, rather than just hoping the AI system connects the dots."
At Farsight, this layer is called the System of Judgement. It captures the preferences, edits, and approval deltas that separate a first draft from the final version. That's where a firm's true standard lives, and it isn't reflected in finished files alone. Systems that capture this "why" over time can generate work that aligns with how the firm operates, not just how the industry operates.
For AI for Finance applications, this shift from generic output to firm-specific reasoning is what separates tools that save time from those that still demand heavy human rework.
Why generic AI tools fall short in finance
General-purpose models are trained on public internet data, public benchmarks, and expert content that reflects broad practice. "All of it describes how the work is done everywhere, not how your firm does it in particular," Faro said. A pitch deck generated by such a model might have the right sections but miss the analyst's preferred data sources, slide architecture, or narrative logic.
Prompt engineering and attached examples help, but they're a stopgap. "You can't change the underlying behavior of the model, so each time you want a deliverable done your way, you're re-teaching it from scratch inside the prompt - your preferences, your formatting, your approach - and paying for that re-education in tokens and inconsistency on every call."
Recent MIT research found that 95% of enterprise AI pilots produce no measurable return, and the study attributed that to a learning gap, not model capability. Models can perform when they understand the firm's standards, but without a system that encodes those standards, the output stays generic.
How AI can learn a firm's judgment
"For an AI system to truly 'learn' a firm, it needs to understand not only how they do things - like create specific deliverables, write different types of emails, or conduct certain depths of research - but also why they do them," Faro said. That means maintaining an understanding of deliverable structure and communication style, while also learning from the feedback loops of edits, approvals, and argument framing.
As more work flows through the system, these two dimensions make the AI more consistent and tailored. Every analyst at the firm can then draw on that collective intelligence, reducing the time spent aligning outputs with firm-specific expectations.
Protecting client data while operationalizing knowledge
Faro draws a sharp line between raw client data and the judgment layer around it. "The thing actually worth capturing from a firm isn't its raw client data, it's the method behind how the work was done with it - how an analyst structures an argument, which source clears the bar to ship, how one framing gets chosen over another." That judgment can be encoded and reused without exposing confidential numbers.
Traceability becomes an asset. When the system keeps a record of where a figure came from, which sources made the final version, and who signed off, the entire process becomes defensible. Teams move faster because they're not starting from scratch, and client trust is maintained because sensitive content is handled in a controlled, auditable environment.
Why this matters for finance professionals
For finance teams, reducing the rewrite cycle isn't just about efficiency - it's about consistency under pressure. Tools that capture a firm's System of Judgement let analysts spend less time reformatting slides and more time refining the underlying argument. The firms that close the learning gap first will deliver client-ready work faster and with fewer errors, turning their institutional knowledge into a competitive advantage that compound over every deal.
Your membership also unlocks:
