Kamba · Research Note · February 2026
The Workflow Gap: Why More Data and More AI Still Isn't Working
Every month a dataset sits outside production is a month
your portfolio couldn't use it.
The bottleneck is almost never access to data or access to AI. It is the workflow between having it and using it — the gap between an AI pilot and a governed production system, between the budget line and the portfolio impact. This paper names it, documents it, and puts a number on what leaving it open costs.
For a $2.5bn fund onboarding six datasets a year at status-quo speed, that delay costs an estimated
~$7.3m in gross alpha opportunity annually.
Illustrative model · Gross, pre-costs · Mid-size tier, 6 datasets/yr, 60 bps good-tier assumption · Results vary materially by strategy and execution
The evidence
The execution gap
66% priority / 16% executing
66% of asset managers have made GenAI a strategic priority. Only 16% have a fully defined strategy and are implementing it. The gap is the workflow.
BCG Global Asset Management, May 2024
The daily frustration
79% cite integration as #1
79% of practitioners say combining data is their most frustrating daily challenge — more than cost, more than coverage gaps.
Exabel / BattleFin, Jan 2025 (n=130, ~$820bn AUM)
The spend paradox
Margins down 5 years running
Asset manager margins have declined for five consecutive years despite rising technology spend.
McKinsey Global Asset Management, Jul 2025
What the paper covers · 7 sections · 16 cited sources
01 · The cost model
Alpha opportunity lost, by firm size
Per dataset, per firm, across four AUM tiers ($750m–$30bn+). One metric, one auditable formula — so your quant can stress-test it in ten minutes.
02 · Two types of cost
P&L impact vs operational waste
Alpha opportunity lost (return not earned) and operational opportunity cost (vendor overspend + zombie pipelines) — defined separately and modelled independently.
03 · The hidden waste
Redundant, decayed, and zombie data
How to find it, quantify it, and remove it safely — with a ranked output and governance-safe deprecation path that won't break anything downstream.
04 · The market data
Where the buy side actually stands in 2026
16 cited sources — Coalition Greenwich, Lowenstein Sandler, Exabel, BCG, McKinsey, SEC, IOSCO — on adoption, spend, AI execution rates, and governance expectations.
05 · AI & agentic workflows
The strategy-to-execution gap
Why 66% of firms have a GenAI strategy and 16% are executing it — and what the firms that are executing differently have in common.
06 · The 2026 benchmark
What good looks like now
Not another AI pilot. A governed end-to-end workflow — from discovery to production — where compliance is built in from the start, not discovered at audit time.