Nearly half of finance leaders name security as the top blocker to expanding AI use in their revenue operations, and 42 percent point to governance and system integration close behind, according to a new Salesforce survey of 865 finance leaders. That gap between AI’s promised return and the reasons teams hesitate to deploy it at scale is becoming the actual battleground in revenue technology, and one vendor’s answer to it points at where the category is heading.
The complexity finance can no longer track by hand
The Salesforce research, conducted through Agentforce Revenue Management, found that 71 percent of finance leaders say their company now sells through more channels than it did a year ago, and 65 percent track deals across more than one revenue model at once. Nearly 9 in 10 expect to add more consumption-based products. Each of those additions multiplies the number of systems a revenue team has to reconcile before it has one true picture of the business.
Sam Chung, Chief Customer Officer for Agentforce Revenue Management at Salesforce, said the CFO’s job has changed shape because of it. “AI decisions don’t reach finance as budget requests anymore, they reach us as questions about risk, control, and accountability. The CFO has moved from signing off on AI to being answerable for it,” Chung said.
Advertisement
The survey found 90 percent of finance leaders who use AI report positive ROI, yet 67 percent say their team still completes at least one in five workflows manually, often in spreadsheets. So the appetite for AI is not the problem. What the data shows is a trust gap: teams that want the productivity are the same teams naming security, governance, and integration as the reasons they have not gone further.
One vendor’s answer: own every layer
Sidetrade, an order-to-cash platform that manages close to 10 trillion dollars in B2B transaction data, launched SAFE, the Sidetrade Agentic Framework for Enterprise, this week as a direct answer to that exact gap. Every agent built on SAFE runs on Sidetrade’s own models, in its own data centers, on its own GPU compute, rather than routing customer financial data through a third-party model provider or hyperscaler.
Olivier Novasque, CEO and founder of Sidetrade, framed the shift as a change in what CFOs are actually asking. “During the first wave of generative AI, the question every enterprise asked was which model was the most powerful. Finance departments are now asking a different question: how do we put AI to work on one of our most sensitive processes, cash flow generation, without handing control of our customer data, operating model and intelligence to someone else,” Novasque said.
The framework is built to answer the same three worries the Salesforce survey isolates. Data sovereignty answers the security concern: sensitive financial data never leaves Sidetrade’s own infrastructure or trains a third-party model. A policy layer with deterministic escalation and full audit trails answers the governance concern, with every agent validated against Sidetrade’s security standards before it can go live. And because Sidetrade sets its own compute pricing rather than passing through volatile per-token rates from a model provider, it answers the cost-visibility concern directly, with multi-year contracts that guarantee what an agent will cost to run.
The underlying asset behind that pitch is over a decade of order-to-cash transaction history: nearly 45 million buying companies and close to 10 trillion dollars in B2B transactions, according to Sidetrade. Luke Hennerley, the company’s VP of AI Operations, said that history is what a frontier model alone cannot provide. “We start from a decade of O2C data that gives our AI context and history at a depth no incumbent can match. SAFE is how we put it to work. We run everything on our own GPUs, so the inference is ours, and now the learning is too,” Hennerley said.
What it means for the revenue leader
The specifics of SAFE matter less to a RevOps buyer than the pattern it represents. Every agentic AI vendor now claims some version of security and governance. Few can show what that claim actually rests on: whose GPUs the agent runs on, whose model weights it calls, and who sets the price per token six months from now. Sidetrade’s launch is notable mainly because it makes those three answers checkable rather than promised.
That is the question a revenue leader evaluating any agentic platform, not just an order-to-cash tool, should now be asking before it reaches a security review: does this vendor own the model and compute the agent runs on, or does it resell someone else’s, and what happens to the contract price when that someone else changes its terms. The Salesforce data suggests most finance and revenue teams are not yet asking that question systematically, since integration and governance still rank as blockers rather than solved problems.
Vertical integration is not free. A vendor that owns its model stack trades access to the fastest-moving frontier models for control over cost and data residency, and Sidetrade is explicit that it is optimizing for predictability over raw model horsepower. For a revenue or finance leader, that trade is the one worth interrogating directly in a vendor evaluation, alongside the pricing terms and the audit trail a vendor can actually produce, rather than the one buried in a security questionnaire nobody reads end to end.
As more categories inside the revenue stack move from AI pilots to AI running production workflows, the vendors that can answer the ownership question directly, the way Sidetrade just did, will set the evaluation bar the rest of the market gets measured against.
Source: GlobeNewswire
Related: A Pricing Benchmark Is Not a Pricing Strategy | The AI Maturity Gap Has One Root Cause: CRM Data
