Artificial intelligence (AI) is already being adopted across government contractors, but adoption alone does not create enterprise value. The next priority is making sure AI efforts are tied to business outcomes, supported by the right governance model, and designed to scale beyond individual users, teams, and one-off pilots.
That was the central focus of a recent Baker Tilly webinar, which brought together professionals from Baker Tilly and Deltek to discuss how government contractors can move from AI experimentation to governed, measurable impact. The discussion drew on findings from Deltek’s Clarity Government Contracting Study, along with practical perspectives on AI governance, enterprise operating models, use case prioritization, workflow automation, and compliance considerations in a highly regulated environment.
AI adoption is widespread, but governance is still catching up
Deltek’s Clarity Government Contracting Study is an industry report based on survey responses from 900 government contractors, providing a snapshot of current priorities across the market and across business functions.
The study highlights a sharp increase in AI adoption across government contractors, with finance, business development, contracts, legal, HR, project delivery and back-office teams all finding practical ways to use AI. But widespread use does not automatically translate into enterprise value. Without a structured approach, AI efforts can remain fragmented, difficult to measure and risky from a compliance standpoint.
That risk is especially important in GovCon, where companies operate under audit requirements, pricing pressure, cybersecurity expectations, DCAA and DCMA scrutiny, CMMC readiness requirements and complex contract obligations. In this environment, AI cannot simply be treated as another productivity tool. It needs to be connected to business value, compliance requirements and enterprise risk management.
One of the important takeaways from the Clarity report is that AI adoption is accelerating faster than AI governance. Many contractors are already using AI to summarize information, analyze data, support reporting, accelerate drafting and reduce manual effort. But many are still determining how to protect sensitive data, establish usage policies, validate outputs and maintain audit trails.
That gap creates risk. If AI use remains fragmented across teams, organizations may struggle to understand what tools are being used, what data is being exposed, how outputs are being reviewed and whether workflows are compliant. The result can be what David Hickey described as “agent chaos,” where different systems, tools, AI agents and informal user practices proliferate without common standards for access, testing, documentation, ownership and monitoring.
With the right framework, however, AI can support faster proposal development, better contract ingestion, stronger financial processes, earlier risk detection and more defensible decision-making.
Governance as a competitive advantage
Clarity findings suggest that organizations with stronger governance practices and more integrated systems are better positioned to scale AI effectively. In GovCon, that matters because speed and control are no longer separate goals. Contractors need to move faster while still protecting sensitive data, maintaining compliance and preserving auditability.
These capabilities can support better pricing confidence, shorter close cycles, more reliable financial monitoring and faster response to opportunities. But those benefits depend on trust. Leaders need confidence that AI-enabled workflows are secure, repeatable, documented and aligned with regulatory requirements.
A mature governance model helps answer practical questions: Which AI use cases should be prioritized? Who owns the process? What data can the AI access? How are outputs validated? What happens when the system flags an exception? When does a human need to review or approve the result? How is the activity documented for audit purposes?
In that sense, governance is not just a risk-management function. It is what allows organizations to scale AI responsibly, avoid fragmented adoption and turn promising use cases into repeatable business capabilities.
Start with business value, not AI possibility
Organizations should not begin with the question, “Where can we use AI?” A better starting point is, “Where does the business need to improve, and what outcome would actually matter?”
This distinction is important because many AI pilots prove that the technology works, but still fail to create enterprise value. They may automate an isolated task, produce an impressive summary, or demonstrate a new capability, without changing the workflow, improving a measurable KPI, or solving a problem that leadership cares about.
For government contractors, those priorities can be organized around four enterprise value drivers: revenue growth, operating margin, asset efficiency and intangible asset value. Those value drivers help organizations move from broad AI interest to specific business outcomes, such as win rate, proposal throughput, days sales outstanding, cost to serve, close cycle time, audit findings or billing readiness.
Once the desired outcome is clear, organizations can identify the workflow, handoff, decision point or control activity that most directly affects it. If the goal is revenue growth, AI may support opportunity analysis, proposal development or basis-of-estimate creation. If the goal is margin improvement, AI may help reduce manual work, rework or exception handling in finance, contracts, procurement or project accounting. If the goal is asset efficiency, AI may support billing readiness, invoice preparation, collections workflows or other working-capital processes.
Only after the value driver, metric and workflow are defined should organizations decide which capability fits the problem. The answer may be generative AI, agentic AI, predictive analytics, intelligent document processing, robotic process automation, workflow orchestration, or a combination of those capabilities with human review.
This value-first approach helps organizations avoid pilots that are interesting but disconnected from business outcomes. A successful AI initiative should do more than show that the technology can perform a task. It should improve a metric, strengthen a workflow, reduce risk, or create a repeatable capability the organization can scale.
Building an AI operating model
To move from experimentation to scale, organizations need an AI operating model.
One approach is to create an automation or AI center of excellence. This group does not need to build every AI solution itself. In many organizations, development may be federated across business units. However, the center of excellence can establish common standards for documentation, coding, testing, security, monitoring, and maintenance.
An effective operating model should also include executive oversight. Leaders need to define the AI strategy, set priorities, approve investment, and ensure alignment with the organization’s broader goals. Information technology (IT) and security teams should help manage data access and system integration. Compliance teams should be involved early so that controls and audit requirements are designed into workflows from the start.
The model can mature over time. Many organizations begin with targeted pilots, prove value in one area, and then expand. As the program scales, governance, standards, and support structures can become more sophisticated.
Where AI can create value
There are several practical AI use cases across the government contracting life cycle.
Proposal development is one major opportunity. AI can help generate outlines, assist with drafting, check compliance against solicitation requirements, and improve consistency across submissions. When connected to existing opportunity and contract data, AI can help teams reduce the time from request for proposal (RFP) receipt to compliant proposal submission. Deltek’s AI orchestrator, Dela, provides AI-assisted proposal capabilities that are embedded into familiar workflows inside Costpoint and also extend those workflows into Microsoft Word to provide continuity throughout the proposal process.
Contract ingestion and project setup are also strong candidates because they directly affect speed to execution, billing readiness, revenue recognition, cash flow and downstream compliance. AI-enabled document processing can extract key contract facts into a structured review format, while specialized agents can help apply configured revenue recognition rules, route accounting conclusions for review and approval, and draft project structure recommendations based on enterprise-specific Costpoint setup guidance. The value is not simply that AI reads the contract; it is that contract evidence becomes structured, reviewable and usable for downstream decisions, helping reduce manual rekeying, setup errors, billing delays and audit support gaps while keeping humans in control of key accounting and project setup conclusions.
Finance and accounting workflows offer additional opportunities. AI can assist with anomaly detection, financial close, expense report analysis, invoice readiness, and cash flow processes. In many cases, AI can work alongside deterministic automation, such as robotic process automation. Bots can manage repeatable, rules-based steps, while AI agents can support interpretation, analysis and exception handling.
Sentinel Control Fabric™ / Approver Zero is another high-value pattern for first-line transaction review. Instead of finding errors after posting, AI can review expenses, invoices, purchase orders, labor transactions and journal entries before the first human approver acts, flagging missing support, coding issues, contract restrictions, allowability concerns or policy exceptions earlier in the workflow. The value is a shift from reactive cleanup to preventative control, helping reduce rework, billing delays, audit findings and compliance leakage.
Compliance must be built in from the beginning
For government contractors, AI-enabled workflows must be designed with compliance in mind. That means mapping requirements across the full process, including Federal Acquisition Regulation (FAR), Defense Federal Acquisition Regulation Supplement (DFARS), contract clauses, internal policies, cost accounting standards, and audit expectations.
Not every decision should be automated. In some cases, the right answer is to route an item for manual review. If an invoice lacks key project information or includes costs that require additional allowability analysis, the workflow should flag the exception rather than force a decision.
Ultimately, auditors will want to understand how the organization knows the AI-enabled process is working. Management review, sampling, internal audit involvement, and remediation procedures should all be part of the control environment.
Bring AI into your enterprise workflows with Baker Tilly
Costpoint and the broader Deltek ecosystem provide a strong foundation for bringing AI and automation into the systems and workflows government contractors already use. Through its Deltek partnership, Baker Tilly helps organizations connect those capabilities to practical use cases such as proposal support, compliance checks, financial close, contract ingestion, project setup and other high-value GovCon workflows.
Listen to the full webinar and get practical ideas for turning AI governance, compliance, and use cases into real business value.
Baker Tilly helps organizations combine Deltek experience with Intelligent Process Automation capabilities that can reduce repetitive work, improve accuracy and support more efficient operations. As a Deltek Platinum-tier partner, Baker Tilly combines Deltek solution experience with government contracting, accounting, business process, and compliance knowledge to help organizations implement, integrate, and optimize systems for long-term value.