Summary
This article addresses IT leaders’ core concerns about AI in ERP: data safety, access control, data handling, and security complexity. It explains how Dela was designed from the ground up to operate within Vantagepoint’s existing security model, inherit permissions natively, and give administrators control over rollout.
Why It Matters: When a new AI capability lands in an ERP, IT leaders get the hard questions: Is data safe? Who can see what? Do we need to reconfigure everything? Getting these answers right determines whether AI becomes an enabler or a blocker in your firm.
Key Takeaways:
- Dela inherits permissions natively at the application, record, and field level. There is no parallel permission layer to configure, and no risk of the AI surfacing data users cannot otherwise access.
- Your project financials, client information, and employee records are used to answer questions. They are not used to train the underlying model or improve AI for other firms.
- Dela Insights comes with granular administrative settings so you control which insight categories are active, which roles receive alerts, and how the system behaves across your organization.
When a new AI capability lands in an ERP platform, the people who feel it most acutely are not always the ones in the press release. It is the IT leaders and system administrators who get the questions: Is our data safe? Who can see what? What happens if the AI is wrong? Do we need to reconfigure everything?
These are the right questions. And for Dela in Deltek Vantagepoint, the answers are more reassuring than you might expect, because Dela was designed from the ground up around the security and control infrastructure that IT leaders care about.
Here is what you need to know.
Dela Does Not Get Its Own Set of Keys
The most important thing to understand about how Dela, your AI orchestrator, works is also the simplest: it does not operate outside of Vantagepoint's existing security model.
Dela inherits permissions natively at the application, record, and field level, including user-defined fields. If a user does not have security permissions to a record or a field in Vantagepoint, they will not be able to retrieve that information in a Dela response. There is no parallel permission layer to configure, no separate access control system to maintain, and no risk of the AI surfacing data that a user could not otherwise access.
This is not a marketing claim. It is an architectural decision. Dela operates entirely within the authenticated user's security context in Vantagepoint, meaning the access rules your team already manages are the ones Dela follows. One set of controls for the entire platform.
Your Data Does Not Leave to Train a Model
AI platforms vary significantly in how they handle customer data, and it is worth being explicit about where Dela stands.
Vantagepoint communicates with the underlying large language model, hosted on Microsoft Azure, via HTTPS/TLS encryption. Requests are stateless, meaning no conversation or query data is retained between sessions. And per Deltek's contract with Microsoft, customer data is not used to train or improve the underlying model.
Your project financials, client information, contract documents, and employee records are used to answer your users' questions. They are not used to make the AI smarter for someone else's firm.
Admins Control What Gets Surfaced and to Whom
Dela Insights, the proactive alert system that flags project risks, expense policy violations, and financial anomalies, comes with granular settings that give system administrators meaningful control over the experience.
In the Dela settings, administrators can configure which insight categories are active, which roles receive which alert types, and how the system behaves across the organization. Insights are role-based by design, so a project manager sees earned-value gaps and margin alerts, enabling early intervention. Finance sees GL trends before close. Business development and marketing teams see client news to nurture the relationships that matter. Each role gets the signals that matter to their job, delivered each morning automatically. That focus reduces noise and increases trust in the system.
This is not a binary on/off switch. It is a configurable system that lets your team roll out Dela Insights deliberately, starting with the use cases that matter most to your firm and expanding from there.
A Materially Different Posture from Some Competitors
Not all AI implementations in the ERP market are built the same way, and the differences matter from a security standpoint.
Some competitor AI tools access customer data through third-party BI connectors, essentially querying your data the same way an external tool like Power BI would. That approach introduces an additional layer outside your ERP's native security context, with its own access controls, its own data handling, and its own risk surface to evaluate and manage.
Dela does not work that way. There is no external query layer, no third-party intermediary between your data and the AI, and no additional system for your team to audit or certify. The security perimeter is Vantagepoint itself.
Dela Was Purpose-Built for Professional Services
Beyond the architectural security advantages, there is another reason Dela does not require months of customization: it was built from the ground up to understand the dynamics of A&E and consulting projects. Project phases. Fee types. Margin drivers. Billing models. That domain knowledge is embedded in the system. When your team deploys Dela, it understands your business on day one. Competitors bolting generic AI onto their platform still require significant tuning before they can flag the project economics issues that matter to you.
What a Responsible Rollout Looks Like
For IT leaders preparing to enable or expand Dela across their firm, a few practical considerations are worth front-loading:
- Start with Insights Settings. Before Dela Insights goes live for your users, review the settings configuration. Decide which categories to enable first and which roles should receive which alerts. A deliberate rollout prevents noise and builds user trust in the system.
- Review your expense and timesheet policies. Dela's policy-based compliance features for expense approvals and timesheet reviews are only as strong as the policies you have uploaded. Natural language policy documents can be configured directly in the application. If your firm has not yet done this, it is the highest-leverage configuration step available today.
- Communicate the permission model to skeptical stakeholders. The most common concern IT teams encounter when rolling out AI is that it will surface information people should not see. The native permission inheritance model is your answer to that concern, and it is worth communicating proactively before questions arise.
- Monitor through the lens you already have. Because Dela operates within Vantagepoint's existing security and audit framework, you do not need a separate monitoring strategy for AI-related access. Your existing processes apply.
The Bigger Picture
The firms that will get the most value from AI in their ERP are the ones whose IT teams are enablers, not blockers. That requires trust in the underlying architecture, confidence in the data handling, and control over the rollout. When your IT team can deploy Dela enterprise-wide with confidence, every project manager gets proactive margin alerts. Every finance team member sees GL trends before close. The entire firm operates on fresher information. That confidence is what unlocks the value.
Dela is built to earn that trust, not to work around it. The security model is native, the data handling is transparent, and the configuration tools exist precisely so that IT leaders can shape how AI shows up for their users.
Customer Town Hall
See What's Next for Deltek Vantagepoint
Join Deltek experts and fellow customers each quarter to discuss product vision, recent releases, and how our Vantagepoint fits your goals.