By Ty Alibhai, Head of AI, VFP Consulting
There is a version of the AI conversation happening in professional services firms right now that goes something like this: “We need to move faster; AI can help, so let’s go.”
That instinct is not wrong. But it is incomplete, and the gap between that instinct and the discipline required to act on it wisely is exactly where services organizations win or lose in the years ahead.
I have spent a considerable amount of time thinking about this challenge, and more importantly, living it. At VFP Consulting, we are not observing the AI shift from a distance. We are building AI into our products, embedding it in our delivery model, and helping clients navigate it on their own implementations. What I keep coming back to is this: the firms that will lead are not necessarily the ones with the most AI tools. They are the ones that have figured out how to accelerate with AI while preserving, and in fact deepening, the judgment that clients pay for in the first place.
That is harder than it sounds.
The pressure to move faster is not going away
Services clients expect more. Margins demand efficiency. The firms that learn to harness AI as a genuine delivery accelerator will create real structural advantages over those still debating whether to try. But acceleration without judgment is just speed in the wrong direction, or worse, speed that quietly erodes the trust clients place in your team’s expertise.
I see the most immediate and meaningful AI opportunity is in the parts of delivery that are repetitive, documentation-heavy, and synthesis-intensive. The Discovery and Design phase. Requirements management. Resource allocation. Project health monitoring. These are not peripheral to what services firms do. They are the connective tissue of every successful implementation, and they are where significant time is lost on every engagement.
I see the most immediate and meaningful AI opportunity is in the parts of delivery that are repetitive, documentation-heavy, and synthesis-intensive.
This was stated plainly from the stage, and it’s something we’ve seen play out directly with our clients. Organizations that invest in AI before getting their data, processes, and fundamentals right tend to accelerate their existing problems rather than solve them. The firms that see the biggest returns from AI are the ones that first earned them through operational maturity.
What we are doing at VFP
We have been applying Claude, Anthropic’s large language model, to accelerate Discovery and Design work, and the results have been tangible. On a recent engagement, we established a dedicated Claude workspace configured with VFP’s standard PSA requirements baseline from Campfire, session transcripts from client discovery workshops, and supplemental documentation added as the engagement progressed.
The outcomes were concrete: over 300 client-specific functional system requirements were produced, each validated against Certinia’s best practices and VFP’s standard requirements framework, and uploaded directly into Campfire, our Implementation Lifecycle Management (ILM) app for consultant and engineer use. Each requirement was enriched with contextual links to relevant Certinia documentation to support downstream configuration. Beyond that, the team identified more than 20 incremental requirements through AI-assisted transcript analysis, surfacing gaps that manual review likely would have missed. Certinia documentation research was conducted directly within Claude, enabling faster feature lookups without switching tools.
Work that would have taken days of manual synthesis, from hundreds of hours of transcripts, took a few hours. But here is the important qualification: every AI-generated requirement was reviewed and validated by experienced consultants before it went into Campfire. The AI produced a working first draft. The humans determined what was accurate, complete, and appropriate for that client. That model is what scales, and it is the only model I am willing to stand behind.
Campfire: built by consultants, for consultants
Campfire did not start as an AI play. It started because our delivery team was struggling with disparate tools and point solutions to manage implementations. In 2018, VFP consultants built what became RADTest, a tool designed specifically for the way implementation teams actually work. By 2024, that internal tooling had evolved into a commercial product, and Campfire was born. It is now live on the Salesforce AppExchange and built for medium- and large-sized professional services organizations running complex implementations.
The problems Campfire solves are the ones that erode services’ profitability and client trust every single day: scope creep that goes uncaptured and unbilled, go/no-go decisions made without full visibility into testing status, and the friction of coordinating across too many disconnected tools. Clients using Campfire report:
- 76% less scope creep
- 34% faster testing cycle
- 27% more revenue from change orders.
Those are not marketing numbers. They reflect what happens when you give a delivery team the right structure at the right moment.
What makes Campfire genuinely different is the AI Engine at its core. It is a combination of Claude and direct Salesforce API calls, all behind a clean, intuitive interface that consultants can pick up without a learning curve. In practice, it’s straightforward: a consultant feeds a batch of system requirements into the Campfire engine, and the engine builds the solution directly in the customer’s target test environment.
The impact on the Build phase is significant. Build cycles are cut by over 40%, which is meaningful on its own, but the more important outcome is what that time buys back. Instead of spending their hours on point-and-click manual configuration, consultants are free to do what clients are actually paying them for: thinking through the architecture, challenging assumptions, and advising on the best possible solution design. That shift, from administrator to advisor, is where the real value of AI in services delivery shows up.
Agentforce and Certinia Veda: solving the operational problems services firms know too well
Our AI work extends beyond our own delivery model. We are now embedding agentic capabilities into client environments through Agentforce and Certinia Veda, and the use cases address pain points that have been part of service operations for as long as I can remember.
Certinia Veda is an intelligent operations engine built natively on Salesforce. Because it operates within the client’s existing Certinia environment, it works directly against their actual project data, resource availability, and financial records without any external synchronization. It operates within a deterministic, rules-bound framework, which means all AI-generated outputs adhere to the organization’s established business logic and security permissions. I think that architecture is exactly right. The guardrails are not a limitation; they are what make the tool trustworthy.
The Spring 2026 release brings three capabilities that address what I hear from services leaders most often. The Work Reallocation Agent handles one of the most operationally disruptive events in any services firm: replacing a resource mid-engagement. Whether due to leave, departure, or competing project demands, staffing reallocations are time-consuming and high-stakes. The AI agent identifies all assignments for the resource being replaced, executes searches against the same criteria, including region, practice, group, role, and skills, and proposes best-match replacements so work in flight can continue without delay. What previously took days now happens in minutes.
The Project Assistant Agent continuously monitors project health, analyzing milestones, tasks, and timecards to surface issues as they emerge rather than after the fact. An overdue milestone, a task running behind schedule, and timecards that do not match the plan. For project managers carrying multiple active engagements, the difference between staying proactive and constantly reacting is significant.
Generative AI Summaries condense detailed project data, financials, and resource information into concise digests, reducing administrative reporting time and giving project managers up to 20 hours of capacity back per month to focus on delivery rather than data reconciliation.
The cumulative effect is that Veda redirects human attention away from administrative overhead and toward the decisions, relationships, and judgment calls that genuinely require people.
The lesson I keep coming back to
Every AI capability we have deployed, either internally or for clients, shares one characteristic: it amplifies human judgment rather than replacing it. That is a deliberate choice on our part, not a hedge.
AI in professional services fails when it is positioned as a decision-maker. Our team learned this directly. When we explored AI-assisted process diagram generation, extensive work on prompting and tooling ultimately produced output that did not meet our project standards. We documented that finding, pulled back from that capability for now, and used manual diagram development instead. That is what responsible AI adoption looks like in practice: test it, evaluate it honestly, apply it where it genuinely works, and be willing to say no when it does not.
The firms that approach AI adoption with that level of rigor will build something durable. Those that chase AI as a positioning exercise, without the governance, validation practices, and human oversight to support it, will create problems at exactly the moments that matter most to clients.
At VFP, we are building toward a future where AI consistently makes our consultants faster, our processes more consistent, and our clients’ implementations more successful. But we are doing it with the same discipline we apply to every engagement. Speed is only valuable if judgment accompanies it.
That is the challenge this industry needs to get right, and I think Services Week is exactly the right moment to say it plainly.
