Certinia’s Veda: A Working Blueprint for Agentic AI in Professional Services

A category at an inflection point

Professional services automation (PSA) has spent two decades knitting together the operational backbone of consulting firms, systems integrators, agencies and software companies running their own services arms. Now, that well-established backbone is being asked to absorb agentic AI at speed. MIT's NANDA initiative reports that roughly 95% of enterprise generative AI pilots show no measurable profit and loss impact, with the failure traced to weak integration into actual workflows rather than model quality; this may be surprising, given the focus on models in AI (Fortune on the MIT NANDA study). Similarly, BCG reports that 74% of companies have yet to generate tangible value from AI, and only 4% are creating substantial value across functions (BCG). Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027, citing rising costs, unclear benefits and what its analysts call “agent washing” (Reuters).

Against that backdrop, Austin-based Certinia announced Veda on 15 April 2026. It is a suite of ten specialist AI agents and 64 intelligent actions, built on Salesforce Agentforce and grounded over fifteen years of services-specific data and business logic (Certinia press release). It deserves close attention because Certinia is directly tackling the integration problem the MIT and BCG numbers actually describe.

What Veda actually is

DJ Paoni, Certinia's CEO, framed it plainly at the press and analyst briefing:

“We aren't just adding AI features. What we're really doing is redefining the velocity of the services industry.”

The name is taken from the Sanskrit for knowledge or wisdom, and Paoni argued the choice was deliberate: “In an era of black box AI, Veda is different. It's built on the foundation of Certinia's two decades of institutional knowledge and domain specific data.”

Veda sits as an orchestration layer above Certinia's Professional Services, Customer Success and Financial Management clouds. There are seven role-based agents (Estimation, Resourcing, Services Delivery, Financials, Customer Success, Health Monitoring and Customer Lifecycle) which coordinate work across the lifecycle. Each agent can call upon specialist agents and intelligent actions for the heavier lifting. Underneath, there is a unified data fabric and the Salesforce platform. There is also a trust and governance hub, which helps to bring together agentic and human collaboration.

The layered design is the most important thing about Veda. As Scott Hebner of theCUBE Research put it, “Veda is not treating the model as the product. It uses the model as the base, then adds the operational context and reasoning needed to make AI useful in knowledge-work-oriented workflows” (theCUBE Research). This distinction is doing real work, in a market that is overcrowded with generic copilots bolted onto generic data.

Raju Malhotra, Certinia's chief product and technology officer, made the engineering thesis explicit. The choice, he said, is to take advantage of “the best of probabilistic capabilities from an LLM, but not really relying 100% of them, matching it with a reasoning layer... based on our best practices and policies that we know work well in professional services.” Probabilistic is used where it is most helpful, and the deterministic approach takes over where consequences bite. This approach is a sensible posture for a domain in which a wrong staffing recommendation can badly affect downstream customer organisations as well as the professional services organisation.

Why a domain-specific engine matters now

IDC's May 2025 SaaSPath survey found that 48% of organisations plan to invest in AI-powered PSA, while 22% intend to replace their current PSA outright if generative AI is not part of the offering (IDC MarketScape, AI-Enabled PSA, 2025–2026). For a category known for sticky, multi-year contracts, it is very striking to see such a high rate of intended displacement. The same study positioned Certinia as a Leader and noted that “Certinia is one of the first vendors to productize autonomous PSA and build real use cases for Generative AI.”

The potential impact is phenomenal, as the PSA software market is forecast to grow from USD 14.41 billion in 2025 to USD 50.33 billion by 2034, a CAGR of 14.91% (Fortune Business Insights). Most of that growth is being pulled forward by AI-enabled functionality rather than traditional workflow automation. PSA has two characteristics that make it a strong proving ground for enterprise AI: the workflows are knowledge-intensive, and the unit economics are transparent enough to measure whether AI is paying back.

That second point is where Veda works hard to tackle the issues of integration found in the MIT and BCG numbers described earlier. Certinia has tied the launch to a specific impact framework, splitting value into staffing, delivery and customer success outcomes on top of the productivity already delivered by the core platform.

Headline figures from the early adopter cohort include an additional ten hours of capacity per resource manager per month, twenty hours for project managers, and a 1% lift in expansion and retention for customer success teams. The company argues that translates into roughly a 1.5% EBITDA improvement at large services firms. It's helpful to note these benefits in a market where the value of AI and agents is being scrutinised, as in the BCG and Gartner reports cited earlier.

Trust as architecture, not afterthought

Hebner's central observation is that Certinia “has elevated trust from a governance afterthought to an architectural principle.” Veda agents are bounded by Salesforce permissions and they sit behind the Einstein Trust Layer. This means that the Veda agents are governed by policies and rules embedded in the reasoning layer. Actions are auditable while users stay accountable, thereby addressing agent and human collaboration. Malhotra was direct: “Humans are not only in the loop, but the users are always in control... accountability in all the world of agentic is on humans and trust is about accountability.”

That matters, because Gartner's agent-washing critique applies to most of the market. Pointing a generalist agent at a system of record without policy guardrails can contribute toward hallucinations, which is precisely why so many proofs of concept stall. Certinia's bet is that services firms will pay a premium for an engine that sits inside its lane on top of a solid platform.

Open ecosystem and a sane commercial model

Veda is built on Salesforce Agentforce and supports Model Context Protocol (MCP), the open standard now being adopted across Salesforce, AWS, Box, IBM and Stripe (Salesforce Agentforce MCP support).

Pricing is a single Veda add-on at USD 30 per user per month on top of an existing Certinia user licence, with Agentforce flex credits covering consumption. It is unfussy by enterprise software standards and avoids the per-action obscurity that has made some agentic offerings difficult to budget.

What it means for the industry

Three things stand out. First, Veda is one of the clearer worked examples of what enterprise AI looks like when integration, not novelty, is the design constraint. As Uriah Hakala of early adopter Diabsolut put it, “the ability to have AI that understands our business context and acts on it, rather than simply reporting on it, is what makes Veda a meaningful step forward.”

Second, combining probabilistic reasoning with deterministic, policy-bound action is the right answer for any consequential knowledge-work domain, including sensitive areas such as legal, healthcare administration and finance operations. Certinia is signalling a template other vertical software vendors will be pressed to match.

Third, commercial discipline matters. With 95% of pilots failing and 40% of agentic projects expected to be cancelled, buyers are tired of paying for promises. Veda's per-user licence, transparent consumption tier and outcome-tied value framework reads as a deliberate response to that fatigue.

Harder questions remain. Can Certinia's reasoning layer keep pace as Salesforce evolves Agentforce? Will the early adopter ROI generalise across firms with different operating models? How quickly will the semantic layer mature toward causal reasoning, which Hebner identifies as the next frontier? These are the questions that determine whether Veda becomes the reference architecture for trusted agentic AI in services, or simply a strong release in a crowded year.

For now, the most useful summary is the one Certinia chose for its own launch slide: AI native to the full services journey, purpose-built to act, not just advise.