cms.quaeris.ai

Category: Data Analysts

  • The four pillars of Governed Agentic Analytics: audit, semantic, agentic, BYOM

    The four pillars of Governed Agentic Analytics: audit, semantic, agentic, BYOM

    Four capabilities separate a governed agentic platform from a chatbot bolted onto a warehouse. Remove any one of them and the structure fails in a specific, predictable way. This piece defines each pillar, explains what breaks without it, and shows how the four reinforce one another when they are built as one system rather than assembled from separate tools.

    Pillar one: the audit trail

    Every question asked of an agent should produce two artefacts: the answer, and the record of how the answer was produced. The audit pillar means logging the full chain on every interaction: who asked, what they asked, which SQL the agent generated, which tables and rows it touched, which model produced the reasoning, and what came back. This is the prompt-level audit trail, and it is the difference between an answer you can act on and an answer you can defend.

    Without it, every agent response is an orphan. When a number in a board deck is challenged three months later, there is no way to reconstruct where it came from. Internal audit cannot test the system. Regulators in banking, insurance, and healthcare cannot accept its outputs as evidence. The absence of an audit trail does not make an agent less useful day to day. It makes the agent indefensible on exactly the day it matters most.

    Pillar two: the semantic layer

    An agent is only as accurate as its understanding of what the business means by its own words. The semantic pillar supplies that understanding: certified definitions of metrics, dimensions, and business terms that the agent consults before it writes a line of SQL. When someone asks about net revenue, the agent does not guess which of the fourteen revenue-adjacent columns in the warehouse to use. It uses the one the business certified.

    QuaerisAI’s Smart Semantic Layer adds a property that distinguishes it from manually authored alternatives: it learns. When an analyst corrects a definition or refines a query, the correction feeds back into the layer, so the next person asking a similar question benefits from it. Without a semantic layer of some kind, an agent hallucinates business logic, and it does so fluently, which is worse than failing loudly. Confidently wrong answers are the fastest way to destroy trust in an analytics deployment.

    Pillar three: the agentic layer

    The agentic pillar is what makes the platform more than search. Real analytical questions are rarely answered by one query. Why did margin dip in the northeast last quarter is a plan: pull the margin trend, decompose it by product line, check input costs, compare against the same period last year, and assemble a finding. The agentic layer is the capability to decompose a question into steps, execute them across structured and unstructured sources, and synthesise a result, autonomously.

    Without it, the platform is a text-to-SQL translator: useful for single-shot lookups, incapable of investigation. The user gets a number when they needed an explanation. The distinction matters commercially because the investigation work is where the senior-staff hours actually go. Based on QuaerisAI customer deployments and published materials, agentic deployment in audit engagements has reduced standard engagement time from 170 hours to 70 hours, a 59 percent reduction, precisely because the multi-step work is where the time was.

    Pillar four: BYOM

    Bring your own model means the platform runs on the language model the enterprise chooses, whether from Anthropic, OpenAI, Google, or Meta, rather than binding the customer to a single vendor’s model. This pillar looks like a procurement detail until the first time it is tested. A compliance team mandates a specific model for data-residency reasons. A model vendor changes pricing. A new model release outperforms the incumbent on SQL generation. In a BYOM architecture, each of these is a configuration decision. In a locked architecture, each is a re-platforming project.

    BYOM also changes the negotiating position. An enterprise that can switch models holds pricing power an enterprise that cannot switch has surrendered. And for regulated buyers, model choice is increasingly a compliance requirement in its own right, because model provenance is becoming part of what auditors and regulators expect organisations to document.

    Why the pillars only work together

    The four pillars are not a feature list. They are load-bearing in combination. The audit trail is only meaningful if the semantic layer makes answers deterministic enough to be worth auditing. The semantic layer only pays for itself when an agentic layer is asking it thousands of questions a day. The agentic layer is only deployable in the enterprise because the audit trail makes its autonomy accountable. And BYOM keeps the whole structure durable as the model market shifts underneath it.

    This is why assembling the equivalent from point solutions rarely works. A logging tool, a metrics store, an agent framework, and a model gateway from four vendors produce four integration seams, and governance fails at seams. Governed Agentic Analytics is one architecture, and the four pillars are its definition.

    Frequently asked questions

    Which pillar should an evaluation test first?

    The audit trail, because it is the hardest to retrofit and the easiest to verify in a demo. Ask the vendor to show the complete record for a single query: the prompt, the generated SQL, the tables touched, the model used, and the result. If that record does not exist or requires engineering effort to produce, the other three pillars are standing on sand.

    Is a semantic layer still necessary if the model is very capable?

    Yes. Model capability improves general reasoning, but no model can know that your finance team excludes intercompany transfers from net revenue, or that fiscal Q1 starts in February. Those are facts about your business, not facts about the world, and they have to be supplied by a governed definition layer. Stronger models make the semantic layer more valuable, not less, because they execute certified definitions more reliably.

    Does BYOM mean the customer manages model infrastructure?

    No. BYOM in the QuaerisAI architecture means the customer selects which model powers the platform, from providers including Anthropic, OpenAI, Google, and Meta, and can change that selection. The platform handles the integration. The customer holds the choice.

  • Your Analytics Team Is Not Slow. Your System Is Fragmented.

    Your Analytics Team Is Not Slow. Your System Is Fragmented.

    A majority of analytics teams are not underperforming. They are working at full capacity inside a system that was never designed to scale with the volume of questions a modern business generates. The problem is not the people. It is the architecture built around them, and understanding that distinction is the first step to fixing it.

    The Repeatable Pattern

    When analytics teams fall behind, the instinctive responses tend to follow a familiar sequence. Leaders call for more dashboards. Procurement teams evaluate new tools. Hiring managers post for more analysts. None of these address the actual problem.

    Most organisations already have more data than they can use, more dashboards than any team trusts, and more incoming requests than a reasonably sized analytics function can clear. And yet, despite all of that investment, the same failure mode keeps appearing: a question gets asked, and the answer does not arrive fast enough for the decision it was meant to inform. That lag, multiplied across every team and every decision cycle, is where the real cost lives.

    What Analysts Actually Do All Day

    There is a large gap between what analytics teams are hired to do and what they actually spend their time on. The honest version of the job description at most organisations looks something like this: answering the same question from a different stakeholder for the twelfth time, rebuilding the same business logic in a slightly different format, chasing down metric definitions that should have been standardised long ago, sitting in meetings explaining numbers that the audience does not fully trust, and working through a request backlog that adds items faster than it clears them.

    This is not insight generation. It is ticket management with a SQL layer on top. The analysts doing this work are not slow or untalented. They are allocated to work that should not exist in the form it currently does, and the best ones feel it most acutely because they can see clearly what they would rather be doing.

    Where the Bottleneck Actually Is

    When decisions take too long, the analytics team is usually the first place leadership looks. This is the wrong diagnosis. The bottleneck is not the analyst. It is the system that requires every business question to travel through a human intermediary before it can be answered.

    The standard path is predictable: a business user has a question, they submit a request, it joins a queue, it eventually gets prioritised, an analyst works on it, and the answer arrives days later, sometimes after the window to act on it has already closed. The system routes every question through a person regardless of whether that question actually requires one. When you build a structure where thinking does not scale, the problem is not the thinkers.

    Why AI Creates Anxiety on Analytics Teams

    When AI tools arrive with the promise of instant answers, the reaction from analytics professionals is often scepticism or outright resistance. This is a rational response. The framing of AI as something that can answer questions automatically sounds, from the analyst’s perspective, like a straightforward argument that the analyst’s role is redundant.

    The more accurate framing is different. If AI takes over any portion of the analytics function, the work it displaces is the work that analytics teams find least valuable and most draining: the repeat questions, the low-complexity extractions, the formatting requests, the work that should never have required a credentialled analyst in the first place. The genuine risk for analytics professionals is not AI replacing them. It is staying inside a system that continues to allocate their time to work that a machine could handle, while the higher-value work they are capable of never gets done.

    The Shift Most Organisations Have Not Made

    The more useful frame for AI in analytics is this: it does not replace analysts, but it does expose how much of what currently occupies an analyst’s day should never have been their job. When the repeat questions are handled automatically, when definitions are standardised and no longer require manual interpretation, when routine extractions no longer require a ticket, what is left is the work that actually requires human judgment, contextual knowledge, and strategic thinking.

    The practical outcomes of this shift are measurable. Request backlogs shrink. Metric definitions stabilise because they are enforced consistently rather than reinterpreted per request. Business users get faster access to answers they can trust. And analysts, freed from the low-value queue, can focus on the analysis that actually influences decisions rather than reporting on decisions that have already been made.

    The Two Options Available

    When an analytics team is consistently overwhelmed, there are two available responses. The first is to add headcount and continue operating the same system, which distributes the load without addressing the structural cause. The second is to fix the system so that the team’s capacity is allocated to work that requires their expertise.

    Most organisations choose the first option because it feels more immediate and more controllable. It also explains why the problem tends to persist regardless of how many analysts are added. The queue expands to meet the available capacity, and the cycle continues.

    What Actually Changes

    When the structural problem is addressed and analysts are no longer the required intermediary for every data question, the nature of their contribution shifts in a meaningful way. They become the owners of data definitions and business context rather than the executors of routine requests. They move from explaining last quarter’s numbers in meetings to helping shape how next quarter’s decisions get made. Their value does not decrease. It becomes more visible because it is being applied to work where it is genuinely differentiated.

    This is also why most AI implementations in analytics fail to deliver on their stated promise. They focus on replacing outputs, generating automated reports and pre-built dashboards, without addressing the underlying system that created the bottleneck. Automating the wrong workflow at higher speed does not fix the problem. It accelerates it.

    What the Goal Actually Was

    The dashboards were never the goal. The analysts were never the goal. The goal was always faster, more reliable decisions made by the people responsible for them. If the current system cannot consistently deliver that, the answer is not to add more of the same components. It is to change what the system asks of the people inside it.

    Fixing the structure around an analytics team is harder and slower than adding headcount. It is also the only approach that actually resolves the problem rather than deferring it.