Every year, companies make high-stakes decisions using tools that can leave gaps: consulting recommendations that may be difficult to audit, dashboards focused on past performance, or AI-generated answers whose underlying evidence is not always clear. KNOWIDEA, a Toronto-based decision intelligence company founded in September 2025, built VeriPIE to close that gap. Its new release, VeriPIE 1.0, is designed to replace that opinion with a prediction a leader can defend line by line, back to the first data point.
What VeriPIE actually does
VeriPIE connects to the systems a company already runs, including ERP, finance, operations and CRM platforms, and layers in external market benchmarks. From there, it looks for the variables that actually drive a given number, forecasts what happens next under each option, and assigns an expected value to each outcome. Before any of that reaches a decision-maker, the engine verifies every number it produced, tracing each one back to its source system, table, rows and dates.
KNOWIDEA uses NeuroSymbolic AI: neural networks that learn patterns from a company’s data, paired with symbolic logic that reasons in explicit, checkable rules. The combination is designed to help VeriPIE produce recommendations supported by verifiable calculations where possible, while clearly identifying what it cannot verify.
The engine runs the same path on every question: it gathers company and market information, isolates the handful of variables that explain the change, predicts the outcome of each option, weighs those predictions against the company’s risk tolerance, and hands the executive a recommendation with any remaining human judgment calls clearly marked. In one example KNOWIDEA points to, a 4.2% gross margin decline was split into three drivers, price realization, cost to serve and product mix, with three other candidate explanations ruled out before a recommendation was made.
Already running inside oil and gas, finance and manufacturing
Parts of VeriPIE are already in production. According to KNOWIDEA, three decisions at a large oil and gas producer—covering acquisitions and integration, spending and budget control, and capital programme planning—represented several million Canadian dollars in first-year value, or several times the software’s licensing cost. At a global financial institution, the company also reports that VeriPIE converts natural-language prompts into SQL queries for complex enterprise databases, with accuracy modestly above a reported baseline on two widely used text-to-SQL benchmarks.
Every major decision is a prediction. The question is whether you can prove it before you bet the company on it. Consultants give you an opinion, dashboards give you the past, and AI gives you an answer you cannot check. VeriPIE gives you the future with the working shown. – Yatharth Sejpal, co-founder and CEO, KNOWIDEA
Sejpal, who studied organizational psychology and business at the University of Waterloo and previously worked on large-scale M&A engagements at Mercer and on internal performance systems at TD Bank, built an early prototype himself before recruiting Brian Zhengyu Li, who is a PhD candidate in Computer Science from Georgia Tech, as technical co-founder and CTO.
Prediction without verification is a guess with better formatting. VeriPIE makes the forecast, then proves every part of it that can be proven, and names the part that cannot. That is what lets an executive act on it. – Brian Zhengyu Li, co-founder and CTO, KNOWIDEA
A fast climb since launch
KNOWIDEA says it passed $1 million in revenue within six months of launching, is growing 350% quarter over quarter, and now serves customers in Canada, the United States, Australia and India. That growth trajectory tracks with what KNOWIDEA has said publicly before this launch: the company reported crossing $500,000 in annual recurring revenue within its first four months, with roughly 60% of that early revenue coming through warm introductions and 30% through partnerships, rather than a dedicated outbound sales motion.
Positioned alongside consultants, dashboards and AI tools, not replacing them
KNOWIDEA is explicit that VeriPIE sits on top of the systems a company already owns rather than replacing them. It does not become the ERP or the BI layer; it turns the data already inside those systems into a verified prediction. The company’s pitch is less about speed for its own sake—a consulting engagement that takes eight to twelve weeks, a dashboard that answers in seconds but primarily describes the past—than about offering output whose supporting calculations can be checked by code.
KNOWIDEA frames the comparison around four questions it says most tools cannot all answer at once: does it look forward rather than just report what already happened, is every number traceable back to its source, are claims checked by code rather than by another model, and does the system say plainly what it cannot prove. A strategy consulting engagement looks forward once, at the end of an eight-to-twelve-week process, and its recommendation reflects judgment rather than a traceable number. A BI dashboard reports the past in seconds but does not forecast. A generic AI prediction tool answers quickly but cannot always show where any given number came from, let alone flag what it does not know. VeriPIE’s answer to each of those four questions, the company says, is yes.
For a business considering it, KNOWIDEA’s own answer to how it fails is as telling as how it works: when the underlying evidence is not there, VeriPIE does not issue a verified recommendation. It names the missing input instead, so the gap is visible before, not after, a decision gets made.
To learn more about VeriPIE or schedule a 30-minute working session focused on a business decision, visit knowidea.dev.