A 35-Year AI Veteran's New Startup Is Wiring Determinism Into Enterprise LLMs

The early-stage startup, backed by operators from Chargebee and Capillary, is betting that regulated industries need more than just another model wrapper.

About CogniSwitch

Published

You open a chat interface embedded in your company’s patient portal, asking it to summarize a recent call with a concerned family member. The AI assistant responds with confidence, citing specific clinical codes and care recommendations. The problem is, the conversation never happened. The model, left to its own probabilistic devices, has confabulated an entire interaction, complete with plausible but fabricated medical advice. This is the moment of vertigo that CogniSwitch is built to prevent.

Founded in 2022 and based in Secaucus, New Jersey, CogniSwitch operates in the narrow, high-stakes alley between generative AI’s promise and its peril. Its product is not another chatbot or a fine-tuned model, but an infrastructure layer designed to sit beneath them. The company converts unstructured enterprise knowledge,SOPs, call transcripts, regulatory documents,into structured, governed inputs, aiming to produce what it calls “deterministic, auditable intelligence” for regulated industries like healthcare and financial services [CB Insights, April 2025]. In a landscape chasing model size and speed, CogniSwitch is betting on certainty.

The architecture of accountability

The technical wedge is a combination of knowledge graphs and what the company terms neuro-symbolic verification [cogniswitch.ai/contextops, retrieved 2026]. Instead of relying solely on a large language model’s next-token prediction, the platform first organizes a company’s internal knowledge into a structured semantic map. When a query comes in, the system grounds the LLM’s response in this verified knowledge base, checking outputs against a symbolic layer of rules and facts. The goal is to reduce the hallucinations and errors that make enterprise adoption in sectors like pharmaceuticals or banking a compliance nightmare [SaaSBOOMi, February 2024]. It’s a bet that the future of enterprise AI isn’t about smarter models, but about more accountable ones.

This focus is crystallized in the company’s open-source offering, KL4A (Knowledge Layer for AI Agents), which is designed to extract obligation-shaped claims from standard operating procedures and tether them to source text [cogniswitch.ai/blog/launching-kl4a-knowledge-layer-for-agents, retrieved 2026]. It’s a building block meant for developers constructing agentic workflows where a wrong decision carries real cost. The company says it can structure 100% of patient conversations into clinical intelligence for care gap analysis, a claim pointing squarely at its telehealth ambitions [cogniswitch.ai/telehealth, retrieved 2026].

A founder with four decades of context

The ambition is backed by a founder whose career arc mirrors the evolution of the problem itself. CEO Dilip Ittyera has approximately 35 years of experience building knowledge-management systems and was working on AI and expert systems in the 1980s [cogniswitch.ai/about, retrieved 2026]. His resume includes senior engineering roles at NIIT, Hexaware, and Zensar, and he once scaled a business unit from zero to $25 million in two years as CTO [cutshort.io/company/coinswitch-98-SguSAP7P, retrieved 2026]. This is not a founder chasing a trend; it’s one returning to a lifelong theme with new tools.

He is joined by co-founders Joshua Thomas, the CTO who leads the verification architecture and has a background in computational linguistics, and Vivek Khandelwal, the Chief Business Officer and an IIT Bombay graduate who previously built the marketing automation platform iZooto [cogniswitch.ai/contextops, retrieved 2026]. The team presents a blend of deep technical AI pedigree and scaled commercial experience.

Role Name Key Background
Founder & CEO Dilip Ittyera ~35 years in knowledge systems; former CTO at Zensar; AI work since the 1980s.
Co-Founder & CTO Joshua Thomas Leads neuro-symbolic verification; decade+ in AI and computational linguistics.
Co-Founder & CBO Vivek Khandelwal IIT Bombay grad; previously built iZooto from scratch to market leadership.

Early traction and angel backing

Public traction metrics are sparse, but the company has disclosed initial customers including a large professional body, a telecom services company, and an electronic ecommerce player [SaaSBOOMi, January 2024]. Its most recent public development is the launch of a podcast, ContextOps, in August 2026, suggesting a focus on community and category building [Apple Podcasts, August 2026].

Financing appears lean and operator-heavy. The clearest public record shows a $100,000 seed raise in September 2022 [CB Insights, April 2025]. The company’s backers are a roster of seasoned SaaS operators, including Vijay Rayapati (AtomicHQ), Krish Subramanian and Rajaraman Santhanam (Chargebee), Aneesh Reddy (Capillary Technologies), and the Upekkha SaaS Accelerator [cutshort.io/company/cogknit-semantics-pvt-ltd-kfqzzF1R]. This suggests a bet on founder capability and market timing rather than a capital-intensive initial burn.

The compliance moat and its challengers

The company’s potential moat is the complexity of compliance itself. Building a knowledge graph that accurately reflects a hospital’s protocols or a bank’s trading regulations is not a weekend project. It requires deep understanding of both the domain and the architecture of trust. CogniSwitch’s early focus on healthcare and financial services,two of the most heavily regulated sectors,positions it in markets where the cost of AI error is existential, not just inconvenient.

Yet, the path is lined with credible counterfactuals. The market for AI governance and safety is becoming crowded, with approaches ranging from vector databases to dedicated evaluation platforms. The core risk for CogniSwitch is that its neuro-symbolic, knowledge-graph approach could be seen as too heavy or specialized compared to lighter-touch retrieval methods that are "good enough" for many use cases. Furthermore, the company’s reported $100,000 in seed funding, while sufficient for a lean start, is a fraction of the war chests available to well-funded incumbents or new entrants in the AI infrastructure space.

The company’s answer likely lies in specificity. Its narrative isn’t about general AI safety, but about enabling mission-critical applications where an audit trail is non-negotiable. The open-source KL4A project is a clever wedge to get developers hands-on with its methodology. And the founder’s four-decade context is a form of institutional knowledge that pure-play AI startups lack.

What to watch in the next act

The next twelve months will test whether this early operator-backed bet can translate into scaled enterprise validation. Key milestones to watch include the naming of a flagship healthcare or financial services customer, a move beyond the initial seed funding, and any expansion of the open-source KL4A community. The company’s podcast initiative indicates a focus on mindshare, but the real proof will be in signed contracts that detail how its infrastructure prevented a costly hallucination.

Ultimately, CogniSwitch is answering a cultural question that every enterprise is now grappling with implicitly: in the rush to deploy AI, where do we draw the line between useful automation and unacceptable risk? Its entire proposition is that the line is drawn not in the model's parameters, but in the infrastructure that feeds it. The product is a argument that for AI to become truly operational, it must first become accountable. The bet is that in boardrooms facing down HIPAA and FINRA, accountability will be the only feature that matters.

Sources

  1. [CB Insights, April 2025] Company Comparison: CogniSwitch vs. Glean | https://www.cbinsights.com/compare/cogniswitch-vs-glean-2
  2. [SaaSBOOMi, February 2024] What next in ChatGPT? | https://saasboomi.org/newsletter/what-next-chatgpt/
  3. [cogniswitch.ai, retrieved 2026] ContextOps | https://cogniswitch.ai/contextops
  4. [cogniswitch.ai, retrieved 2026] About | https://cogniswitch.ai/about
  5. [cutshort.io, retrieved 2026] Company Recruiting Profile | https://cutshort.io/company/coinswitch-98-SguSAP7P
  6. [SaaSBOOMi, January 2024] Riding the Gen AI Wave by Dilip Ittyera, CogniSwitch | https://saasboomi.org/saas/growth/riding-the-gen-ai-wave-by-dilip-ittyera-cogniswitch/
  7. [cogniswitch.ai/blog, retrieved 2026] Launching KL4A (Knowledge Layer for Agents) | https://cogniswitch.ai/blog/launching-kl4a-knowledge-layer-for-agents
  8. [Apple Podcasts, August 2026] ContextOps podcast | https://podcasts.apple.com/us/podcast/contextops/id6793630750
  9. [cutshort.io, retrieved 2026] Investor Backing Details | https://cutshort.io/company/cogknit-semantics-pvt-ltd-kfqzzF1R

Read on Startuply.vc