MindWalk (NASDAQ: HYFT): Enterprise AI Has Found Its Ontology Layer. Biology Needs One Grounded in Biological Function

ReefIQ licensing is metered on stored data and compute usage, unlike MindWalk’s historical fee-for-service projects and seat-based subscriptions.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

As data platforms and industry analysts converge on the "ontology layer" as the missing piece of agentic AI, MindWalk outlines why life sciences needs a biological context layer of its own, and why ReefIQ™, powered by HYFT® Technology, is that layer

AUSTIN, Texas--(BUSINESS WIRE)-- MindWalk Holdings Corp. ("MindWalk") (NASDAQ: HYFT), a Bio-Native AI company, today outlined why the ontology layer now being built across enterprise AI does not reach biological function, and why life sciences requires a biological context layer of its own. ReefIQ™, MindWalk's biological context layer for life sciences, supplies what a horizontal ontology cannot: a governed, model-agnostic representation of what biology does, available from the first day of a deployment rather than modeled in over years, enriching a pharmaceutical company's drug discovery data as it lands, and compounding its intellectual property with every program run on it. For investors, the distinction matters because the context layer, not the model, is where durable platform value accrues, and it is the layer MindWalk sells.

"The enterprise has discovered that an agent without an ontology guesses, and a guessing agent is a liability. That is exactly right, and it is exactly half of pharma's problem," said Dr. Jennifer Bath, President and Chief Executive Officer of MindWalk. "An enterprise ontology tells an agent what a purchase order, a patient or a clinical site is. It cannot tell an agent what a protein does, which regions of an antibody will tolerate modification, or why a program that failed three years ago matters to the one running today. The competitive question is no longer which model or agent you run; every model becomes a commodity. The question is what the model or agent runs on, because that is where the value ends up. In the enterprise, that is an ontology of the business. In biology, it has to be a representation of biology itself, built from what evolution preserved. The client's own programs, including the ones that failed, have to keep compounding inside it. That is what we built, and we are not aware of anyone else delivering it as a governed layer any model or agent can reason over."

The layer the market has converged on

Four years into the generative AI cycle, the enterprise has learned that large language models are becoming interchangeable ¹ and that deploying them inside a real organization disappoints far more often than the demonstrations suggest. MIT's 2025 State of AI in Business study put the share of enterprise generative AI pilots delivering no measurable return at 95 percent and located the cause not in model quality but in systems that do not retain context or integrate into workflows, ² and Gartner predicts that through 2026, 60 percent of AI projects unsupported by AI-ready data will be abandoned. ³ Few environments carry a higher cost for a confident wrong answer than drug discovery.

The market's answer is a new layer between an organization's data and the models and agents that use it, called an ontology layer, a semantic layer or a context layer. Whatever the label, it defines the entities an organization cares about, how they relate, what actions can be taken on them, and who may take them. Leading enterprise data platforms have built this layer for business data, the accounts, orders, sites and products their agents act on, ⁴ and Gartner now predicts that by 2030 universal semantic layers will be treated as critical infrastructure alongside data platforms and cybersecurity. ⁵ The market no longer debates whether the layer matters. It debates who owns it, in each domain.

Where the enterprise ontology stops

Every version of this layer describes the organization. The vendor supplies the machinery for objects, links, actions and governance, and the client models the content, deployment by deployment. That is the right design for an enterprise, where the entities are business entities. It is the wrong design for biology, where the meaning that matters is what a sequence or a structure does, and that meaning is not something a client can model in.

A protein sequence stored in an enterprise ontology is a string in a property. A structure is a file. Nothing in the layer knows that two sequences with no surface similarity perform the same function, or that a particular region of an antibody will not tolerate modification. That knowledge is not in the enterprise's data to be modeled; it is in biology, and it has to be brought to the data. Biomedical literature networks and vocabulary services annotate what is already known about a catalogued molecule; they cannot infer function for a sequence no one has characterized. Lab informatics systems register molecules and assays but do not encode function. Protein foundation models carry learned representations of function, but they are sold as models, not as a governed layer that any model or agent can reason over.

The layer for biology

ReefIQ occupies the intersection none of these reach. HYFT® Technology encodes biology as function-aware pattern-objects spanning sequence and structural biology. Refined over 20 years of curation, they form a continuously evolving biological representation of 660 million patterns and 25 billion relationships. When a client's discovery data lands in ReefIQ it is enriched at ingestion against that representation, so sequences, structures, assays, omics and program history become governed, queryable biological context. LensAI™, and any model or agent a client chooses to run, reasons over that same context, embedded in the client's own workflow, governance and human review. The client's own intellectual property compounds there. Client data remains the client's, inside their governance. Every program a client runs enriches every prior program, because the biological representation refines continuously and that refinement reaches data already on the platform. Programs that failed become queryable context for the programs that follow, and historical and ongoing work sit in one frame. The value a client builds compounds in the HYFT representation layer, not in any individual model or agent that runs on top of it.

MindWalk is a Bio-Native AI company built on more than 40 years of biology heritage, founded in 1983. It occupies the data structure and orchestration layer of biological AI, in the one domain where a confident wrong answer costs the most. ReefIQ has been commercially available since June 2026, and LensAI is in contracted, recurring arrangements with life sciences clients today.

Because ReefIQ is model-agnostic and governed, it is complementary to the enterprise data estate pharma already owns. Horizontal ontology platforms, cloud data platforms and lab informatics systems hold operational and instrument data; ReefIQ supplies the biological context those systems cannot generate. MindWalk expects the two layers to be deployed together as pharma moves agentic AI from pilots into discovery workflows.

What a context-layer deployment means commercially

In pharma, the work of building this layer is usually procured under an unglamorous name: data management. Unifying sequence, structure, assay and program data into a governed, AI-ready foundation is the first phase of every context-layer deployment, and it is the same work the enterprise now recognizes as building the ontology layer. It is also where the position is won. Once a client's discovery data is enriched and governed in ReefIQ, each subsequent program builds on that context instead of recreating it, and the models and agents running on it can be changed without disturbing it. A ReefIQ deployment is therefore structured differently from the fee-for-service projects and seat-based subscriptions that have made up most of MindWalk's historical revenue. Licensing is metered on the data a client holds in the platform and the compute it runs against that data, so revenue scales with discovery activity rather than headcount. The context a client builds there cannot be exported to another vendor without starting over, which is why these relationships are multi-year by design. A deployment typically begins with data management and widens into analytical tools and custom models running on the same context. ReefIQ agreements should not be benchmarked against MindWalk's historical LensAI agreements for any of these reasons.

About MindWalk Holdings Corp.

MindWalk Holdings Corp. (NASDAQ: HYFT) is a Bio-Native AI company building the BioIntelligence infrastructure that life sciences AI and agentic AI require, integrating AI, data, and advanced wet lab capabilities into one connected discovery ecosystem. At its core is HYFT® Technology, a proprietary, function-aware representation of biology. Its HYFT pattern-objects span sequence and structural biology and, refined over 20 years of curation, form a continuously evolving biological representation of 660 million patterns and 25 billion relationships. This enriched biological representation is the architecture behind ReefIQ™, the biological data substrate that provides context for life sciences, enriching data at ingestion and growing more valuable with each program run on it, and LensAI™, the reasoning and application layer for target discovery, candidate diligence, portfolio decision support, and the agentic AI workflows pharmaceutical companies are now deploying. By design, value compounds in this HYFT representation layer, not in any individual AI model that runs on top of it.

References

[1] Stanford Institute for Human-Centered Artificial Intelligence, 2025 AI Index Report, Technical Performance chapter, April 2025. The performance gap between the top-ranked and tenth-ranked model fell from 11.9 percent to 5.4 percent in one year; the gap between open-weight and closed models narrowed from 8 percent to 1.7 percent. hai.stanford.edu/ai-index/2025-ai-index-report

[2] MIT NANDA, The GenAI Divide: State of AI in Business 2025, July 2025. Based on 52 executive interviews, a survey of 153 leaders and 300 public deployments.

[3] Gartner, "Lack of AI-Ready Data Puts AI Projects at Risk," press release, February 26, 2025. Gartner.com/en/newsroom

[4] Public product documentation: Palantir Foundry Ontology (object, link and action types); Snowflake Cortex Agents, ontology-grounded, June 2026; Databricks Genie Ontology, June 2026; Microsoft Fabric IQ ontology, preview.

[5] Gartner, "Gartner Announces Top Predictions for Data and Analytics in 2026," press release, March 11, 2026. By 2030, universal semantic layers will be treated as critical infrastructure alongside data platforms and cybersecurity; 44 percent of data and analytics leaders have implemented semantic layers and a further 48 percent plan to by 2027. gartner.com/en/newsroom

Forward-Looking Statements

This press release contains forward-looking statements within the meaning of applicable United States and Canadian securities laws. Forward-looking statements in this press release include, without limitation: statements regarding the evolution of enterprise and life sciences AI toward ontology, semantic or context layers; the expected role, capabilities and commercial model of ReefIQ™ and LensAI™; the expectation that the value of ReefIQ to clients increases with use; and the expectation that horizontal and biological context layers will be deployed together. Forward-looking statements are based on management's current expectations and assumptions and are subject to known and unknown risks, uncertainties, and other factors that may cause actual results to differ materially, including: the pace and scale of market acceptance and adoption of ReefIQ™ and LensAI™ by life sciences clients; the evolution of the enterprise and life sciences AI market, including the strategies of larger, better-resourced data platform, ontology and semantic-layer vendors and the possibility that such vendors extend their offerings into biological data; statements regarding the continued standardization of AI model capability, which may not develop as anticipated; the expectation that horizontal and biological context layers will be deployed together, which may not occur; the ability of MindWalk to maintain and defend its competitive positioning and intellectual property, including pending trademark and patent applications; the technical performance of AI-based discovery methods and of the underlying compute infrastructure; the build-out and certification of compliance capabilities for regulated workloads; the ability to convert engagement into contracted, recurring arrangements; the structure, duration and economics of ReefIQ agreements, which may differ from those described; competition; regulatory determinations; and capital markets conditions. Additional information is available in MindWalk's Annual Report on Form 20-F and other filings on SEDAR+ (sedarplus.ca) and EDGAR (sec.gov/edgar). Except as required by law, MindWalk undertakes no obligation to update any forward-looking statement. Nothing in this press release constitutes investment advice. This press release does not constitute an offer to sell or the solicitation of an offer to buy any securities.

Trademarks

HYFT® is a registered trademark of MindWalk Holdings Corp. LensAI™ and ReefIQ™ are trademarks of MindWalk Holdings Corp. or its subsidiaries; ReefIQ™ registration is pending. All other trademarks are the property of their respective owners.

View source version on businesswire.com: https://www.businesswire.com/news/home/20261006642943/en/

Investor Relations Contact
Louie Toma, CPA, CFA
Managing Director, CoreIR
investors@mindwalkai.com

Source: MindWalk Holdings Corp.

Key Terms

model-agnostic technical

An approach, tool, or test that works with any predictive or analytical model rather than depending on a single algorithm or vendor. For investors, model-agnostic methods let you compare, audit, and trust results across different forecasting systems—like using a universal charger that fits many phones—reducing reliance on one provider, making performance easier to validate, and lowering the risk that a single model’s quirks drive investment decisions.

semantic layer technical

A semantic layer is a translation layer that turns raw, technical data into consistent, business-friendly terms and metrics so non-technical users can ask questions and get reliable answers. For investors it matters because it makes financial reports and analytics comparable and easier to trust—like a common dictionary that ensures everyone is using the same definition of revenue, costs or customer counts, reducing confusion and speeding decision-making.

protein foundation models technical

Large machine-learning models trained on massive databases of protein sequences and structures to serve as a general-purpose foundation for many downstream protein tasks. They learn statistical patterns of amino-acid sequences and structural motifs during a self-supervised pretraining phase, produce numeric embeddings or predicted properties (such as structure, contacts, or functional annotations), and are then fine-tuned or adapted to specific uses like predicting protein folding, designing new sequences, estimating effects of mutations, or screening candidates. Like other foundation models, they provide reusable representations that accelerate many workflows but produce probabilistic predictions that require experimental validation and domain-specific evaluation.

omics medical

Omics is the study of large sets of biological information — for example all genes, proteins or chemical products inside cells — treated like a complete inventory rather than a single item. Think of it as reading the full instruction manual or catalog for a living system; that broad view helps identify new drug targets, diagnostic tests or efficiency gains in biotech. Investors care because omics-driven discoveries can create new products, revenue streams and competitive advantages, but also carry technical, development and regulatory risks.