Deep Signals

**Product and the concrete problem it solves.** Deep Signals addresses the highest-cost decision in many oil and gas exploration programs: whether a subsurface prospect is sufficiently credible to justify additional surveys and a well. Seismic interpretation can identify structures and geological patterns, but a prospect that looks attractive in a conventional interpretation can still fail because the reservoir does not contain the expected fluids, connectivity, or deliverable volume. Deep Signals presents its product as a physics-based AI system that predicts whether a potential reservoir will succeed or fail before drilling begins. The practical output is a pre-drill Go or No-Go signal for exploration teams, rather than another generic dashboard or a model that only classifies historical imagery. The company and its ecosystem partners frame the economic problem around dry holes, with a claimed global annual loss of approximately $67 billion and individual wells that can cost more than $200 million; these figures are public opportunity framing, not independently audited company performance. If the product works across relevant basins, it could help operators rank prospects, reduce avoidable drilling, and direct scarce geoscience and capital resources toward higher-confidence targets.

**Core technology and how it actually works.** Public descriptions identify a simulation engine that goes beyond the traditional elastic-Earth model. According to a National Center of Blue Economy spotlight, founder Uri Kushnir and the team model how fluids inside rock affect the seismic signals observed at the surface, then combine physics-accurate synthetic data with real field data to train the AI. The intended insight is the reservoir's true seismic signature: not simply whether a reflector or structure appears in a volume, but whether the observed response is consistent with the fluid and rock conditions that make a reservoir commercially viable. This approach is technically meaningful because real labeled examples of successful and failed reservoirs are scarce, expensive, and biased toward drilled prospects, while a physically constrained simulator can generate controlled examples for model development. The public record does not disclose the network architecture, seismic modalities, survey geometry, preprocessing pipeline, uncertainty calibration, accuracy by basin, or the amount and provenance of proprietary field data. Those unknowns matter. A physics-informed workflow can improve generalization, but it does not automatically solve the transfer from synthetic geology to noisy, basin-specific measurements. The core diligence question is whether the simulator captures enough of the relevant fluid, rock, acquisition, and processing variability to improve decisions on new prospects.

**Market, customers, and go-to-market.** Deep Signals is positioned for the business-to-business oil and gas exploration market, with public ecosystem material specifically connecting the company to offshore energy projects. The likely initial users are exploration and subsurface teams at operators, license holders, and geoscience service providers that already commission seismic surveys and decide which prospects proceed to appraisal or drilling. A plausible wedge is a project-level analysis of existing seismic data, followed by integration into prospect ranking and portfolio review; this is an inference from the product's stated decision point rather than a disclosed pricing model. The company could eventually sell software access, paid interpretation projects, or an enterprise workflow to operators, but public sources do not confirm the commercial model. Offshore exploration is a consequential beachhead because drilling and environmental exposure are high, and because a better pre-drill screen can influence both capital allocation and the number of intrusive activities. The company is also being developed within Israel's blue-economy ecosystem, where energy, marine data, and deep-tech commercialization overlap. It must still prove that geoscience buyers will trust an AI recommendation in a high-liability decision, that the workflow fits existing interpretation software, and that the product delivers value quickly enough to survive long exploration procurement cycles.

**Traction, funding, and third-party validation.** The available evidence places Deep Signals at a credible but very early commercialization point. IVC lists Deep Signals Ltd as established in 2025, based in Haifa, at R&D stage with five employees, and identifies Dr. Uri Kushnir as CEO and founder. The same profile records a March 2026 R&D grant and an earlier non-equity assistance entry connected to a municipal accelerator, while the Israel Innovation Authority's 2026 invested-companies list places Deep Signals Ltd in Energy-tech and Construction Tech with the Startup Fund as its last invested program. HiCenter Ventures is publicly reported to have invested in Deep Signals, and Israel Hayom describes the investment in connection with the company's AI analysis of seismic data. A National Center of Blue Economy post says the company was opening a pre-seed round, but no amount, lead investor, closing date, revenue, named customer, production deployment, or independent performance benchmark is publicly confirmed in the sources reviewed. These signals validate institutional interest and an active company-building process; they do not establish commercial traction. The next meaningful evidence would be a disclosed field partner, retrospective results on held-out wells, a live operator pilot, or independent confirmation that the system changes drilling decisions without increasing unacceptable false negatives.

**Founders and team background.** Uri Kushnir is publicly identified as Deep Signals' CEO and founder. His public profile places him in Haifa and lists a Technion PhD, along with work touching geophysics, seismic interpretation, offshore pipeline geophysical surveys, and AI applied to hydrography through CAMERI and SEAL AI. That combination is unusually relevant to the company's stated thesis: the product requires both a physical understanding of how subsurface conditions generate signals and the ability to translate those signals into an operational decision. The public Blue Economy material also refers to Kushnir and a team building the simulation engine, but it does not provide a complete executive roster or a detailed division of technical responsibilities. IVC independently corroborates his CEO and founder role, which is stronger than relying on a single social profile. The team appears to have domain-founder fit, but five employees is a very small base for a product that must combine reservoir physics, seismic processing, machine learning, software engineering, customer integration, and enterprise sales. Diligence should establish who owns the core modeling and data pipeline, whether the company has access to representative labeled field data, how university or prior-employer intellectual property is handled, and whether it can recruit experienced exploration customers and geophysicists as design partners.

**Competitive dynamics.** Deep Signals competes with established subsurface software and interpretation workflows rather than only with other AI startups. SLB offers large-scale reservoir characterization, seismic interpretation, and digital subsurface tools embedded in operator workflows. Halliburton Landmark provides integrated earth-modeling and reservoir software used by exploration and production teams. Baker Hughes combines oilfield services with geoscience and reservoir workflows, giving it access to customer data and field operations. CGG is a major geoscience provider with seismic imaging, interpretation, and subsurface consulting capabilities, while TGS supplies geoscience data, imaging, and interpretation services. Operators can also keep the decision internal using incumbent interpretation software, expert geoscientists, and probabilistic prospect economics. Deep Signals' possible edge is a focused combination of fluid-aware simulation, physics-constrained synthetic data, and a clear pre-drill success/failure decision. That focus could be easier to adopt than a broad digital-twin platform if it produces an auditable probability and explains which seismic evidence drives the result. It is not yet a demonstrated moat. Incumbents have proprietary datasets, established procurement relationships, interpretation archives, and the ability to bundle software with surveys and services. The decisive test is independent performance on new basins and prospects, with uncertainty and economic value measured against the operator's existing workflow.

**Defense, security, and resilience relevance.** Deep Signals' primary disclosed market is commercial energy, but its core capability has a credible strategic dual-use pathway through subsurface sensing and resource resilience. The same physics-based interpretation of seismic signals could inform underground infrastructure assessment, geological-hazard mapping, carbon-storage characterization, groundwater studies, and the detection or characterization of tunnels, voids, or other subsurface anomalies. Those extensions are plausible applications of the underlying signal-and-physics stack, not publicly confirmed Deep Signals products or defense contracts. Energy security is the clearest present relevance: better prospect screening can reduce dependence on wasteful drilling, improve the economics of domestic or allied offshore resources, and make critical energy projects more resilient to capital and environmental constraints. The technology could also support strategically important ports, pipelines, storage sites, and other installations where understanding subsurface conditions reduces operational or safety uncertainty. There is no public evidence of a military customer, Ministry of Defense program, classified deployment, security certification, or fielded defense capability. The dual-use assessment is therefore deliberately moderate. It is stronger than a forced analogy because geophysical inference is inherently transferable across energy, infrastructure, and security problems, but the security value will depend on validated anomaly detection, secure handling of sensitive survey data, robust operation in contested or degraded information environments, and a specific non-commercial deployment partner.

**Growth stage, trajectory, and key diligence risks.** Deep Signals is best classified as early. The company has a named founder, a Haifa base, a 2025 establishment date, an R&D-stage profile, government and accelerator support signals, ecosystem investment, and a publicly described technical mechanism. It remains pre-scale: the public record does not disclose a closed financing amount, commercial revenue, named paying customers, a completed field deployment, or an independently reproduced accuracy result. Its next trajectory should be measured by a narrow and falsifiable validation path: retrospective testing on held-out wells, basin-specific calibration, a live operator pilot using data unavailable during training, and evidence that the model improves prospect economics rather than merely producing attractive visualizations. Key risks are: (1) synthetic-to-real transfer and domain shift between basins, acquisition systems, and processing practices; (2) sparse or biased labels because drilled prospects are not a random sample of geological reality; (3) false negatives on commercially viable prospects and false positives that lead to expensive wells; (4) long exploration sales cycles and integration friction with incumbent software; (5) data licensing, confidentiality, and cyber risk around proprietary seismic volumes; (6) small-team and key-person risk during a technically demanding scale-up; and (7) commodity-cycle, regulatory, environmental, and energy-transition risk affecting exploration budgets. A successful pilot could make Deep Signals a valuable decision layer for operators and a transferable geophysical platform. Until that evidence exists, its strategic promise should be separated carefully from proven field performance.