All analysis was generated autonomously, without human review. Scores are analytical opinions drawn from the cited public sources, without hands-on testing. They are not audits, certifications, investment reports, purchasing advice, or evaluations of quality.
Mirror Security sells encryption software and AI agent controls to regulated industries such as banking, healthcare and government. Its VectaX product lets an AI model compute on a customer's data without decrypting it, and the customer keeps the keys. Founded in 2024, it raised a 2.5 million dollar pre-seed round led by Sure Valley Ventures and Atlantic Bridge. It names Nvidia, Intel, Yotta, G42 Inception, Accops, Qdrant and MongoDB among its customers. Yotta will deploy Mirror's platform on its cloud for Indian enterprises and government. Inception AI, part of Abu Dhabi-based G42, will deploy it across its enterprise and government platforms. Its named customers are technology and infrastructure companies, and it reaches the regulated buyers it targets through partners such as these.
| Description | Mirror Security sells an encryption and governance platform for AI systems, built so models run inference and retrieval on ciphertext while the customer keeps the keys. | [f1] |
|---|---|---|
| Founded | 2024 | [f2] |
| HQ | Dublin, Ireland | [f3] |
| Funding | $2.5M total | [f3] |
| Latest funding | Pre-seed, led by Sure Valley Ventures and Atlantic Bridge | [f4] |
| Product | What it does |
|---|---|
| VectaX | Encryption engine that keeps prompts, context, embeddings and inference output as ciphertext while a model computes on them. |
| AgentIQ | Control plane that gives AI agents an identity, scopes their tool and MCP access, and applies policy and guardrail checks on each action. |
| DiscoveR | Automated adversarial testing of AI systems, with findings mapped to the OWASP Top 10 for LLMs and MITRE ATLAS. |
| Zero | Agent-estate governance that finds sanctioned and unsanctioned AI agents and records eleven attributes for each one. |
| Gateway | Hosted endpoint that routes model and agent calls, applies policy at the door, and selects an encryption lane per route. |
| CodePrism | Coding assistance and review that runs on encrypted source, with decryption confined to the developer's own machine. |
| Encrypted Token Factory | OpenAI-compatible API for open-weight models that runs inference on encrypted prompts, with an optional local node on the developer's machine. |
AI Defense Matrix
| Govern | Identify | Protect | Detect | Respond | Recover | |
|---|---|---|---|---|---|---|
| AI-Workload Platforms Inference servers, training platforms, vector DB platforms, and the model-loading supply chain. | ||||||
| AI Coding and Orchestration Tools AI coding tools and agentic orchestration tools on user devices, plus their plugins, skills, hooks, system prompts, scaffolding, harnesses, configuration settings, and MCP clients. | ||||||
| AI-Generated Code Code produced by AI tools, AI-assisted reviews, AI-generated infrastructure-as-code and tests, and vibe-coded apps that bypass CI/CD. | ||||||
| AI Traffic Paths to external and self-hosted AI services, MCP tool channels, agent-to-agent calls, model-registry downloads, and egress to unapproved AI services. AI gateways, LLM routers, and MCP gateways steer traffic along those paths. The steering decisions and their inputs belong here too: routing policies, MCP server registries, and agent naming and discovery services. | ||||||
| AI Model Model weights, fine-tuning checkpoints, model cards, registries, an AI bill of materials (AIBOM), and the third-party LLMs your enterprise consumes. | ||||||
| Training Data Datasets used for training, fine-tuning, and continued learning. | ||||||
| Runtime AI Data User prompts, inference inputs, RAG content, vector DB content, persistent agent memory, and interaction history. | ||||||
| AI Agent Identities AI agents as non-human principals, plus credentials, keys, permission scopes, service accounts, and delegation chains across agents and tools. |
VectaX keeps prompts and inference output encrypted while a model computes on them. The platform also covers agent governance, runtime guardrails, the request path, shadow-AI discovery and encrypted code review. These products are mapped to the AI Defense Matrix. [f1]
How well the company can compete in its security market, scored across eight dimensions against public evidence.
| Dimension | Score | Rationale |
|---|---|---|
| Problem Clarity How precisely the company defines its problem, with evidence the problem exists at the scale claimed. | 3/5 | Mirror names regulated buyers and one specific pain: other approaches decrypt data to compute on it, which Mirror calls the in-use gap after encryption at rest and in transit. ThinkBusiness describes the same target, safeguarding proprietary data during model training and inference. None of the cited sources contains an independent measure of how many organisations delay AI projects over that pain, so the problem is clear and its scale is Mirror's own account. [s1, s4, s14] |
| Capability Depth How specific the technical capabilities are, with evidence beyond marketing claims such as docs and third-party validation. | 3/5 | Mirror's product pages give concrete detail. They describe encrypted vector, keyword and hybrid retrieval on the same key hierarchy as inference, a policy engine with more than 32 guardrail models, and a benchmark band for Llama-2 models served with vLLM. Mirror also names the NVIDIA libraries its inference platform uses. The benchmark figures are Mirror's own, and Mirror keeps the per-configuration rows in a technical brief. None of the cited sources contains a third-party evaluation or customer technical writeup, so the depth is documented without an external validation point. [s3, s4, s21] |
| Market Timing Whether the market is ready for this product, with evidence that buyers are actively seeking solutions. | 3/5 | Mirror incorporated in May 2024, and ThinkBusiness calls safeguarding proprietary data during model training and inference one of the most pressing challenges in enterprise AI adoption. The demand signals come from partners and an investor. Silicon Republic reported the Inception AI agreement, Yotta announced a service aimed at banking, healthcare and government, and Mindflair reported Mirror's selection for Google's Gemini Startup Forum. These announced deployments are partner adoption signals, and the cited sources document no separate kind of buyer demand, so demand stays indirect. [s1, s12, s14, s15, s19, s24] |
| Team Credibility Demonstrated domain expertise with public signals such as prior exits, publications, and industry recognition. | 3/5 | Silicon Republic and EU-Startups both name Pankaj Thapa and Dr Aditya Narayana K as founders. Mirror's about page credits Thapa with two decades on enterprise and nation state security products and leadership roles at startups it says were acquired. It credits Narayana with two decades across security, cryptography, privacy and machine learning. That is verifiable in-domain experience. The acquisitions and the board member Mirror describes as a former RSA Data Security operating chief appear on Mirror's own pages. No cited independent source documents those prior builds. [s2, s11, s12, s13] |
| GTM Proof Evidence of actual traction (customers, revenue signals, partnerships) beyond stated intentions. | 3/5 | Mirror's homepage names Nvidia, Intel, Yotta, G42 Inception, Accops, Qdrant, MongoDB, OWASP GenAI and AAIF under a Trusted by heading, and Mirror has announced partnerships with Yotta, Inception AI and Intel among others. The Yotta service appears in a joint release Yotta also publishes and in an investor's statement, and Silicon Republic and ThinkBusiness relayed Mirror's own Inception AI announcement. The cited outlets do not independently quantify the commercial scale of these relationships. [s1, s12, s14, s19, s20, s23] |
| Funding Efficiency Whether funding matches go-to-market ambition, with signs of capital-efficient growth. | 2/5 | Mirror raised 2.5 million dollars in a pre-seed round led by Sure Valley Ventures and Atlantic Bridge. It said it would use the money to expand teams in Ireland, the United States and India, develop encrypted inference and secure fine-tuning, and expand into United States enterprise markets. That is an enterprise sales motion on pre-seed money, and none of the cited sources discloses revenue or margin against it. Mirror offers the Encrypted Token Factory, and the mismatch between the motion and the raise is what holds the score at 2. [s1, s9, s12, s16] |
| Category Clarity Whether the company creates or fits a recognizable category that buyers can quickly place in their stack. | 3/5 | Silicon Republic, EU-Startups and ThinkBusiness each describe Mirror as an AI encryption company, and ThinkBusiness names fully homomorphic encryption as its core technology. Placement still needs explanation, because Mirror's product list runs from encrypted inference and an encrypted token service through agent governance, adversarial testing and encrypted code review. [s1, s12, s13, s14] |
| Incumbent Defensibility How vulnerable the core value proposition is to absorption as a feature by a platform vendor. | 3/5 | Encrypted inference and retrieval is cryptography integrated with model serving, and none of the cited sources reports a platform vendor offering it as a feature. In the Intel collaboration Mirror describes, the agent and the Mirror SDK run inside an Intel Confidential Computing TEE, so an adjacent hardware control already sits with a platform vendor. Integration and cryptographic complexity give some friction to absorption, and none of the cited sources describes a structural moat that bundling could not eventually replicate. [s3, s10, s21] |
Mirror Security aims at one point in the AI pipeline. On its homepage, Mirror says encryption at rest and in transit was locked down decades ago, and that the open gap is data in use, which everyone else decrypts to compute on. ThinkBusiness describes the same target as safeguarding proprietary data during model training and inference.
The buyers Mirror names are the ones regulation applies to. AgentIQ ships policy packs for finance, healthcare, GDPR and HIPAA. Mirror's partners page names regulated banks, sovereign clouds, defence agencies and healthcare networks as the customers its partners serve. Mirror also cites residency requirements under GDPR, HIPAA and sovereign AI mandates as the reason DiscoveR tests without access to model weights or training data.
Mirror frames the problem in compliance terms too. It says it maps signed evidence to NIST AI RMF, ISO 42001, the EU AI Act, SOC 2, HIPAA, GDPR, the OWASP Top 10 for LLMs and MITRE ATLAS. None of the cited sources measures how many organisations delay AI projects over data exposure, so the size of the problem is Mirror's own account. [s1, s4, s5, s14, s18]
Mirror lists six platform products and a newer Encrypted Token Factory. VectaX encrypts context, prompts, embeddings and inference output end to end, and runs encrypted vector search, encrypted keyword search and encrypted hybrid retrieval on the same key hierarchy as inference. AgentIQ checks agent actions against a natural-language policy engine and more than 32 pre-trained guardrail models. Zero surfaces sanctioned and shadow agents and maps eleven dimensions for each. Gateway picks an encryption lane per route for model and agent calls. DiscoveR runs automated adversarial testing, and CodePrism encrypts code at the IDE boundary so coding assistants process ciphertext.
Mirror describes the Encrypted Token Factory as an OpenAI-compatible API for open-weight models, where changing one base URL lets existing tools keep working. A developer can also run its node on their own machine, and Mirror says prompts are sealed there before any byte leaves. Mirror lists encrypted memory and encrypted guardrails for the service as coming soon.
The published performance numbers are Mirror's own. Mirror reports encrypted inference on open-weight Llama-2 7B and 70B models served with vLLM. It reports time to first token under 350 milliseconds and accuracy within one point of plaintext inference on four benchmark tests. It publishes that band on the VectaX page and places the per-configuration rows in a technical brief. Mirror also names the NVIDIA libraries its production inference platform uses, among them CUDA, cuBLAS, NeMo Retriever and TensorRT-LLM.
Mirror narrows the headline claim itself. It says pure FHE is too slow for production serving, so it spends its cryptographic budget on user data and keeps open-weight models in the clear. [s1, s3, s4, s5, s6, s7, s8, s16, s21]
Mirror positions VectaX against hardware enclaves. On the VectaX page, Mirror describes a TEE as covering one surface, the inference enclave, and says VectaX moves with the data. Mirror still uses enclave hardware in its Intel collaboration, where the agent application and the Mirror SDK run inside an Intel Confidential Computing TEE with attestation from Intel Tiber Trust Authority.
Zama has raised more for homomorphic encryption. EU-Startups set Mirror's pre-seed round beside European companies working in adjacent areas of cryptography and AI security. It reported that Zama secured 49 million euro to scale fully homomorphic encryption for confidential smart-contract execution. Mirror raised 2.1 million euro for a different application of the same mathematics. [s3, s10, s13]
Mirror's homepage names Nvidia, Intel, Yotta, G42 Inception, Accops, Qdrant, MongoDB, OWASP GenAI and AAIF under a Trusted by heading. Mirror's news page lists partnership announcements for several of the same organisations.
Both companies document the Yotta relationship. Mirror and Yotta announced a partnership to offer encrypted AI inference as a service on Yotta's Shakti Cloud, aimed at banking, financial services, healthcare, legal services and government agencies. Yotta publishes that announcement on its own site. Mindflair holds its interest in Mirror through Sure Valley Ventures' third fund. It said Mirror's encrypted inference platform had moved into full production and that Yotta will deploy it for Indian enterprises, government and defence organisations.
Silicon Republic reported Mirror's announcement of a multimillion-dollar strategic agreement with Inception AI, under which Mirror will deploy its security stack across Inception's enterprise and government ecosystem. ThinkBusiness placed Inception within Abu Dhabi based G42. Mirror's own announcement says Inception will integrate and deploy Mirror's platform and co-develop AI security products for government entities and enterprises. Silicon Republic also listed earlier partnerships with Intel, MongoDB, Qdrant, SiSys AI and Accops, and Credio announced that it would integrate Mirror's technology into its platform.
Mirror also runs a partner programme. It says partners co-build, co-sell and co-deliver with Mirror, with discount tiers and revenue share negotiated at signing. [s1, s12, s14, s17, s18, s19, s20, s22, s23, s26]
Silicon Republic and EU-Startups both name Pankaj Thapa and Dr Aditya Narayana K as Mirror's founders. Silicon Republic reports that the company spun out of University College Dublin and originated from Narayana's thesis. Mirror's about page says Thapa leads strategy, partnerships and sales after more than two decades building enterprise, security and nation state security products. It says Narayana leads research and product direction after more than two decades across security, cryptography, privacy and machine learning.
The specifics that would strengthen that record appear on Mirror's own pages. The about page credits Thapa with leadership roles at three startups it says were acquired: Azingo by Motorola, Mobiliya by QuEST Global, and Digital14, now KATIM, by Edge Group. It credits Narayana with 16 patents and more than 50 research publications. Silicon Republic repeats a figure of 23 team-held patents that also appears in Mirror's funding announcement.
Mirror announced Albert Sisto as an independent board member. Its announcement describes him as a former Chief Operating Officer of RSA Data Security with more than 40 years across public companies, startups, venture capital and private equity. [s2, s9, s11, s12, s13]
Mirror states that it is SOC 2 Type 1 certified and GDPR and HIPAA ready. A probe of likely trust surfaces found no public trust portal: the trust and security subdomains did not resolve, and the trust, security and compliance paths on the main site returned 404. No report or scope document appeared on the probed surfaces.
Mirror's trust argument is architectural. It says keys stay with the customer in every deployment mode, from SaaS through a customer VPC and a sovereign region to a fully air-gapped network, and that nothing calls home. The about page adds that Mirror cannot see, use or reconstruct customer data, and that customer data and its insights never become training fuel or context for anyone else's models.
Mirror also offers its product's output as evidence. It says every inference and retrieval through VectaX leaves a tamper-evident signed receipt capturing the request hash, the policy version applied, the model invoked and the encryption posture. None of the cited sources reports an auditor, regulator or customer acting on those receipts. [s1, s2, s3, s27]
| Company | Relationship | Note | Compare |
|---|---|---|---|
| Zama | adjacent | EU-Startups named Zama among European companies that secured funding in areas adjacent to Mirror, reporting a 49 million euro raise to scale fully homomorphic encryption for confidential smart-contract execution. |
Add analyzed competitors to compare them side by side with Mirror Security.
A closer look at the company's product strategy, measuring how defensible it is against market forces and examining the eight areas behind it.
reinforce or reposition
Mirror Security sells encryption software that lets AI models compute on data they never see, alongside agent controls, adversarial testing and encrypted code review. Customers hold their own keys, and Mirror says their data never trains other models. Its funding announcement cites 23 patents held by its team. Its SOC 2 Type 1 certification is ordinary enterprise preparation, so it does not block a replacement. Customers replacing Mirror would need to change integrations with its software development kit, policy engine and encrypted indexes.
| Dimension | Score | Rationale |
|---|---|---|
| Value Delivery Does the product sell software as the product, or judgment, trust, or accountability with software as the delivery mechanism. | 1/3 | Mirror sells software products: encrypted inference, agent controls, adversarial testing and encrypted code review. It delivers them as SaaS, as the hosted Gateway, and as VectaX running in a customer VPC or air-gapped network. In each mode what the customer buys is the software and its features. |
| Switching Cost How expensive leaving is for a customer: data portability, integrations, learned workflows, network effects, regulatory data residency. | 2/3 | Leaving Mirror means undoing real integration. For Gateway, encryption happens at the source through Mirror's SDK, agent rules live in AgentIQ's policy engine, and VectaX runs encrypted vector search and retrieval on the same key hierarchy as inference. The cited sources document these integrations but do not quantify the work needed to replace them. |
| Compliance Moat Whether certifications, liability acceptance, or audit trails block an easy replacement. | 1/3 | Mirror states a SOC 2 Type 1 certification and GDPR and HIPAA readiness. Those are ordinary enterprise-market preparation. None of the cited sources describes a product-carried authorisation, mandate or retained liability that would block a replacement. |
| Problem Complexity Whether the product requires ML, optimization, real-time systems, or years of specialized expertise. | 3/3 | VectaX runs encrypted inference and retrieval inside model-serving paths such as vLLM and TGI, and Mirror's Intel write-up names similarity-preserving encryption and hybrid homomorphic encryption among its techniques. That is cryptography integrated with real-time model serving. Narayana's two decades across security, cryptography, privacy and machine learning match the specialised expertise that work calls for. |
| Buyer Profile Whether buyers are SMB operators, mid-market IT teams, or regulated enterprises and governments with procurement gates. | 3/3 | Mirror builds for regulated buyers, with policy packs for finance, healthcare, GDPR and HIPAA. On its partners page, Mirror says its partners ship encrypted AI to regulated banks, sovereign clouds, defence agencies and healthcare networks. Silicon Republic relays Mirror's announcement that it will deploy its security stack across Inception's enterprise and government ecosystem. The organisations named in the cited sources are technology and infrastructure companies such as Nvidia, Intel and Yotta, and the regulated buyers in the record are those Mirror says its partners serve. |
| Layer Whether the product is an end-user application, a platform with application features, or infrastructure other applications depend on. | 2/3 | VectaX, Gateway and the Encrypted Token Factory are infrastructure that other AI applications call, including an OpenAI-compatible API that existing tools reach by changing one base URL. Mirror also ships application products, with DiscoveR's testing, Zero's agent governance and CodePrism's coding assistance, which makes it a platform with application features. |
| Proprietary Data, Content, or IP Whether the product accumulates datasets, content licenses, or IP that a rival cannot recreate from scratch. | 1/3 | The asset kind Mirror puts forward is intellectual property, and the cited sources name counts: 23 patents its funding announcement calls team-held, and 16 patents on the about page. None of the cited sources identifies a granted patent assigned to Mirror. On data, Mirror says customers hold their own keys and that customer data never becomes training fuel for other models, and none of the cited sources names a dataset Mirror retains. |
Mirror aims at organisations that regulation keeps from moving data. AgentIQ ships policy packs for finance, healthcare, GDPR and HIPAA, and Mirror deploys as SaaS, in a customer VPC, in a sovereign region or fully air-gapped, with keys under customer control. Its partners page names regulated banks, sovereign clouds, defence agencies and healthcare networks as the customers Mirror's partners serve.
Several of Mirror's announced routes into that segment run through platforms. The Yotta partnership targets banking, financial services, healthcare, legal services and government agencies on Yotta's Shakti Cloud. Silicon Republic reported that Mirror will deploy its stack across Inception AI's enterprise and government ecosystem. The Encrypted Token Factory adds a developer segment, bringing encrypted inference to coding CLIs and copilots.
Mirror covers three geographies at an early stage. Its funding announcement puts team expansion in Ireland, the United States and India, and the SoloCheck record shows incorporation on 1 May 2024 and a micro company size class.
VectaX is Mirror's encryption engine. It encrypts context, prompts, embeddings and inference output end to end, and adds encrypted vector search, encrypted keyword search and encrypted hybrid retrieval on the same key hierarchy as inference. Mirror reports encrypted inference on open-weight Llama-2 7B and 70B models served with vLLM, with time to first token under 350 milliseconds and accuracy within one point on four standard tests.
It says pure FHE is too slow for production serving, so it spends its cryptographic budget on user data and keeps open-weight models in the clear. Mirror's Intel write-up names similarity-preserving encryption for embedding storage and hybrid homomorphic encryption for similarity computations among its techniques.
The Encrypted Token Factory offers the same approach as a service. Mirror describes an OpenAI-compatible API for open-weight models and a local node that seals prompts on the developer's machine, and lists encrypted memory and encrypted guardrails as coming soon.
Mirror's other products cover agent controls and testing. AgentIQ pairs a natural-language policy engine with more than 32 pre-trained guardrail models, and Mirror says its runtime checks complete in under 25 milliseconds inline. DiscoveR simulates attacker access patterns against a customer's AI system without access to model weights or training data. Every figure is Mirror's own, and none of the cited sources contains a third-party benchmark or audit.
Mirror's homepage names Nvidia, Intel, Yotta, G42 Inception, Accops, Qdrant, MongoDB, OWASP GenAI and AAIF under a Trusted by heading. Its news page lists partnership announcements for several of the same organisations.
Both companies document the Yotta relationship. Mirror and Yotta announced a partnership to offer encrypted AI inference as a service on Yotta's Shakti Cloud, and Yotta publishes the announcement on its own site. Mindflair, which holds its interest in Mirror through Sure Valley Ventures' third fund, said Yotta will deploy Mirror's platform for Indian enterprises, government and defence organisations.
Silicon Republic and ThinkBusiness each reported Mirror's announcement of a multimillion-dollar strategic agreement with Inception AI, part of G42, to deploy Mirror's security stack across Inception's enterprise and government ecosystem. Mirror's own announcement adds that the two companies will co-develop AI security products for government entities and enterprises.
Mirror also runs a partner programme in which, it says, partners co-build, co-sell and co-deliver, with discount tiers and revenue share negotiated at signing. UK Investor Magazine reported Mindflair's statement that Mirror is one of 33 cybersecurity startups selected for Google's Gemini Startup Forum.
None of the captured pages gives rates or paid tiers. Mirror's Encrypted Token Factory page says exact variants, pricing and current availability live in a signed-in catalog, and Mirror's homepage header offers Book a Demo and Start for Free.
For partners, Mirror says discount tiers and revenue share are negotiated at signing. None of the captured pages gives a per-call, per-seat or per-token figure for a buyer estimating cost.
Mirror offers both hosted and customer-controlled delivery. It deploys as SaaS, in a customer VPC, in a sovereign region or fully air-gapped, with keys staying with the customer in every mode and nothing calling home. Gateway is hosted, while VectaX can run entirely in a customer VPC or air-gapped network. The Encrypted Token Factory adds a local node a developer runs on their own machine.
The operational figures Mirror publishes are its own latency and capacity claims. Mirror says AgentIQ's runtime checks complete in under 25 milliseconds inline, and reports time to first token under 350 milliseconds for VectaX on Llama-2 models served with vLLM. For Gateway, it reports sustained throughput across 1,000 to 10,000 concurrent agents on a single Mirror cluster. For Gateway, Mirror's SDK encrypts at the source before the calling app sends the request to the Mirror endpoint.
For partners, Mirror offers architecture review with a Mirror solutions engineer and a dedicated solutions engineer for a partner's initial three deals. None of the cited sources describes an implementation timeline or a service commitment to direct customers, so a buyer sizing the operating burden has Mirror's own figures to work from.
Mirror states that it is SOC 2 Type 1 certified and GDPR and HIPAA ready. A probe of likely trust surfaces found no public trust portal: the trust and security subdomains did not resolve, and the trust, security and compliance paths on the main site returned 404. No report or scope document appeared on the probed surfaces.
Mirror's stronger trust argument is architectural. Mirror says the customer holds the keys in every deployment mode. Its about page says Mirror cannot see, use or reconstruct customer data, and that customer data never becomes training fuel for anyone else's models. For a buyer whose objection is that the vendor could read the data, key custody addresses that objection directly.
Mirror also offers its product's output as evidence. It says every inference and retrieval through VectaX leaves a tamper-evident signed receipt. Mirror also says it maps signed evidence to NIST AI RMF, ISO 42001, the EU AI Act, SOC 2, HIPAA, GDPR, the OWASP Top 10 for LLMs and MITRE ATLAS. Mirror says DiscoveR's testing maps to the OWASP LLM Top 10 and MITRE ATLAS. None of the cited sources reports an auditor, regulator or customer acting on that output.
Mirror presents its products as one platform. Gateway picks an encryption lane per route for model and agent calls. AgentIQ applies identity and policy to agent actions, Zero surfaces agents, DiscoveR tests AI systems and CodePrism encrypts code at the IDE boundary. VectaX supplies the encryption, and the Encrypted Token Factory exposes encrypted inference through an OpenAI-compatible API.
Mirror builds on other vendors' hardware and clouds. In the Intel collaboration Mirror describes, the agent application and the Mirror SDK run inside an Intel Confidential Computing TEE with attestation from Intel Tiber Trust Authority. Mirror's production inference platform uses NVIDIA's CUDA, cuBLAS, NeMo Retriever and TensorRT-LLM, and Mirror and Yotta offer the encrypted inference service on Yotta's Shakti Cloud.
Silicon Republic lists partnerships with Intel, MongoDB, Qdrant, SiSys AI and Accops. Mirror announced a collaboration with SiSys AI to develop an FHE co-processor. A homepage testimonial describes pairing VectaX with Accops HySecure, and Credio announced that it would integrate Mirror's technology into its platform.
Mirror describes seven partner shapes in its partner programme, each running Mirror's products underneath the partner's own offer. None of the cited sources describes exclusivity terms in the named partnerships.
Silicon Republic and EU-Startups both name Pankaj Thapa and Dr Aditya Narayana K as Mirror's founders, and Silicon Republic reports that the company originated from Narayana's thesis. Mirror's about page says Thapa leads strategy, partnerships and sales after more than two decades building enterprise, security and nation state security products. It says Narayana leads research and product direction after more than two decades across security, cryptography, privacy and machine learning.
Its about page credits Thapa with leadership roles at three startups it says were acquired: Azingo by Motorola, Mobiliya by QuEST Global, and Digital14, now KATIM, by Edge Group. It credits Narayana with 16 patents and more than 50 research publications.
Mirror announced Albert Sisto as an independent board member. It describes him as a former Chief Operating Officer of RSA Data Security with more than 40 years across public companies, startups, venture capital and private equity. SoloCheck classifies Mirror Security Limited as a micro company.
| Id | Source | Tier | Accessed |
|---|---|---|---|
| f1 | Mirror Security: homepage | official | 2026-09-15 |
| f2 | SoloCheck: Mirror Security Limited company record | other | 2026-09-15 |
| f3 | Silicon Republic: Dublin's Mirror Security raises $2.5m for AI encryption tech | press | 2026-09-15 |
| f4 | EU-Startups: UCD spin-out Mirror Security secures 2.1 million euro for FHE-based AI security | press | 2026-09-15 |
| Id | Source | Tier | Accessed |
|---|---|---|---|
| s1 | Mirror Security: homepage “Mirror deploys as SaaS, in a customer VPC, in a sovereign region, or fully air-gapped, while customer keys remain under customer control. Mirror Security is SOC 2 Type 1 Certified.” | official | 2026-09-15 |
| s2 | Mirror Security: About Us “Spun out of University College Dublin's NovaUCD Ireland's centre for deep-tech research ventures.” | official | 2026-09-15 |
| s3 | Mirror Security: VectaX product page “VectaX retains most of plaintext throughput on production AI workloads. On open-weight Llama-2 7B and 70B served with vLLM on datacenter GPUs, encrypted inference runs at production speed with TTFT under 350 ms. Accuracy delta across MMLU, HellaSwag, HumanEval, and GSM8K stays under 1 point.” | official | 2026-09-15 |
| s4 | Mirror Security: AgentIQ product page “Control plane for AI agents. Identity, policy, 32+ guardrails, and a signed verdict on every call.” | official | 2026-09-15 |
| s5 | Mirror Security: DiscoveR product page “DiscoveR runs systematic, automated adversarial testing mapped to OWASP LLM Top 10 and MITRE ATLAS, the standard frameworks for AI adversarial testing.” | official | 2026-09-15 |
| s6 | Mirror Security: Zero product page “Discover, Map, Govern, Enforce. Discover surfaces every agent (sanctioned or shadow) across the environment. Map captures the agent's 11 dimensions: owner, identity, runtime, model, prompt, tools, MCP, plugins, permissions, data, browser.” | official | 2026-09-15 |
| s7 | Mirror Security: Gateway product page “Policies fire on encrypted metadata where possible; the lane (Plain · E2E · FHE · Local) is picked per route.” | official | 2026-09-15 |
| s8 | Mirror Security: CodePrism product page “Work on your code. Without ever decrypting it.” | official | 2026-09-15 |
| s9 | Mirror Security: pre-seed funding announcement “has raised $2.5 million in pre-seed funding to scale its breakthrough encryption platform for AI Security.” | official | 2026-09-15 |
| s10 | Mirror Security: Intel collaboration write-up “Intel and Mirror Security have collaborated to address the growing security challenges posed by autonomous AI agents.” | official | 2026-09-15 |
| s11 | Mirror Security: board appointment announcement “Mirror Security, the deep tech AI security company pioneering encrypted AI inference for regulated industries, today announced the appointment of Albert (Al) Sisto as an Independent Board Member.” | official | 2026-09-15 |
| s12 | Silicon Republic: Dublin's Mirror Security raises $2.5m for AI encryption tech “Dublin-based cyber start-up Mirror Security has announced today (2 November) a pre-seed fundraise of $2.5m to scale its encryption platform for AI security.” | press | 2026-09-15 |
| s13 | EU-Startups: UCD spin-out Mirror Security secures 2.1 million euro for FHE-based AI security “a research-driven cybersecurity company spun out of University College Dublin, has raised €2.1 million ($2.5 million) in pre-Seed funding to scale its encryption platform for AI Security.” | press | 2026-09-15 |
| s14 | ThinkBusiness: Mirror Security raises $2.5m to secure AI with breakthrough encryption “The Dublin-based company is tackling one of the most pressing challenges in enterprise AI adoption: safeguarding proprietary data during model training and inference.” | press | 2026-09-15 |
| s15 | SoloCheck: Mirror Security Limited company record “Mirror Security Limited was set up on Wednesday the 1st of May 2024. Their current partial address is Dublin, and the company status is Normal.” | other | 2026-09-15 |
| s16 | Mirror Security: Encrypted Token Factory page “An OpenAI-compatible API for open-weight models: change one base URL and your existing tools keep working.” | official | 2026-09-15 |
| s17 | Mirror Security: Articles and News index “Mirror Security and Yotta Partner to Launch Encrypted AI Inference as a Service on Shakti Cloud” | official | 2026-09-15 |
| s18 | Mirror Security: Partners Program page “Mirror partners ship encrypted AI to customers most platforms can't reach. Regulated banks, sovereign clouds, defence agencies, healthcare networks.” | official | 2026-09-15 |
| s19 | Mirror Security: Yotta partnership announcement “Mirror Security Limited and Yotta Data Services Private Limited today announced a strategic partnership to jointly offer Secure encrypted AI Inference as a Service on Yotta's Shakti Cloud - India's largest sovereign AI infrastructure with 16,000+ GPUs.” | official | 2026-09-15 |
| s20 | Yotta: Mirror Security and Yotta encrypted AI inference press release “Mirror Security Limited and Yotta Data Services Private Limited today announced a strategic partnership to jointly offer Secure encrypted AI Inference as a Service on Yotta's Shakti Cloud - India's largest sovereign AI infrastructure with 16,000+ GPUs.” | official | 2026-09-15 |
| s21 | Mirror Security: NVIDIA encrypted inference announcement “Mirror Security’s platform integrates NVIDIA AI inference across the accelerated computing stack, including NVIDIA CUDA and cuBLAS, the NVIDIA NeMo framework with NVIDIA NeMo Retriever , and NVIDIA TensorRT-LLM for optimized inference and serving.” | official | 2026-09-15 |
| s22 | Mirror Security: Inception AI agreement announcement “G42's Inception AI and Mirror Security Announce Strategic Agreement to Partner & Co-Develop Next-Generation AI Security Solutions” | official | 2026-09-15 |
| s23 | Vox Markets: Mindflair investee Mirror Security launches NVIDIA powered AI inference platform “has noted that portfolio company Mirror Security has moved its Encrypted AI Inference platform into full production, powered by NVIDIA accelerated computing infrastructure.” | press | 2026-09-15 |
| s24 | UK Investor Magazine: Mindflair investee Mirror Security picked for Google AI cybersecurity forum “Mindflair has said one of its investee companies, Mirror Security, has been chosen for Google’s Gemini Startup Forum for cybersecurity, a flagship programme backing the next generation of AI-enabled security firms.” | press | 2026-09-15 |
| s25 | Mirror Security: SiSys AI co-processor collaboration announcement “today announced a strategic collaboration with industry leaders to develop and commercialize the MIRROR (Modular Infrastructure for Real-time Reasoning Over Encrypted Data) co-processor.” | official | 2026-09-15 |
| s26 | Mirror Security: Credio partnership announcement “This partnership will integrate Mirror Security's advanced, privacy-preserving technologies into Credio's platform.” | official | 2026-09-15 |
| s27 | Mirror Security trust probe: trust and security subdomains and a nonsense control fail DNS, etf subdomain resolves, /trust /security /compliance return 404 | official | 2026-09-15 |
| Id | Source | Tier | Accessed |
|---|---|---|---|
| s1 | Mirror Security: homepage “Mirror deploys as SaaS, in a customer VPC, in a sovereign region, or fully air-gapped, while customer keys remain under customer control. Mirror Security is SOC 2 Type 1 Certified.” | official | 2026-09-15 |
| s2 | Mirror Security: About Us “Spun out of University College Dublin's NovaUCD Ireland's centre for deep-tech research ventures.” | official | 2026-09-15 |
| s3 | Mirror Security: VectaX product page “VectaX retains most of plaintext throughput on production AI workloads. On open-weight Llama-2 7B and 70B served with vLLM on datacenter GPUs, encrypted inference runs at production speed with TTFT under 350 ms. Accuracy delta across MMLU, HellaSwag, HumanEval, and GSM8K stays under 1 point.” | official | 2026-09-15 |
| s4 | Mirror Security: AgentIQ product page “Control plane for AI agents. Identity, policy, 32+ guardrails, and a signed verdict on every call.” | official | 2026-09-15 |
| s5 | Mirror Security: DiscoveR product page “DiscoveR runs systematic, automated adversarial testing mapped to OWASP LLM Top 10 and MITRE ATLAS, the standard frameworks for AI adversarial testing.” | official | 2026-09-15 |
| s6 | Mirror Security: Zero product page “Discover, Map, Govern, Enforce. Discover surfaces every agent (sanctioned or shadow) across the environment. Map captures the agent's 11 dimensions: owner, identity, runtime, model, prompt, tools, MCP, plugins, permissions, data, browser.” | official | 2026-09-15 |
| s7 | Mirror Security: Gateway product page “Policies fire on encrypted metadata where possible; the lane (Plain · E2E · FHE · Local) is picked per route.” | official | 2026-09-15 |
| s8 | Mirror Security: CodePrism product page “Work on your code. Without ever decrypting it.” | official | 2026-09-15 |
| s9 | Mirror Security: pre-seed funding announcement “has raised $2.5 million in pre-seed funding to scale its breakthrough encryption platform for AI Security.” | official | 2026-09-15 |
| s10 | Mirror Security: Intel collaboration write-up “Intel and Mirror Security have collaborated to address the growing security challenges posed by autonomous AI agents.” | official | 2026-09-15 |
| s11 | Mirror Security: board appointment announcement “Mirror Security, the deep tech AI security company pioneering encrypted AI inference for regulated industries, today announced the appointment of Albert (Al) Sisto as an Independent Board Member.” | official | 2026-09-15 |
| s12 | Silicon Republic: Dublin's Mirror Security raises $2.5m for AI encryption tech “Dublin-based cyber start-up Mirror Security has announced today (2 November) a pre-seed fundraise of $2.5m to scale its encryption platform for AI security.” | press | 2026-09-15 |
| s13 | EU-Startups: UCD spin-out Mirror Security secures 2.1 million euro for FHE-based AI security “a research-driven cybersecurity company spun out of University College Dublin, has raised €2.1 million ($2.5 million) in pre-Seed funding to scale its encryption platform for AI Security.” | press | 2026-09-15 |
| s14 | ThinkBusiness: Mirror Security raises $2.5m to secure AI with breakthrough encryption “The Dublin-based company is tackling one of the most pressing challenges in enterprise AI adoption: safeguarding proprietary data during model training and inference.” | press | 2026-09-15 |
| s15 | SoloCheck: Mirror Security Limited company record “Mirror Security Limited was set up on Wednesday the 1st of May 2024. Their current partial address is Dublin, and the company status is Normal.” | other | 2026-09-15 |
| s16 | Mirror Security: Encrypted Token Factory page “An OpenAI-compatible API for open-weight models: change one base URL and your existing tools keep working.” | official | 2026-09-15 |
| s17 | Mirror Security: Articles and News index “Mirror Security and Yotta Partner to Launch Encrypted AI Inference as a Service on Shakti Cloud” | official | 2026-09-15 |
| s18 | Mirror Security: Partners Program page “Mirror partners ship encrypted AI to customers most platforms can't reach. Regulated banks, sovereign clouds, defence agencies, healthcare networks.” | official | 2026-09-15 |
| s19 | Mirror Security: Yotta partnership announcement “Mirror Security Limited and Yotta Data Services Private Limited today announced a strategic partnership to jointly offer Secure encrypted AI Inference as a Service on Yotta's Shakti Cloud - India's largest sovereign AI infrastructure with 16,000+ GPUs.” | official | 2026-09-15 |
| s20 | Yotta: Mirror Security and Yotta encrypted AI inference press release “Mirror Security Limited and Yotta Data Services Private Limited today announced a strategic partnership to jointly offer Secure encrypted AI Inference as a Service on Yotta's Shakti Cloud - India's largest sovereign AI infrastructure with 16,000+ GPUs.” | official | 2026-09-15 |
| s21 | Mirror Security: NVIDIA encrypted inference announcement “Mirror Security’s platform integrates NVIDIA AI inference across the accelerated computing stack, including NVIDIA CUDA and cuBLAS, the NVIDIA NeMo framework with NVIDIA NeMo Retriever , and NVIDIA TensorRT-LLM for optimized inference and serving.” | official | 2026-09-15 |
| s22 | Mirror Security: Inception AI agreement announcement “G42's Inception AI and Mirror Security Announce Strategic Agreement to Partner & Co-Develop Next-Generation AI Security Solutions” | official | 2026-09-15 |
| s23 | Vox Markets: Mindflair investee Mirror Security launches NVIDIA powered AI inference platform “has noted that portfolio company Mirror Security has moved its Encrypted AI Inference platform into full production, powered by NVIDIA accelerated computing infrastructure.” | press | 2026-09-15 |
| s24 | UK Investor Magazine: Mindflair investee Mirror Security picked for Google AI cybersecurity forum “Mindflair has said one of its investee companies, Mirror Security, has been chosen for Google’s Gemini Startup Forum for cybersecurity, a flagship programme backing the next generation of AI-enabled security firms.” | press | 2026-09-15 |
| s25 | Mirror Security: SiSys AI co-processor collaboration announcement “today announced a strategic collaboration with industry leaders to develop and commercialize the MIRROR (Modular Infrastructure for Real-time Reasoning Over Encrypted Data) co-processor.” | official | 2026-09-15 |
| s26 | Mirror Security: Credio partnership announcement “This partnership will integrate Mirror Security's advanced, privacy-preserving technologies into Credio's platform.” | official | 2026-09-15 |
| s27 | Mirror Security trust probe: trust and security subdomains and a nonsense control fail DNS, etf subdomain resolves, /trust /security /compliance return 404 | official | 2026-09-15 |
This site is an experimental research aid created by Zeltser Security Corp. All its data gathering and analysis was performed autonomously without human review, and it can contain errors of fact, interpretation, and judgment that a human reviewer might catch.
The analyses are statements of opinion, not statements of fact. Machine analysis produced the scores, summaries, and matrix placements by weighing the public sources each page cites, and reasonable people can weigh the same sources differently. Where a page states a fact, it cites the public source and the date it was checked, and the statement is only as accurate as that source. Unless a profile expressly says otherwise, the analysis involves no hands-on testing and no independent validation of any company's products or services.
Nothing here is professional, security, legal, financial, investment, or purchasing advice, and nothing here is a recommendation to invest in, do business with, or avoid any company. Inclusion of a company is not an endorsement, and absence of a company is not a judgment about it. Reading this site creates no advisory or client relationship. Verify any detail you plan to act on against the vendor's current materials.
The content is provided "as is" and "as available," with all warranties disclaimed, express or implied, including merchantability, fitness for a particular purpose, accuracy, and non-infringement. No entry is warranted to be complete, current, or correct. Companies change, vendors update their claims, sources can be wrong, and automated analysis can misread them.
To the fullest extent permitted by law, the operator, Zeltser Security Corp, is not liable for any damages that arise from using this site or relying on its content, including direct, indirect, incidental, special, and consequential damages and lost profits, even if advised that such damages were possible. If you are dissatisfied with the site or disagree with these terms, your remedy is to stop using it.
Entries link to vendor pages, press coverage, and other external sites that Zeltser Security Corp does not control and is not responsible for. A link is not an affiliation with the destination or an endorsement of it. Product and company names and trademarks are the property of their owners, used here nominatively to identify the companies described. Short quotations from cited sources appear for identification and commentary.
Use, quotation, automated retrieval, and redistribution of the content are governed by the Terms of Use at cybercompanyprofiles.com/terms, which permit personal and internal business use with attribution and prohibit republication and resale.