Track AI-vs-human invention contributions automatically so deep-tech startups never lose a patent on inventorship grounds.

1. Lightbringer's 300% YoY growth and $10M Series A proves deep-tech founders are actively buying patent-ops software, validating the market for upstream provenance tooling.
2. Lightbringer's named customers—defence drones, quantum sensing, infrastructure—are the same sectors adopting AI simulation and generative design at the fastest rate, making undocumented AI contributions an imminent patent-invalidity risk.
3. The market is moving toward full-lifecycle patent ownership, but no incumbent or emerging tool addresses the pre-filing inventorship documentation gap that prosecution software leaves behind.
4. Deep-tech startups already face the two-month filing lag that Lightbringer targets; inventorship documentation adds another multi-week bottleneck on top, compounding urgency for any tool that eliminates both.

Catalyst. Lightbringer's $10M Series A and 300% growth validates the deep-tech patent software market just as AI tools flood the exact R&D workflows those same companies depend on for competitive IP.

An IDE- and lab-tool plugin monitors which AI systems touched which design artifacts during R&D, tagging each AI-generated suggestion or output at creation time. When an inventor disclosure is triggered, the platform generates a structured inventorship package—showing the human decisions, modifications, and creative leaps that distinguish the invention from the AI substrate—ready for review by patent counsel. The platform integrates with Lightbringer, established prosecution tools, and USPTO EFS-Web so the provenance record travels with the application. A real-time contribution dashboard lets CTOs see inventorship risk across their entire patent portfolio before a due diligence request lands.

What's different. Every existing patent tool—including Lightbringer—begins when a human inventor submits a disclosure. This platform begins when the first AI tool is opened in the R&D environment, capturing contribution evidence before any attorney is involved. The resulting provenance graph is a continuous, tamper-evident record rather than a retroactive declaration, making it structurally more defensible in litigation or USPTO examination. Deep integration with the specific AI tools used in quantum and defense R&D creates a switching cost that prosecution-only tools cannot replicate.

Jobs to be done

flowchart LR
  RD[R&D Engineer] --> Monitor[Provenance Monitor]
  Monitor --> ContribGraph[Contribution Graph]
  ContribGraph --> Package[Inventorship Package]
  Package --> Counsel[Patent Counsel]
  Counsel --> Filing[Patent Filing]
  ContribGraph --> Dashboard[CTO Dashboard]

Executive takeaways

• The near-term wedge is real: patent teams increasingly accept AI inside R&D, but the legal standard still anchors inventorship in human conception, creating a new evidence gap rather than a new filing shortcut.
• This is a compliance workflow sale, not a generic AI copilot sale. Buyers care about auditability, counsel acceptance, and secure deployment more than model novelty.
• Incumbents own downstream patent administration, not upstream invention provenance. That leaves room for a narrow but urgent product that starts before the invention disclosure form exists.
• The beachhead is attractive but not huge; a credible year-three outcome is a low-single-digit-million ARR business unless the company expands from startup patent teams into law firms, biotech, semis, and enterprise R&D labs.
• Go-to-market should ride existing trust networks: deep-tech counsel, AI-native patent firms, and investors who already diligence IP quality in financing and M&A.

Market definition

Software that creates an audit-grade record of human versus AI contribution during R&D and turns that record into filing-ready inventorship support for counsel. It sits upstream of patent management, docketing, and prosecution systems: those tools begin at disclosure or filing, while this product begins when AI is used to shape the underlying invention.

Customer and buyer

Primary users are CTOs, engineering leads, and patent operations owners at patent-active deep-tech startups using AI in daily design or simulation work. The economic buyer is usually the CTO or General Counsel, with outside patent counsel as a powerful technical validator because counsel must be willing to rely on the generated provenance package.

Buying triggers

• Financing, M&A, or investor diligence forces the company to defend patent quality and exposes that current inventor declarations are retrospective and fragile. [24][35][36]
• A live filing or inventorship review requires evidence that only natural persons conceived the claimed invention and that any error can be corrected before it becomes a litigation problem. [1][20][21][22][29]
• Defense-adjacent or controlled-technology programs require stronger audit trails and secure handling of technical data, making ad hoc email workflows harder to defend. [17][18][19]

Willingness to pay

Willingness to pay exists when the tool replaces outside-counsel coordination and lowers existential IP risk. Tech Funding News says Lightbringer sells on cutting filing cycles from about two months to days at roughly half the cost, while Baker Botts and McDermott show that inventorship mistakes can undermine patent value. SimpleIP and Triangle IP both market buyers away from spreadsheets and opaque admin cost, implying that budget already exists for software that reduces legal and operational drag. [25][35][36][50][103][110]

Category dynamics

Growth signal 22.2% YoY deep-tech startup growth (StartUs tracker)

Validation signals

• Lightbringer reports 300% YoY growth and more than 200 deep-tech customers across 17 countries, showing active demand for patent-ops tooling.
• Tech Funding News says Lightbringer sells on cutting filing time from about two months to days at roughly half the cost, confirming cycle-time pain.
• StartUs tracks 57,669 deep-tech startups and 22.2% YoY category growth, supporting a meaningful if specialized prospect base.
• EPO data shows computer technology is the largest filing field, reinforcing that AI-adjacent invention activity is not niche.

Regulatory & technical constraints

• Only natural persons can be inventors; the product must help prove human conception rather than imply the AI system was an inventor or joint inventor.
• Incorrect inventorship has formal correction paths, but unresolved errors can still threaten enforceability and value.
• Defense or dual-use customers may require NIST 800-171 and CMMC-aligned handling of sensitive technical information.
• The product must preserve evidence at the level of human decisions and edits, not only raw prompt logs.

Competition clusters into five camps: AI-native patent service platforms, enterprise IP management suites, global IP workflow and services platforms, docketing-first law-firm tools, and patent-intelligence/search vendors. None of them visibly owns continuous AI-to-human provenance capture inside the R&D workflow; most begin at invention disclosure, docketing, search, or prosecution.

Why incumbents do not win by default

• Patent management suites. Lightbringer, Anaqua, and Questel already own downstream invention disclosure, portfolio, and prosecution-adjacent workflow, so a startup does not win by being another system of record; it wins by creating evidence before those systems begin.
• Law-firm workflow tools. Docketing and matter-management tools serve counsel operations well, but they are optimized for deadlines and records after a matter exists, not for reconstructing AI-assisted conception inside engineering workflows.
• Patent intelligence platforms. PatSnap- and Questel-style platforms are strong at search, landscapes, and analytics, but that strength does not automatically translate into admissible inventorship provenance or counsel-ready attestation packages.
• Manual counsel and spreadsheets. The default substitute remains attorney-led declarations, spreadsheets, and disclosure forms. That path is entrenched because it is familiar, but it is slow, expensive, and weak at producing contemporaneous evidence.

This company sells compliance-grade invention provenance software to patent-active deep-tech startups that use AI inside R&D and cannot afford an inventorship challenge during fundraising, M&A, or prosecution. The initial customer is a Series A-B quantum or defense hardware startup with 5+ active patent applications, daily use of simulation or generative tools, and an imminent financing or filing event that forces IP diligence. The product wedge is not patent drafting or portfolio management; it is continuous capture of AI-versus-human contribution evidence before the invention disclosure form exists. That wedge fits the research: patent offices still require human inventors, Lightbringer has validated demand for patent-ops software, and current substitutes remain manual declarations, spreadsheets, and outside-counsel coordination. The first proof point is counsel acceptance of a machine-generated provenance package on a live filing or diligence review, because distribution, pricing, and retention all depend on that trust threshold. Go-to-market should therefore start with direct founder sales into acute diligence moments and referral distribution through deep-tech patent counsel and AI-native patent firms, not broad top-of-funnel SaaS marketing. Market size appears real but specialized, with research estimating roughly $77.9M TAM, $23.4M SAM, and a credible year-three SOM near $3.0M, so expansion into biotech, semiconductors, and enterprise R&D labs is necessary for venture scale. A key evidence gap remains how often investors or acquirers already request AI-contribution documentation, so the early company must test demand intensity before scaling headcount or building broad integrations.

Problem

• Deep-tech teams increasingly use AI in simulation, design, and analysis, but patent law still requires proof of human conception and current workflows do not capture that evidence contemporaneously.
• Manual inventor declarations assembled by engineers and outside counsel are slow, expensive, and weak in diligence or litigation because they reconstruct inventorship after the fact.

Solution

• Instrument the R&D workflow to record which AI systems touched which artifacts, which humans accepted or modified outputs, and which decisions support claimed invention.
• Generate editable counsel-ready inventorship packages and portfolio risk dashboards that plug into existing disclosure and prosecution systems instead of replacing them.

Why we win

• Incumbents own downstream filing, docketing, and portfolio workflow; this product starts upstream at the moment AI enters R&D, where no visible category leader has a strong position.
• If counsel accepts the provenance package format, the company compounds a proprietary corpus of reviewed evidence chains, correction patterns, and secure deployments that generic patent tools lack.

Milestones

flowchart LR
  Wedge[Financing or filing trigger] --> MVP[Single-application provenance MVP]
  MVP --> Proof[Counsel-accepted package on live matter]
  Proof --> Expansion[Portfolio coverage plus law-firm referrals]

Founding team

Experiment roadmap

Risk assessment

What must be true

• At least half of qualified pilots convert to annual coverage after the first live package is delivered.
• Outside patent counsel accepts the package format with limited manual rewrite on real matters.
• The first 20 target accounts concentrate workflow usage in a small enough tool set to support with fewer than five core integrations.
• Buyers continue to pay per-application pricing because it is cheaper than repeated manual counsel coordination.
• Adjacent incumbents do not ship equivalent upstream provenance capture before the startup earns trusted distribution through counsel and partners.

Open diligence questions

• How often do recent deep-tech financing or M&A diligence checklists ask about AI-assisted inventorship evidence?
• Which exact tools create the evidentiary surface in the beachhead: notebooks, CAD, simulation, EDA, or lab systems?
• What level of editability and review control do outside counsel require before relying on a machine-generated package?
• Is private cloud sufficient for defense-adjacent buyers, or do the earliest deals require on-prem or air-gapped deployment?
• Can a counsel or AI-native patent-firm channel produce lower CAC than founder-led direct sales after the first pilots?

Model sanity

• Revenue engine. The base case grows from 6 active paying accounts at Y1 exit to 150 by Q4Y3, with roughly $21K of recognized annual revenue per account and most acceleration coming from counsel referrals in Y2-Y3.
• Must go right. Pilot-to-annual conversion and counsel acceptance must clear the BP threshold of 50%+ or the customer ramp never reaches the $3.15M exit ARR implied by the base case.
• Model breaks if. A one-quarter sales-cycle slip plus ARPU compression toward $19K pulls downside cash toward roughly $120K and would force an earlier seed raise.
• Next-round proof. The seed story is credible once the company reaches 25 active paying accounts, has repeatable counsel acceptance, and shows private-cloud delivery without blowing up margin.

Scenarios

Sensitivity

flowchart LR
  Triggers[Financing or filing triggers] --> Pilots[Paid pilot accounts]
  Pilots --> Conversions[Annual covered accounts]
  Conversions --> Revenue[Per-account revenue]
  Revenue --> GrossProfit[Gross profit after delivery cost]
  GrossProfit --> Cash[Cash available for hiring and runway]

Flags: The base case requires referrals and founder sales to carry net account growth from 50 at Q4Y2 to 150 at Q4Y3, so channel proof cannot stay anecdotal. · Y3 recognized revenue is only $2.05M, so the financing story still depends more on exit ARR and channel efficiency than on current profitability. · Gross margin only reaches the BP target in Y3, leaving little room for major private-cloud or counsel-review overruns before the next round.

• Evolving legal standard. USPTO inventorship guidance and appellate case law are still forming; the documentation standard the product targets may shift, requiring constant product updates. Mitigation: Retain a CAFC-specialized patent attorney as a regulatory intelligence advisor and build the contribution graph around durable principles—capturing human creative decisions—rather than specific form requirements that change.
• Patent counsel adoption friction. Outside patent counsel may resist a tool that partially automates their inventorship review work, producing evidence they did not supervise and slowing enterprise sales. Mitigation: Sell to CTOs first as an engineering governance product and position patent counsel as the expert reviewer of provenance packages, not as a role being replaced.
• Narrow initial beachhead. The first addressable segment—AI-heavy deep-tech startups with active patent portfolios—is narrow enough that early pipeline could dry up before the biotech and semiconductor expansion is ready. Mitigation: Run parallel discovery with IP law firms serving deep-tech clients, who aggregate demand across dozens of portfolio companies and can accelerate pipeline beyond the direct startup path.