Probably Genetic
Genomics company identifying rare disease patients through online discovery, machine learning, and at-home genetic testing.
Website: https://www.probablygenetic.com/
Cover Block
Publicly reported
| Field | Value |
|---|---|
| Company | Probably Genetic |
| Tagline | Genomics company identifying rare disease patients through online discovery, machine learning, and at-home genetic testing [Y Combinator] |
| Headquarters | San Francisco, US [Crunchbase] |
| Founded | 2019 [TechCrunch, June 2019] |
| Stage | Seed [TechCrunch, June 2019] |
| Business model | B2B2C |
| Industry | Healthtech |
| Technology | AI / Machine Learning [Y Combinator] |
| Geography | North America |
| Growth profile | Venture Scale |
| Founding team | Solo Founder, Lukas Lange [TechCrunch, June 2019] |
| Funding label | Undisclosed [TechCrunch, June 2019] |
| Total disclosed | $10,720,000 [CB Insights, November 2023] |
Links
Publicly reported
Summary and Signal
PUBLIC Probably Genetic is a San Francisco genomics company building a B2B2C patient-identification and testing platform for rare disease, and it merits investor attention because the current public record points to a business that sits at the intersection of AI-enabled patient discovery, at-home diagnostics, and pharma-sponsored access programs, with a reported reach of more than 200,000 patient profiles and a recent reported ARPA-H award of up to $10 million [Y Combinator] [Seedtable] [Simplify, August 2026]. The company emerged in 2019 around Lukas Lange, identified as co-founder and CEO, when TechCrunch described the initial product thesis as testing children with autism for more than 15 severe genetic diseases that were often missed inside a broader autism diagnosis [TechCrunch, June 2019].
The product has since broadened into a platform that, according to company and recruiting materials, discovers undiagnosed patients online, analyzes disease state using machine learning and at-home testing, and supports compliant communication with patients, while also running sponsored-testing programs for drug developers and patient-advocacy groups [Y Combinator] [Consider, October 2026]. That positioning is the clearest point of differentiation in the available sources: Probably Genetic is not presented simply as a testing provider, but as a demand-generation and patient-routing layer for rare-disease diagnosis and downstream research or treatment development [Y Combinator] [Crunchbase].
On team quality, the public record is still founder-led but directionally credible. TechCrunch identified Lange as a PhD candidate in bioinformatics and genetics at the University of Oxford in 2019, and later profile data ties him to prior roles at the Bill & Melinda Gates Foundation, The Boston Consulting Group, and PBD Project, although those later biographical details are drawn from directory-style sources rather than named editorial reporting [TechCrunch, June 2019] [RocketReach] [Success.ai].
Funding visibility is partial rather than complete. TechCrunch reported seed backing in 2019 from Khosla Ventures, TenOneTen Ventures, Oxford Angel Fund, and angel investors, with the amount undisclosed, while CB Insights records a $10.72 million Series A on November 1, 2023; Y Combinator and Threshold also appear in current company materials [TechCrunch, June 2019] [CB Insights, November 2023] [Y Combinator]. The business model appears B2B2C, with commercial relationships oriented around drug developers and advocacy groups while patients receive free, subsidized, or insurance-billed testing depending on program structure [Y Combinator] [Consider, October 2026] [Probably Genetic, accessed 2026].
Over the next 12 to 18 months, the key watchpoints are straightforward: whether the company can convert reported patient reach into repeatable enterprise relationships, whether the sponsored-testing model yields durable customer economics, and whether the reported ARPA-H award translates into externally validated product or diagnostic milestones rather than remaining a company-circulated announcement [Seedtable] [Simplify, August 2026]. Hiring signals, including openings across engineering, data science, partnerships, product, and patient experience, suggest the company is still building toward scale, but the public evidence remains thin on named customers, revenue, and independently verified clinical outcomes [Ashby] [Y Combinator].
One source, partially checked -- Section relies on one independently reported company profile from TechCrunch, one funding database entry from CB Insights, and several company or recruiting materials for current product and traction claims.
Taxonomy Snapshot
| Axis | Value |
|---|---|
| Stage | Seed |
| Business Model | B2B2C |
| Industry / Vertical | Healthtech |
| Technology Type | AI / Machine Learning |
| Geography | North America |
| Growth Profile | Venture Scale |
| Founding Team | Solo Founder |
| Funding | Undisclosed, with $10.72M Series A recorded and seed amount undisclosed |
Company Overview
PUBLIC
Probably Genetic appears to have been formed in 2019 and is based in San Francisco, with the company positioned around rare-disease patient identification through online discovery, machine learning, and genetic testing [Crunchbase]. The public record is thin on legal-entity detail, but the operating profile is consistent across startup databases and company materials that describe a genomics business serving patients and partners in rare disease [Crunchbase] [Y Combinator].
The chronology that can be verified from the permitted sources is narrow but useful. Crunchbase lists the company as founded in 2019 and headquartered in San Francisco [Crunchbase]. The company later surfaced as a Y Combinator-backed startup with investors including Khosla Ventures, Threshold, TenOneTen Ventures, Oxford Angel Fund, and Y Combinator, while Crunchbase also associates the business with a November 1, 2023 Series A financing of $10.72 million [Crunchbase]. Taken together, the visible arc is a 2019 formation followed by continued venture backing and a clearer financing marker in late 2023, though public company-site materials do not provide a more detailed milestone history in the sources reviewed [Crunchbase] [Y Combinator].
One source, partially checked -- Relies primarily on Crunchbase for founding date, headquarters, and financing chronology, with partial corroboration from Y Combinator on company identity and investor set.
The Product and the Stack
MIXED
The product story is clearest when separated into two surfaces: patient intake and sponsor-facing discovery. Probably Genetic describes its platform as one that "discovers undiagnosed patients online, analyzes their disease state using machine learning and at-home testing, and enables compliant communication with patients" [Y Combinator]. In plain terms, the company appears to start with online patient discovery, move into symptom or disease-state assessment, and then route qualified users into genetic testing and follow-up pathways, often through no-cost or insurance-covered programs according to company materials [Probably Genetic, accessed 2026].
Public sources show the model has broadened since launch, but the common thread is still rare-disease case finding. At launch in 2019, TechCrunch reported that the initial product was intended to test children with autism for more than 15 severe genetic diseases that were often grouped under an autism diagnosis [TechCrunch, June 2019]. More recent public-facing materials describe free or sponsored genetic testing programs for specific cohorts, including individuals in the U.S. with symptoms suggestive of frontotemporal dementia and people with seizure or developmental-delay related disorders [Crunchbase] [TESS Research Foundation, accessed 2026]. Consider's recruiting materials also describe sponsored-testing programs in which drug developers subsidize testing at little or no cost, which is consistent with the company's stated B2B2C positioning and suggests the software layer is tied to patient recruitment and program operations, not only diagnostic interpretation [Consider, October 2026] [Y Combinator].
The technology stack itself is only partly visible in public evidence, so any precision beyond machine learning should be treated carefully. The company repeatedly frames its system as an AI or machine-learning platform for rare-genetic-disease diagnosis and patient identification, but the available sources do not verify a named proprietary model, a published clinical-validation dataset, or a specific laboratory workflow [LinkedIn] [BioSpace, August 2026] [Unite.AI, August 2026]. Hiring signals point to a software-heavy operation spanning data science, engineering, partnerships, and patient operations, while roles such as Senior Data Scientist, Solutions Architect, Head of Engineering, and Growth Manager imply internal tooling for data workflows, outreach automation, and sponsor integrations (inferred from job postings) [Y Combinator] [Ashby] [LinkedIn, accessed 2026]. The practical takeaway is that the differentiation, if it holds up, likely rests less on a single diagnostic test and more on the combination of patient acquisition, eligibility assessment, compliant engagement, and testing orchestration in a fragmented rare-disease market [Y Combinator] [Consider, October 2026].
One source, partially checked -- Product workflow is corroborated across Y Combinator, TechCrunch, Crunchbase, and program pages, but the deeper AI-stack claims remain partly company-described and not independently validated.
The Market They Are Entering
PUBLIC The market matters now because rare-disease diagnosis remains slow and fragmented, while falling tolerance for diagnostic delay is pushing payers, researchers, advocacy groups, and drug developers toward earlier patient identification and lower-friction testing pathways [TechCrunch, June 2019] [Y Combinator].
Public evidence on market size is thin in the source set, so the cleaner way to frame this market is by function rather than by a single headline TAM. Probably Genetic sits at the intersection of rare-disease diagnostics, patient identification software, and sponsored genetic-testing programs for biopharma and advocacy partners [Y Combinator] [Consider, October 2026]. The available materials support that the company sells into institutions, specifically drug developers and patient-advocacy groups, while patients are the end users in a B2B2C model [Y Combinator]. That matters for market definition: the immediate spend pool is less consumer genomics and more research recruitment, diagnostic access, and programmatic testing budgets tied to rare disease care and drug development [Y Combinator] [Consider, October 2026].
Demand drivers are visible even without a formal third-party market model. The company launched around a concrete diagnostic gap, screening children labeled with autism for more than 15 severe genetic diseases that might otherwise be missed inside a broader clinical category [TechCrunch, June 2019]. Its current positioning extends that same logic to undiagnosed rare-disease patients found online, then triaged through machine learning, at-home testing, and compliant patient communication workflows [Y Combinator]. If that workflow works at scale, the demand is supported by three linked needs: families want answers sooner, advocacy groups want earlier case finding, and drug developers want more efficient patient identification for treatment development and clinical research [Y Combinator] [Consider, October 2026].
The adjacent markets are easier to see than the core market's formal boundaries. One substitute is traditional clinic-led genetics referral, where the patient enters through neurologists, developmental specialists, or genetic counselors rather than through online discovery and remote intake [Crunchbase] [Probably Genetic, accessed 2026]. Another is direct sponsored testing tied to a single therapeutic program, which can overlap with Probably Genetic's model when pharma underwrites testing at little or no cost [Consider, October 2026]. A third adjacency is broader AI-enabled patient recruitment infrastructure, where the commercial value sits less in the test itself and more in identifying, qualifying, and activating hard-to-find patient cohorts for trials or longitudinal research [Y Combinator].
Regulation and macro conditions cut both ways. The favorable side is that compliant communication, physician-ordered testing, and genetic counseling make the service easier to integrate into regulated care pathways than a pure consumer genomics product [Y Combinator] [Crunchbase]. The constraint is that genetics workflows are exposed to reimbursement rules, clinical evidence thresholds, consent requirements, and privacy expectations around health and genomic data, which can slow conversion even when patient interest is high [Crunchbase] [Probably Genetic, accessed 2026]. The reported ARPA-H award, if confirmed through a primary government source later, would also fit a broader public-sector tailwind toward using AI to shorten the rare-disease diagnostic journey, though that specific award remains company-attributed in the available evidence [Simplify, August 2026].
| Cited market signal | What it suggests |
|---|---|
| Drug developers and patient-advocacy groups use the platform [Y Combinator] | Commercial demand may come from institutional program budgets rather than consumer testing spend alone |
| Sponsored-testing programs funded by drug developers [Consider, October 2026] | Biopharma can act as a payer in niches where standard access to testing is limited |
| Initial product targeted more than 15 severe genetic diseases within autism-labeled populations [TechCrunch, June 2019] | The company's wedge is condition-specific case finding, not broad population screening |
| More than 200,000 patients reached or profiled, according to company-linked sources [Seedtable] [Probably Genetic, accessed 2026] | There is at least directional evidence of patient demand, though the figure is not independently audited |
The picture that emerges is a market defined by diagnostic friction and stakeholder misalignment, not by a simple one-line TAM. For investors, the important read-through is that adoption likely depends less on top-of-funnel awareness than on whether the company can repeatedly convert identified patients into completed tests and clinically useful outcomes inside regulated partner programs.
No independent source found -- This section relies primarily on company materials and secondary aggregators for market framing, with limited independent third-party market data in the provided source set.
The Competitive Field
MIXED Probably Genetic appears to sit in a narrower lane than broad consumer genetics or clinical lab players: its public positioning is less about selling a test kit outright and more about finding likely rare-disease patients online, qualifying them algorithmically, and routing them into sponsored or low-cost testing programs for drug developers and advocacy groups [Y Combinator] [Consider, October 2026] [TechCrunch, June 2019].
That creates a competitive map with three distinct groups, even if the available source set does not name direct startup rivals. First are incumbent clinical genetics channels, meaning physicians, hospital systems, and traditional genetic-testing workflows, which remain the default path for many families seeking a diagnosis. Probably Genetic is trying to intercept patients earlier, before or outside that referral chain, through online discovery and at-home testing workflows [Y Combinator] [Crunchbase]. Second are consumer-facing testing and telehealth pathways, which can lower access friction but are not necessarily built around rare-disease case finding or compliant pharma-sponsored recruitment. Third are disease-specific nonprofits and advocacy groups, which often have community trust and patient reach but may lack the software, machine-learning, and operational infrastructure to identify undiagnosed patients at scale. In public materials, the company positions itself at the intersection of those layers rather than squarely inside any one of them [Y Combinator] [TESS Research Foundation] [Consider, October 2026].
The edge that is visible today is distributional rather than purely technical. Public sources repeatedly describe a workflow that combines online patient discovery, machine-learning analysis, compliant communication, and at-home or no-cost testing, with drug developers and advocacy groups as commercial counterparts and patients as end users [Y Combinator] [Consider, October 2026] [Probably Genetic, accessed 2026]. If that workflow is functioning as described, the asset is not simply an AI model. It is the ability to acquire hard-to-find patients, convert them into completed profiles or tests, and do so in a way that sponsors can use for research or treatment-development programs. The reported reach of more than 200,000 patients suggests some scale in top-of-funnel awareness, although the figure is company-reported or database-repeated rather than independently audited [Probably Genetic, accessed 2026] [Seedtable]. That edge could prove durable if repeated program activity generates proprietary phenotype, intake, and engagement data. It is also perishable because patient acquisition channels, symptom-keyword discovery, and sponsored-testing operations can be copied if larger labs, pharma service firms, or better-capitalized startups decide the economics are attractive enough.
The company is most exposed where trust, reimbursement, and clinical integration matter more than discovery alone. A traditional provider-led genetics network has an advantage in physician relationships, care continuity, and diagnostic authority, particularly for patients who already sit inside neurology, pediatrics, or specialty-care pathways. Likewise, a large established testing laboratory, though unnamed in the present source set, would typically control assay operations, payer contracting, and brand recognition with clinicians in ways a younger B2B2C platform may not. Probably Genetic's own public materials point to free testing, insurance-billed care, and sponsored programs, which implies the model still depends on external funding sources, clinical partners, or both to keep patient acquisition economics attractive [Probably Genetic, accessed 2026] [Consider, October 2026]. That makes the company comparatively weaker in channels it does not own outright, especially if advocacy partnerships or sponsor budgets slow.
Over the next 18 months, the most plausible competitive outcome is a split market rather than a single category winner. Probably Genetic is the likely winner if rare-disease drug developers continue shifting budget toward earlier patient identification and sponsor-supported testing, because its public pitch is already aligned to that use case and the company is still hiring across partnerships, data, product, and patient experience functions [Y Combinator] [Ashby] [LinkedIn, accessed 2026]. The more likely loser in that scenario is the slower, referral-only diagnostic pathway, where patients wait to reach a genetics specialist before testing begins. The reverse case is also clear: the referral-based incumbents win if payers, clinicians, or pharma sponsors insist on deeper integration with established care networks and validated lab infrastructure, while Probably Genetic loses ground if online discovery proves expensive to sustain or if company-reported patient reach does not translate into repeatable sponsor value [Probably Genetic, accessed 2026] [TechCrunch, June 2019] [Y Combinator].
No independent source found -- This section relies heavily on company, accelerator, and recruiting materials, with limited independent named-publisher corroboration beyond TechCrunch's 2019 launch coverage and secondary database references [TechCrunch, June 2019] [Y Combinator] [Consider, October 2026].
Opportunity
PUBLIC
The prize here is not a better consumer genetics experience, it is the chance to become a durable patient-intelligence layer for rare disease diagnosis and therapy access if the company can turn patient discovery into a repeatable enterprise workflow for pharma and advocacy groups [Y Combinator] [Consider, October 2026].
The headline opportunity is straightforward: Probably Genetic could become the default front door through which undiagnosed rare-disease patients are identified, triaged, tested, and then matched into care or research pathways. That outcome is still early, but it is not purely aspirational in the available record. The company has described a platform that discovers undiagnosed patients online, analyzes disease state using machine learning and at-home testing, and enables compliant communication with patients [Y Combinator]. Public materials also indicate commercial relevance on both sides of the market: drug developers and patient-advocacy groups are presented as users of the platform, while sponsored-testing programs suggest a business model in which third parties subsidize testing to surface eligible patients earlier [Y Combinator] [Consider, October 2026]. If the company can sit upstream of diagnosis and downstream of therapy development, it has a path to becoming infrastructure rather than a single-point testing vendor.
That possibility is easier to take seriously because the company appears to have moved beyond a single-condition concept. TechCrunch's 2019 reporting described an initial autism-focused at-home testing plan screening for more than 15 severe genetic diseases [TechCrunch, June 2019]. Later public materials point to programs in frontotemporal dementia and seizure or developmental-delay-related disorders, alongside a broader claim of reaching more than 200,000 patients [Crunchbase] [TESS Research Foundation] [Seedtable]. The evidence is uneven and some of it remains company-led, but the directional signal is consistent: Probably Genetic is trying to build a repeatable case-finding and testing motion across multiple rare-disease cohorts rather than a one-off diagnostic product.
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Sponsored testing becomes the wedge | The company becomes a standard outsourced channel for pharma-sponsored genetic testing in selected rare-disease indications, with patient discovery, intake, testing, and compliant follow-up handled in one workflow. | More disease-specific sponsored programs similar to the public descriptions already tied to drug-developer-funded testing [Consider, October 2026]. | The company already publicly frames sponsored testing as part of its operating model and says drug developers use the platform for treatment development and patient identification [Consider, October 2026] [Y Combinator]. |
| Rare-disease intake becomes a patient funnel | Probably Genetic becomes a high-intent intake layer for undiagnosed patients and families, monetizing through testing reimbursement, sponsored programs, and research participation. | Expansion of free assessment and no-cost testing programs across additional symptom clusters or advocacy partnerships [Probably Genetic, accessed 2026] [TESS Research Foundation]. | Public materials show condition-specific programs, free eligibility assessment, and a patient reach claim above 200,000 profiles, which at minimum suggests early top-of-funnel scale [Probably Genetic, accessed 2026] [Seedtable]. |
| Government-backed diagnostic infrastructure | The company uses public-sector support to strengthen its AI and operational stack, then becomes a preferred partner for large-scale rare-disease case finding. | Conversion of the reported ARPA-H award into validated clinical or operational outputs that increase credibility with health-system, payer, or biopharma partners [Simplify, August 2026]. | The reported award, while not independently verified from ARPA-H primary materials in the provided sources, indicates external interest in the thesis that AI-assisted rare-disease diagnosis could be scaled through this platform [Simplify, August 2026]. |
What compounding looks like is a data and distribution flywheel, though it remains only partially evidenced in public. Each additional patient profile can improve symptom-pattern recognition and outreach efficiency, especially if machine-learning models are being used to analyze disease state before testing [Y Combinator]. Each additional sponsored-testing program can widen the set of disease areas in which the company has both patient acquisition pathways and variant interpretation experience [Consider, October 2026]. If patients get answers in weeks, as the company claims in recruiting and social materials, that can also improve referral and advocacy-group credibility, which in turn lowers acquisition friction for the next cohort [LinkedIn, accessed 2026].
There is also an operational compounding effect in the hiring pattern. Current openings span software, data science, product, partnerships, patient experience, content, and CRM automation, which reads less like a single lab service and more like a company trying to scale a full-stack acquisition and care-navigation system [Ashby] [Y Combinator]. That does not prove product-market fit, but it does support the idea that management sees growth coming from repeatable patient sourcing, partner delivery, and workflow automation rather than from selling isolated tests one by one.
The size of the win is difficult to benchmark cleanly from the provided public record because no reliable market-sizing report or direct public comparable is included in the source set. Still, the rough shape of the upside can be stated conditionally. If Probably Genetic were to become a category-level platform sitting between rare-disease patients, advocacy groups, and biopharma programs, the value would likely be driven by repeat enterprise contracts, proprietary patient-intelligence data, and embedded program workflows rather than test revenue alone [Y Combinator] [Consider, October 2026]. In that scenario, not a forecast, the company could plausibly grow into a strategic asset worth several hundred million dollars or more to a diagnostics, life-sciences tooling, or biopharma-services acquirer. The key reason is not the number of tests, but control of a scarce workflow: identifying hard-to-find patients before competitors do, and doing so in a compliant, scalable way. That is the kind of position that can matter disproportionately in rare disease, where patient discovery is often the bottleneck to both care and drug development [Y Combinator] [TechCrunch, June 2019].
No independent source found -- This section relies materially on company, recruiting, and platform descriptions, with limited independent corroboration from TechCrunch, Seedtable, Crunchbase, TESS Research Foundation, and Consider.
Sources
Publicly reported
[Y Combinator] Probably Genetic: Patient Intelligence Platform for Genetic Disease | https://www.ycombinator.com/companies/probably-genetic
[Crunchbase] Probably Genetic - Crunchbase Company Profile & Funding | https://www.crunchbase.com/organization/probably-genetic
[TechCrunch, June 2019] Probably Genetic helps families identify genetic conditions | https://techcrunch.com/2019/06/18/probably-genetic/
[Seedtable] Probably Genetic | Funding, Investors & Team | https://www.seedtable.com/startups/probablygenetic
[Simplify, August 2026] Probably Genetic Awarded Up to $10M ARPA-H Contract to End the Rare Disease Diagnostic Odyssey with AI | https://simplify.jobs/c/Probably-Genetic
[Consider, October 2026] Consider | https://consider.com/boards/vc/tenoneten-ventures/jobs/probably-genetic
[LinkedIn, Accessed 2026] Probably Genetic | LinkedIn | https://www.linkedin.com/company/probablygenetic
[Ashby] Probably Genetic careers | https://jobs.ashbyhq.com/probablygenetic
Articles about Probably Genetic
- A Free Genetic Test for Seizure Disorders Wins Over 200,000 Patient Profiles — The Y Combinator alum, backed by Khosla and Threshold, uses sponsored testing and machine learning to connect patients, advocacy groups, and drug developers.