ProgrammableMarketing/origin.bio (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 6
  • This quarter · 12
  • Big bets · 4
  • Wildcards · 1
  • How we can help
Generated Nov 10, 2026
  1. Programmable
  2. Playbooks
  3. Origin Bio
Origin Bio logo

growth playbook · Nov 10, 2026

origin.bio (opens in a new tab)

Origin Bio growth playbook

Diagnosis

Origin Bio (YC W26, team of 8) has moved from AI-designed regulatory DNA to licensing matched spatial, H&E and clinical tumour data, and it is hiring its first commercial lead to close $100K+ licences. The website has not caught up with that move.

The homepage argues the problem well (response rates of 1.0% to 34.3%, a $2.8 billion median drug cost) but never says "license", "dataset" or "pharma", gives no sample or cohort counts, ends in a bare request form, has the title "What if · Origin Bio" and lists only regulatory DNA posts under Research.

Rivals such as Enable Medicine lead with numbers and buyer teams. This playbook first makes the offer readable: a data catalogue, a qualifying form, a clear title, a research section about tumours and a plain public summary. Next it builds the evaluation path (a public sample slice, a data quality card, a standard evaluation agreement, a guided workspace tour), then the bets: an AI benchmark, buyer-funded cohorts and morphology-only prediction on sponsors' own archived slides.

Growth Score

6/10

Biggest Opportunity

Turn the strong problem story into a buyable product: what data exists, in which cancers and at what quality, and how a pharma or AI team evaluates it, so the new Head of Data Partnerships starts with warm, qualified requests.

Biggest Risk

A seed-stage company with no published scale sells against Owkin (25 million patients, 840 hospitals), Enable Medicine and Noetik. If buyers cannot see cohort size and quality, they default to the bigger names, and the regulatory DNA history can blur what Origin sells.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Say what a buyer can license: a data catalogue section in place of a bare "Request access" form

    Write the cohort table: indication, sample count (current and target), modalities and availability date.

    High impact First result in 2 weeks
  2. 2

    Ask three questions on the access form and route answers to the Head of Data Partnerships

    Add the three fields to the form.

    High impact First result in 2 weeks
  3. 3

    Rename the homepage from "What if · Origin Bio" to what Origin sells

    Rewrite the homepage title and description.

    Medium impact First result in 2 weeks

02

Your idea portfolio

23 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 8 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 4Big bets · 10Fill-ins · 4Reconsider · 5
Effort →

Showing 23 of 23 ideas

This week 6

Quick wins with a clear first action.

  • 01

    Origin's business is licensing matched spatial, H&E and clinical tumour data to pharma and AI companies, with contracts from $100K. Yet the homepage never uses the words "license", "dataset" or "pharma", gives no count of samples, patients or cancer types, does not name the spatial platform, and ends in a four-field "Request access" form that does not say what access gives. A translational or AI team that reaches the bottom of the page knows the problem but not what it would be buying. Enable Medicine, which sells to the same teams, leads with scale numbers and names its buyer teams.

    High impactLow effort First result in 2 weeks Head of Data Partnerships

    First action

    Write the cohort table: indication, sample count (current and target), modalities and availability date.

    How to measure it

    Access requests from pharma, biotech or AI teams naming an indication

    Now
    –
    Target
    10 a month
    By
    90 days

    Steps

    1. 1

      List the cohorts

      Indication, samples profiled and planned, modalities, retrospective or prospective, availability.

      Computational Biologist, Spatial Transcriptomics 2 days

    2. 2

      Decide what to disclose

      Exact counts or ranges, and which cohorts are open for exclusivity.

      CEO 1 day

    3. 3

      Write the spec sheet

      One page per cohort: sample fields, QC summary, clinical labels, licence options.

      Head of Data Partnerships 3 days

    4. 4

      Add the section

      "Data catalogue" above the form, plus a /data page linked from the nav.

      CTO 2 days

    5. 5

      Rename the CTAs

      "Request a cohort spec and sample data" in the header and at the form.

      CTO 1 hour

    6. 6

      Send the spec on request

      Auto-reply with the relevant spec sheet and a booking link.

      Head of Data Partnerships 1 day

    7. 7

      Track

      Requests by buyer type and indication, and request-to-NDA rate.

      Head of Data Partnerships monthly

    First result
    2 weeks
    Cost
    Staff time only
    Owner
    Head of Data Partnerships

    Smallest test

    Add a five-row cohort table above the form and compare request quality for a month.

    Why it works

    Add a "Data catalogue" section above the form, and a matching /data page, with: the cancer types and number of samples per cohort (in profiling or available now), what each sample includes (Xenium spatial transcriptomics, aligned H&E whole-slide image, pathologist annotations, clinical labels), retrospective or prospective, how licensing works (evaluation agreement, then licence; exclusivity and early-access options) and a one-page cohort spec sheet to download after the form. Change the form heading to "Request a cohort spec and sample data" and the header CTA to match. The new Head of Data Partnerships then starts from requests that already know what Origin sells.

    Data buyers shortlist suppliers on cohort fit and quality before they talk to anyone. Showing indication, size and modality turns a curious visitor into a specific request, and it lets Origin compete on the page with rivals that publish their scale.

    Ready to use

    ChecklistData catalogue section

    [ ] Heading: "Data you can license"
    [ ] One row per cohort: cancer type, samples (now / target), retrospective or prospective, availability
    [ ] Per sample: Xenium spatial transcriptomics, aligned H&E whole-slide image, pathologist annotations, clinical labels
    [ ] QC summary line per cohort
    [ ] How licensing works: NDA, evaluation agreement, licence; exclusivity and early access
    [ ] Who it is for: translational, computational biology, AI/ML teams
    [ ] Form: "Request a cohort spec and sample data"
    [ ] Auto-reply with the spec sheet

    Risks and how to handle them

    • Small early numbers look weak next to large rivals

      Lead with depth per sample (matched spatial, H&E and clinical from one lab) and show the profiling target and date.

    • Disclosing cohorts helps competitors

      Publish indication and size only; keep site, sampling and pricing details for the spec sheet behind the form.

    Evidence

    • Your websiteHomepage wording (checked November 9, 2026): the words "license", "licensing", "dataset" and "pharma" do not appear; there is no count of samples, patients or cancer types, and the spatial platform is not named; the closing "Request access" form asks only for first name, last name, work email and company, with no text saying what access gives.Source
    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Y Combinator profileOrigin, "AI and Data for Cancer Therapeutics": Winter 2026 batch, founded 2025, team of 8, San Francisco. The profile says it produces patient-derived cancer datasets for licensing to pharma and AI companies, combining spatial transcriptomics, H&E imaging and clinical labels, runs its own lab and sources samples from biobank and hospital networks. Its YC launch post, "Safer Cell & Gene Therapies with AI", pitched the earlier regulatory DNA work.Source
    • Competitor researchEnable Medicine's homepage: "The data to enable better medicine"; it licenses an indexed clinical, spatial, cellular and molecular dataset, speaks to three buyer teams (discovery, translational, data and AI), shows its scale ("20+ disease areas", "200+ clinical sites", "125M+ cells") and sells Tailored Cohorts.Source
  • 02

    The homepage title, which search results, browser tabs and link previews show, is "What if · Origin Bio". It says nothing about cancer, spatial data or licensing, and "Origin" is a crowded name in biotech. A translational scientist searching for spatial tumour datasets has no reason to click.

    Medium impactLow effort First result in 2 weeks CTO

    First action

    Rewrite the homepage title and description.

    How to measure it

    Non-brand search clicks (Search Console)

    Now
    –
    Target
    +50%
    By
    90 days

    Steps

    1. 1

      Write the title

      Brand plus category: spatial, H&E and clinical tumour data.

      CEO 30 minutes

    2. 2

      Write the description

      Who it is for and what they get.

      Head of Data Partnerships 30 minutes

    3. 3

      Set preview tags

      Social title, description and a workspace image.

      CTO 1 hour

    4. 4

      Fix other pages

      Clear titles on research posts and the new /data page.

      CTO 2 hours

    5. 5

      Verify Search Console

      Submit the homepage and a sitemap.

      CTO 1 hour

    6. 6

      Measure

      Clicks and queries monthly.

      CTO monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    CTO

    Smallest test

    Change the title today and check the preview in a LinkedIn post draft.

    Why it works

    Change the title to something like "Origin Bio: matched spatial, H&E and clinical tumour data for oncology" and write a description that names the buyer and the offer. Check the social preview, then give the Axis, Switch and /data pages their own clear titles.

    The title is the first line of text most buyers see. Naming the category makes brand and category searches land on Origin, and makes shared links self-explanatory in email and on LinkedIn.

    Risks and how to handle them

    • Losing the brand's tone

      Keep "What if" as the H1; change only the title.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): page title "What if · Origin Bio"; H1 "What if we made as many cancer drugs next year as we have in the last decade?"; a six-slide argument on trial costs and response rates comes before the data engine; nav Workspace, Data engine, Research; CTAs "Get Access", "Get in touch" and "See the platform"; a live workspace showing one lung sample ("57,252 predicted cells"); three named academic advisors; investors Y Combinator, Topology, Glasswing and 15 individuals; LinkedIn and X links.Source
    • Your websiteHomepage wording (checked November 9, 2026): the words "license", "licensing", "dataset" and "pharma" do not appear; there is no count of samples, patients or cancer types, and the spatial platform is not named; the closing "Request access" form asks only for first name, last name, work email and company, with no text saying what access gives.Source
  • 03

    The access form asks for name, work email and company only. The first commercial lead is hired to find the teams that control oncology data budgets across pharma, biotech and AI labs, but the form does not say whether a request comes from a translational team, an AI lab or a student, or which cancer they care about.

    High impactLow effort First result in 2 weeks Head of Data Partnerships

    First action

    Add the three fields to the form.

    How to measure it

    Access requests reaching NDA

    Now
    –
    Target
    25%
    By
    90 days

    Steps

    1. 1

      Add the fields

      Team type, cancer types, intended use.

      CTO 2 hours

    2. 2

      Set routing

      Pharma and AI lab requests to the partnerships inbox; academics to a standard reply.

      CTO 2 hours

    3. 3

      Write replies

      One per team type, each with the right spec sheet.

      Head of Data Partnerships 1 day

    4. 4

      Set the response time

      One working day for buyer requests.

      Head of Data Partnerships ongoing

    5. 5

      Log demand

      Cancer types and uses requested, by month.

      Head of Data Partnerships monthly

    6. 6

      Review

      Share demand with the lab team when choosing the next cohort.

      CEO monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Head of Data Partnerships

    Smallest test

    Add only the team-type dropdown this week.

    Why it works

    Add three short fields: team type (translational, computational biology, AI/ML, external innovation, academic), cancer types of interest, and intended use (biomarker discovery, model training, trial design, benchmark). Route pharma and AI lab requests to the Head of Data Partnerships within one working day with the matching cohort spec, and send academics a lighter reply.

    At $100K+ per contract, a few qualified conversations matter more than volume. The answers also tell Origin which cohorts buyers want next, which the role is asked to feed back.

    Risks and how to handle them

    • More fields lower submissions

      Use dropdowns and make cancer type optional.

    Evidence

    • Your websiteHomepage wording (checked November 9, 2026): the words "license", "licensing", "dataset" and "pharma" do not appear; there is no count of samples, patients or cancer types, and the spatial platform is not named; the closing "Request access" form asks only for first name, last name, work email and company, with no text saying what access gives.Source
    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
  • 04

    The homepage sells spatial tumour data, but all three posts in its Research section are about regulatory DNA: Axis, an optimizer test and 10,000 designed sequences. The Axis post still offers free API access through a form, and the YC launch post pitched cell and gene therapy. A pharma buyer checking Origin's depth finds work on a different product.

    Medium impactLow effort First result in 1 month CEO

    First action

    Outline the first tumour-data post from the comp bio QC workflow.

    How to measure it

    Research-to-request rate

    Now
    –
    Target
    5%
    By
    60 days

    Steps

    1. 1

      Outline two posts

      Sample journey and QC.

      Computational Biologist, Spatial Transcriptomics 2 days

    2. 2

      Write and review

      Figures from the workspace sample.

      CTO 3 days

    3. 3

      Regroup Research

      Tumour data first; DNA posts under their own label.

      CTO 2 hours

    4. 4

      Update Axis

      A note on current focus and whether the API offer stands.

      CEO 1 hour

    5. 5

      Link to the catalogue

      Each tumour post ends with the cohort spec request.

      Head of Data Partnerships 1 hour

    6. 6

      Keep a cadence

      One tumour-data post a month.

      CEO monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    CEO

    Smallest test

    Regroup the Research section and add one tumour post.

    Why it works

    Write two short tumour-data posts first (how one sample becomes matched spatial, H&E and clinical data; what QC each sample passes) and put them at the top of Research. Label the DNA posts "Regulatory DNA (Switch)" in a separate group, and add a line to the Axis post saying where the company's focus is now and whether the API offer is still open.

    Buyers read a supplier's own writing to judge depth and focus. Matching the research to the product removes a question before it is asked, while keeping the DNA work visible for the people it serves.

    Risks and how to handle them

    • DNA collaborators feel dropped

      Keep Switch live and say it continues.

    Evidence

    • Your websiteResearch section: three posts, all on regulatory DNA: "Introducing Axis" (October 8, 2025), "Does Muon improve regulatory DNA learning?" (March 5, 2026) and "10,000 AI-designed regulatory DNA sequences" (March 13, 2026). None covers the spatial tumour data the rest of the homepage sells.Source
    • Your websiteIntroducing Axis (October 8, 2025): a model that generates regulatory DNA elements and also predicts their function, aimed at safer gene and cell therapies; it offers free API access to research institutions through a form and asks engineers and researchers to email a hiring address. The site has no careers page.Source
    • Y Combinator profileOrigin, "AI and Data for Cancer Therapeutics": Winter 2026 batch, founded 2025, team of 8, San Francisco. The profile says it produces patient-derived cancer datasets for licensing to pharma and AI companies, combining spatial transcriptomics, H&E imaging and clinical labels, runs its own lab and sources samples from biobank and hospital networks. Its YC launch post, "Safer Cell & Gene Therapies with AI", pitched the earlier regulatory DNA work.Source
  • 05

    Origin's robots.txt rightly reserves its content against AI training and blocks dozens of named training crawlers, and it keeps the /platform workspace out of search. That protects the data, but most public descriptions of Origin elsewhere still mix the old DNA pitch with the new tumour-data one. Buyers increasingly ask AI search tools for "spatial tumour data providers", and those tools work from whatever they can read.

    Medium impactLow effort First result in 1 month CTO

    First action

    Write a 150-word plain description of what Origin sells.

    How to measure it

    Answers to "what does Origin Bio do" that mention tumour data licensing

    Now
    –
    Target
    4 of 5 tools tested
    By
    90 days

    Steps

    1. 1

      Write the summary

      150 words: offer, modalities, buyers, how to request.

      CEO 1 hour

    2. 2

      Publish /about

      Plain text, linked from the footer.

      CTO 2 hours

    3. 3

      Review crawler rules

      Keep training blocks; check that search and live-answer agents can read public pages.

      CTO 2 hours

    4. 4

      Align profiles

      YC, LinkedIn and X bios use the same wording.

      CEO 1 hour

    5. 5

      Test answers

      Ask five AI search tools what Origin Bio does.

      Head of Data Partnerships 1 hour

    6. 6

      Recheck

      Every quarter.

      Head of Data Partnerships quarterly

    First result
    1 month
    Cost
    $0
    Owner
    CTO

    Smallest test

    Update the YC profile description today and retest in a month.

    Why it works

    Keep the training block. Add a short, text-only /about page (what Origin sells, modalities, buyer teams, how to request data) and confirm that search engines and retrieval agents used for live answers, as opposed to training crawlers, can read it and the homepage. Update the YC profile and LinkedIn summary to the same wording.

    One consistent, readable description gets repeated by search snippets, AI answers and directories, so buyers hear the current pitch instead of the cell and gene therapy one.

    Risks and how to handle them

    • Opening the wrong crawlers

      Change only rules for search and live-answer agents; keep every training reservation.

    Evidence

    • Your websiterobots.txt: allows search indexing (search=yes) and reserves against AI training (ai-train=no); blocks dozens of named AI training crawlers; disallows /platform/, /api/ and /static/ for every crawler, so the workspace demo is not indexed.Source
    • Y Combinator profileOrigin, "AI and Data for Cancer Therapeutics": Winter 2026 batch, founded 2025, team of 8, San Francisco. The profile says it produces patient-derived cancer datasets for licensing to pharma and AI companies, combining spatial transcriptomics, H&E imaging and clinical labels, runs its own lab and sources samples from biobank and hospital networks. Its YC launch post, "Safer Cell & Gene Therapies with AI", pitched the earlier regulatory DNA work.Source
  • 06

    Origin's workspace already renders striking spatial maps (one lung sample shows 57,252 predicted cells), and the homepage renders sections from public CC0 and TCGA slides. The Head of Data Partnerships is asked to be Origin's voice on LinkedIn and at conferences. A visual every week is the cheapest way to be remembered by computational biology and translational teams.

    Medium impactLow effort First result in 1 month Head of Data Partnerships

    First action

    Export three maps from the public workspace sample.

    How to measure it

    Access requests from LinkedIn

    Now
    –
    Target
    3 a month
    By
    90 days

    Steps

    1. 1

      Pick sources

      Public CC0 or TCGA-derived sections, or samples cleared for display.

      Computational Biologist, Spatial Transcriptomics 1 day

    2. 2

      Make a template

      Map, two-line finding, modality, link.

      CTO 1 day

    3. 3

      Batch four posts

      One month ahead.

      Head of Data Partnerships 1 day

    4. 4

      Post weekly

      Rotate founders and partnerships lead.

      Head of Data Partnerships weekly

    5. 5

      Reply and connect

      Follow up with commenters from buyer teams.

      Head of Data Partnerships weekly

    6. 6

      Review

      Requests and engagement by topic.

      Head of Data Partnerships monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Head of Data Partnerships

    Smallest test

    Post one map this week and track profile visits.

    Why it works

    Each week post one map from public or licensed-for-display data with a two-line finding (an immune-excluded region, a tumour-stroma boundary, a cell neighbourhood), the modality behind it and a link to the cohort spec. Rotate between the CEO, CTO and partnerships lead.

    Spatial biology is visual and shareable, and repeated, specific posts build recognition with a small, well-connected buyer community.

    Risks and how to handle them

    • Showing data that partners have not cleared

      Use only public datasets or samples with written clearance.

    Evidence

    • Your websiteHomepage (checked November 9, 2026): page title "What if · Origin Bio"; H1 "What if we made as many cancer drugs next year as we have in the last decade?"; a six-slide argument on trial costs and response rates comes before the data engine; nav Workspace, Data engine, Research; CTAs "Get Access", "Get in touch" and "See the platform"; a live workspace showing one lung sample ("57,252 predicted cells"); three named academic advisors; investors Y Combinator, Topology, Glasswing and 15 individuals; LinkedIn and X links.Source
    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source

This quarter 12

The core plays for the next three months.

  • 07

    Every deal runs NDA, then a data evaluation agreement, then a licence, and the partnerships lead is asked to package sample datasets. Before an NDA, a buyer has only one rendered lung sample in a browser. A downloadable slice of a few samples, with the real file formats, lets a computational team check fit in an afternoon.

    High impactMedium effort First result in 6 weeks Computational Biologist

    First action

    Choose samples with clean consent terms and the best QC.

    How to measure it

    Download-to-NDA rate

    Now
    –
    Target
    20%
    By
    1 quarter

    Steps

    1. 1

      Pick samples

      Three to five, consent allowing release.

      Computational Biologist, Spatial Transcriptomics 1 week

    2. 2

      Check terms

      Partner agreements and de-identification.

      CEO 2 weeks

    3. 3

      Package

      Files, data dictionary, QC report.

      Computational Biologist, Spatial Transcriptomics 1 week

    4. 4

      Write the notebook

      Load, view and reproduce one figure.

      Computational Biologist, Spatial Transcriptomics 3 days

    5. 5

      Set the licence

      Research only, no model training, no redistribution.

      CEO 2 days

    6. 6

      Gate and publish

      Behind the qualifying form on /data.

      CTO 2 days

    7. 7

      Follow up

      Call every buyer-team download within two days.

      Head of Data Partnerships ongoing

    First result
    6 weeks
    Cost
    Staff time and legal review
    Owner
    Computational Biologist, Spatial Transcriptomics

    Smallest test

    Share the slice privately with three prospects first.

    Why it works

    Release three to five samples (Xenium output, aligned H&E, annotations and synthetic or consented clinical labels) under a research-only, no-training licence, with a data dictionary and a notebook that reproduces one figure. Gate the download with the qualifying form so each download is a lead.

    Technical buyers trust data they have loaded themselves. A slice shortens evaluation, filters out poor-fit buyers early and gives the partnerships lead a reason to follow up.

    Risks and how to handle them

    • Consent or partner terms do not allow release

      Use samples bought with release rights, or a public-data example with the same format.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • HiringComputational Biologist, Spatial Transcriptomics (San Francisco or remote, $140K-$170K): owns post-run QC and secondary analysis of Origin's Xenium data (transcript detection, segmentation, artefacts, run-to-run variation); aligns H&E whole-slide images to Xenium DAPI images; brings pathologist annotations into the spatial data; analyses cell states, neighbourhoods and tumour-immune interactions; turns analyses into reproducible tasks for evaluating AI models.Source
    • Your websiteHomepage (checked November 9, 2026): page title "What if · Origin Bio"; H1 "What if we made as many cancer drugs next year as we have in the last decade?"; a six-slide argument on trial costs and response rates comes before the data engine; nav Workspace, Data engine, Research; CTAs "Get Access", "Get in touch" and "See the platform"; a live workspace showing one lung sample ("57,252 predicted cells"); three named academic advisors; investors Y Combinator, Topology, Glasswing and 15 individuals; LinkedIn and X links.Source
  • 08

    Origin is hiring a computational biologist to build reproducible QC for its Xenium data: transcript detection, background, segmentation, transcript assignment, imaging artefacts, run-to-run variation and H&E-to-DAPI registration accuracy. Buyers of spatial data worry most about exactly these problems, and no rival homepage reviewed publishes them.

    High impactMedium effort First result in 6 weeks Computational Biologist

    First action

    List the QC metrics the workflow already computes.

    How to measure it

    Data evaluations moving to licence talks

    Now
    –
    Target
    50%
    By
    2 quarters

    Steps

    1. 1

      Pick metrics

      Eight that buyers ask about most.

      Computational Biologist, Spatial Transcriptomics 2 days

    2. 2

      Automate

      Compute per sample in the QC pipeline.

      Computational Biologist, Spatial Transcriptomics 2 weeks

    3. 3

      Design the card

      One page, with ranges and what they mean.

      CTO 3 days

    4. 4

      Validate

      Pathologist check on annotation coverage.

      Computational Biologist, Spatial Transcriptomics 1 week

    5. 5

      Publish

      Summary on /data, full card in specs.

      Head of Data Partnerships 1 day

    6. 6

      Refresh

      With every new batch.

      Computational Biologist, Spatial Transcriptomics per batch

    First result
    6 weeks
    Cost
    Staff time only
    Owner
    Computational Biologist, Spatial Transcriptomics

    Smallest test

    Produce the card for the workspace lung sample.

    Why it works

    Turn the QC workflow into a one-page card per cohort: median transcripts per cell, segmentation quality, registration error, failed-section rate, batch effects across runs and pathologist annotation coverage. Show a summary on /data and the full card in the spec sheet, and refresh it with each new batch.

    Running everything in one lab is Origin's edge over data aggregators, and a quality card turns that edge into a number a scientist can compare.

    Risks and how to handle them

    • Honest metrics show weak batches

      Publish thresholds and exclude failed sections, as buyers will find them anyway.

    Evidence

    • HiringComputational Biologist, Spatial Transcriptomics (San Francisco or remote, $140K-$170K): owns post-run QC and secondary analysis of Origin's Xenium data (transcript detection, segmentation, artefacts, run-to-run variation); aligns H&E whole-slide images to Xenium DAPI images; brings pathologist annotations into the spatial data; analyses cell states, neighbourhoods and tumour-immune interactions; turns analyses into reproducible tasks for evaluating AI models.Source
    • Competitor researchEnable Medicine's homepage: "The data to enable better medicine"; it licenses an indexed clinical, spatial, cellular and molecular dataset, speaks to three buyer teams (discovery, translational, data and AI), shows its scale ("20+ disease areas", "200+ clinical sites", "125M+ cells") and sells Tailored Cohorts.Source
    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
  • 09

    The homepage lists seven approved drugs with their markers, eligible groups and response rates, from 1.0% to 34.3%, and argues that response "could be predicted before enrollment". It stops at the argument. One worked example on a retrospective cohort would show a buyer what the data actually reveals.

    High impactMedium effort First result in 2 months CEO

    First action

    Check which archived cohorts have treatment and response labels.

    How to measure it

    Qualified downloads

    Now
    –
    Target
    40
    By
    1 quarter

    Steps

    1. 1

      Find the cohort

      Treatment and response labels available.

      Computational Biologist, Spatial Transcriptomics 1 week

    2. 2

      Analyse

      Responder vs non-responder spatial features.

      Computational Biologist, Spatial Transcriptomics 3 weeks

    3. 3

      Advisor review

      Ask a scientific advisor to check the claims.

      CEO 1 week

    4. 4

      Write

      Short web version and a PDF.

      CEO 1 week

    5. 5

      Publish

      Link from the drug table and Research.

      CTO 1 day

    6. 6

      Use in outreach

      Send to teams working in that indication.

      Head of Data Partnerships ongoing

    First result
    2 months
    Cost
    Staff time only
    Owner
    CEO

    Smallest test

    A one-figure post from the workspace sample.

    Why it works

    Pick one indication where Origin has archived tissue from patients treated with a listed drug class, and show what separates responders from non-responders spatially (immune infiltration patterns, neighbourhoods, a candidate marker), with honest limits on sample size. Publish a short case study and a longer PDF for the spec sheet.

    Pharma translational teams buy evidence that the data can answer their question. A worked case is easier to forward internally than a pitch slide.

    Risks and how to handle them

    • Small cohort gives weak signal

      Present it as hypothesis-generating and say what a larger cohort would add.

    Evidence

    • Your websiteDrug table: seven approved cancer drugs (Keytruda, Opdivo, Cabometyx, Tecentriq, Yervoy, Erbitux, Stivarga) with marker, eligible group, response and 2025 revenue, each linked to its pivotal trial; response rates within trial-eligible groups range from 1.0% to 34.3%; bringing a cancer drug to market costs "a median of $2.8 billion" (JAMA).Source
    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
  • 10

    Origin's data engine starts with tissue "from hospital and biobank partners" and processes retrospective archives as well as prospective cohorts. Data scale is where Origin looks smallest next to Owkin (840 hospitals) and Enable Medicine (200+ clinical sites), yet the site speaks only to buyers and has nothing for the institutions that supply tissue.

    High impactMedium effort First result in 1 quarter CEO

    First action

    Write down the current partner terms that could be offered publicly.

    How to measure it

    Signed hospital or biobank partners

    Now
    –
    Target
    3
    By
    2 quarters

    Steps

    1. 1

      Define the offer

      Data back, co-authorship and fees.

      CEO 1 week

    2. 2

      Write the page

      Process, consent, benefits and FAQs.

      CEO 1 week

    3. 3

      Legal check

      Consent and data-use language.

      CEO 1 week

    4. 4

      Publish

      Linked from the data engine section.

      CTO 1 day

    5. 5

      Reach out

      Pathology departments and biobanks via advisors and investors.

      CEO ongoing

    6. 6

      Track

      Conversations, agreements and samples received.

      CEO monthly

    First result
    1 quarter
    Cost
    Staff time and legal review
    Owner
    CEO

    Smallest test

    Send the draft offer to two current partners for feedback.

    Why it works

    Publish a /partners page for pathology departments, biobanks and cancer centres: what Origin does with FFPE blocks, what the institution gets back (matched spatial data on their own samples, co-authorship, revenue share or research credit), consent and de-identification approach, and a contact form. Promote it to pathology and biobank networks.

    Every new tissue source grows the catalogue that buyers pay for. Offering data back to the institution is a stronger pitch than payment alone for academic centres.

    Risks and how to handle them

    • Long institutional approval cycles

      Start with commercial biobanks while hospital agreements move.

    Evidence

    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
    • Competitor researchOwkin's homepage: data from 25 million patients across 840 hospitals, a patient data network, an AI scientist product and wet-lab validation, with pharma partners including Sanofi, AstraZeneca and Boehringer Ingelheim.Source
    • Competitor researchEnable Medicine's homepage: "The data to enable better medicine"; it licenses an indexed clinical, spatial, cellular and molecular dataset, speaks to three buyer teams (discovery, translational, data and AI), shows its scale ("20+ disease areas", "200+ clinical sites", "125M+ cells") and sells Tailored Cohorts.Source
  • 11

    The data engine says Origin processes prospective cohorts for "bespoke data requirements", and Enable Medicine sells the same idea as Tailored Cohorts. On Origin's site it is one line, with no way to ask for it.

    High impactMedium effort First result in 1 quarter Head of Data Partnerships

    First action

    Write the feasibility checklist used for the five-day answer.

    How to measure it

    Quotes issued

    Now
    –
    Target
    5
    By
    1 quarter

    Steps

    1. 1

      Define inputs

      Indication, treatment, size, endpoints, timelines.

      Head of Data Partnerships 2 days

    2. 2

      Feasibility checklist

      Tissue access, lab capacity, consent.

      CTO 3 days

    3. 3

      Price bands

      Per-sample and project fees, exclusivity premium.

      CEO 1 week

    4. 4

      Publish

      Offer and spec form on /data.

      CTO 2 days

    5. 5

      Answer in five days

      Feasibility and quote.

      Head of Data Partnerships ongoing

    6. 6

      Review

      Win rate and margin per cohort.

      CEO quarterly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Head of Data Partnerships

    Smallest test

    Offer the custom cohort in the next three buyer calls.

    Why it works

    Create a "Custom cohort" offer: the buyer specifies indication, treatment history, sample count and endpoints; Origin returns a feasibility answer in five working days and a quote with timeline. Publish the inputs, typical sizes and turnaround on /data, with a short spec form.

    Custom cohorts are high-value and sticky, and a named offer with a clear first step is easier for a pharma team to bring into budget planning.

    Risks and how to handle them

    • Lab capacity overrun

      Cap concurrent custom cohorts and show the next start date.

    Builds on

    Evidence

    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
    • Competitor researchEnable Medicine's homepage: "The data to enable better medicine"; it licenses an indexed clinical, spatial, cellular and molecular dataset, speaks to three buyer teams (discovery, translational, data and AI), shows its scale ("20+ disease areas", "200+ clinical sites", "125M+ cells") and sells Tailored Cohorts.Source
  • 12

    The partnerships role is asked to close Origin's first licences and to price exclusivity, early access and multi-year terms. First reference customers matter more than first-year price for a seed-stage supplier competing with companies that name large pharma partners.

    High impactLow effort First result in 1 quarter Head of Data Partnerships

    First action

    Set the founding terms and the reference ask.

    How to measure it

    Signed founding licensees

    Now
    –
    Target
    3
    By
    2 quarters

    Steps

    1. 1

      Set terms

      Early access, roadmap input, locked price, reference.

      CEO 1 week

    2. 2

      Draft the template

      Standard founding licence.

      Head of Data Partnerships 1 week

    3. 3

      Shortlist

      Ten buyer teams from inbound and investor intros.

      Head of Data Partnerships 1 week

    4. 4

      Pitch

      With the spec sheet and quality card.

      Head of Data Partnerships 1 month

    5. 5

      Close

      Through evaluation to licence.

      Head of Data Partnerships 1 quarter

    6. 6

      Announce

      With each licensee's permission.

      CEO on signing

    First result
    1 quarter
    Cost
    Discount on early contracts
    Owner
    Head of Data Partnerships

    Smallest test

    Offer founding terms in the next two late-stage conversations.

    Why it works

    Offer a capped founding-licensee programme: three slots, early access to each new cohort, input on what is profiled next, a multi-year licence at a locked price, in return for a public reference or co-presented result. Describe it on /data without naming prices.

    Scarcity and influence over the roadmap help a buyer justify an early supplier internally, and the references unlock the next deals.

    Risks and how to handle them

    • Locked prices too low

      Lock per-sample price for a fixed term, not forever.

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Competitor researchOwkin's homepage: data from 25 million patients across 840 hospitals, a patient data network, an AI scientist product and wet-lab validation, with pharma partners including Sanofi, AstraZeneca and Boehringer Ingelheim.Source
  • 13

    Origin's pitch is predicting responders before a trial is designed. The sponsors who need that most are running or planning trials in the cancers Origin profiles, and trial registries list them publicly. The partnerships lead is asked to find the teams that control oncology data budgets.

    High impactMedium effort First result in 1 month Head of Data Partnerships

    First action

    Pull the first list for one indication.

    How to measure it

    Meetings with translational or computational teams

    Now
    –
    Target
    8 a month
    By
    1 quarter

    Steps

    1. 1

      Pick indications

      Cohorts available now.

      Head of Data Partnerships 1 day

    2. 2

      Pull trials

      Phase 1-2, active or recently completed.

      Head of Data Partnerships 2 days

    3. 3

      Find the team

      Translational and computational leads.

      Head of Data Partnerships 1 week

    4. 4

      Write the note

      Their indication, Origin's cohort, one figure.

      Head of Data Partnerships 2 days

    5. 5

      Send and follow up

      Three touches over three weeks.

      Head of Data Partnerships 3 weeks

    6. 6

      Refresh

      New trials each month.

      Head of Data Partnerships monthly

    First result
    1 month
    Cost
    Staff time; optional contact data tool
    Owner
    Head of Data Partnerships

    Smallest test

    Twenty notes for one indication.

    Why it works

    Each month, pull sponsors with active or recently completed phase 1-2 oncology trials in Origin's indications from public registries, flag those with low or biomarker-dependent response, and reach out to their translational leads with a note tied to their indication and the relevant cohort spec.

    A message tied to a sponsor's own indication and trial stage is far more relevant than a generic data pitch, and the list refreshes itself.

    Risks and how to handle them

    • Messages read as presumptuous about a sponsor's trial

      Refer only to public registry facts and offer data, not judgments.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Your websiteDrug table: seven approved cancer drugs (Keytruda, Opdivo, Cabometyx, Tecentriq, Yervoy, Erbitux, Stivarga) with marker, eligible group, response and 2025 revenue, each linked to its pivotal trial; response rates within trial-eligible groups range from 1.0% to 34.3%; bringing a cancer drug to market costs "a median of $2.8 billion" (JAMA).Source
  • 14

    Buyers comparing Origin with Enable Medicine, Owkin and Noetik have no neutral way to compare spatial tumour data: matched modalities, platform, QC, consent, exclusivity. Noetik also develops its own cancer drugs, which matters to a pharma buyer choosing a data partner.

    Medium impactMedium effort First result in 1 month CEO

    First action

    Draft the 12 questions with the partnerships lead.

    How to measure it

    Qualified downloads

    Now
    –
    Target
    50
    By
    1 quarter

    Steps

    1. 1

      Draft questions

      Twelve buyer checks.

      Head of Data Partnerships 2 days

    2. 2

      Answer for Origin

      Honest, with links to the quality card.

      CEO 3 days

    3. 3

      Review

      Scientific and legal check of claims.

      CTO 3 days

    4. 4

      Publish

      Web page plus PDF.

      CTO 1 day

    5. 5

      Promote

      LinkedIn and outbound notes.

      Head of Data Partnerships 2 weeks

    6. 6

      Update

      Twice a year.

      CEO 6 months

    First result
    1 month
    Cost
    Staff time only
    Owner
    CEO

    Smallest test

    Share the checklist as one LinkedIn post.

    Why it works

    Write a guide with a checklist of 12 questions to ask any spatial data supplier (matched H&E and clinical data, single-lab production, QC metrics, consent scope, exclusivity, model-training rights, whether the supplier develops competing drugs) and answer them for Origin at the end. Keep rivals unnamed or describe them only from their public pages.

    Teaching buyers how to choose sets the criteria on which Origin is strongest, and the guide ranks for searches like "spatial transcriptomics data licensing".

    Risks and how to handle them

    • Claims about rivals go out of date

      Keep the guide about questions, not competitor scorecards.

    Builds on

    Evidence

    • Competitor researchNoetik's homepage: an "AI-native biotechnology company" with "one of the largest collections of multimodal tumor data", used for trials that enrich for likely responders, drug positioning and target discovery.Source
    • Competitor researchEnable Medicine's homepage: "The data to enable better medicine"; it licenses an indexed clinical, spatial, cellular and molecular dataset, speaks to three buyer teams (discovery, translational, data and AI), shows its scale ("20+ disease areas", "200+ clinical sites", "125M+ cells") and sells Tailored Cohorts.Source
    • Competitor researchOwkin's homepage: data from 25 million patients across 840 hospitals, a patient data network, an AI scientist product and wet-lab validation, with pharma partners including Sanofi, AstraZeneca and Boehringer Ingelheim.Source
  • 15

    The homepage opens with a live workspace of one lung sample ("Click any cell to see more information", 57,252 predicted cells) and "See the platform" points back to it. A visitor can click cells, but nothing guides them to what a buyer needs to see: data depth per sample, QC and how to get more.

    Medium impactMedium effort First result in 1 month CTO

    First action

    Script the three stops.

    How to measure it

    Tour completions that submit a request

    Now
    –
    Target
    8%
    By
    1 quarter

    Steps

    1. 1

      Script stops

      Modalities, finding, QC.

      Computational Biologist, Spatial Transcriptomics 2 days

    2. 2

      Build the tour

      Overlay on the workspace.

      CTO 1 week

    3. 3

      Add the CTA

      Request this cohort's spec.

      CTO 1 day

    4. 4

      Track

      Starts, completions, requests.

      CTO 1 day

    5. 5

      Test with buyers

      Three prospects try it on a call.

      Head of Data Partnerships 2 weeks

    6. 6

      Iterate

      Fix drop-off points.

      CTO monthly

    First result
    1 month
    Cost
    Engineering time
    Owner
    CTO

    Smallest test

    Add three static captions beside the workspace.

    Why it works

    Add a three-stop guided tour on the workspace: (1) the matched modalities for this sample, (2) a cell neighbourhood or tumour-immune finding, (3) the QC summary, ending with "Request this cohort's spec". Track tour completion and requests from it.

    Interactive demos convert when they tell a story; a guided path turns an impressive visual into a sales argument.

    Risks and how to handle them

    • Tour slows the page

      Load it only after the workspace is ready.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): page title "What if · Origin Bio"; H1 "What if we made as many cancer drugs next year as we have in the last decade?"; a six-slide argument on trial costs and response rates comes before the data engine; nav Workspace, Data engine, Research; CTAs "Get Access", "Get in touch" and "See the platform"; a live workspace showing one lung sample ("57,252 predicted cells"); three named academic advisors; investors Y Combinator, Topology, Glasswing and 15 individuals; LinkedIn and X links.Source
    • Your websiterobots.txt: allows search indexing (search=yes) and reserves against AI training (ai-train=no); blocks dozens of named AI training crawlers; disallows /platform/, /api/ and /static/ for every crawler, so the workspace demo is not indexed.Source
  • 16

    Every licence passes through an NDA and a data evaluation agreement before the licence itself. For a seed-stage supplier with $100K+ contracts, each week spent negotiating evaluation terms is a week a buyer's champion can lose interest.

    High impactMedium effort First result in 1 quarter Head of Data Partnerships

    First action

    Draft the evaluation agreement template.

    How to measure it

    Median NDA-to-decision time

    Now
    –
    Target
    30 days
    By
    2 quarters

    Steps

    1. 1

      Draft the template

      Scope, access period, no-training, deletion.

      Head of Data Partnerships 1 week

    2. 2

      Legal review

      Outside counsel.

      CEO 1 week

    3. 3

      Build the data room

      Per cohort: samples, dictionary, quality card.

      Computational Biologist, Spatial Transcriptomics 1 week

    4. 4

      Write the plan

      Kickoff, mid-point, decision.

      Head of Data Partnerships 1 day

    5. 5

      Announce

      On /data and in outreach.

      CTO 1 day

    6. 6

      Measure

      Cycle time and conversion.

      Head of Data Partnerships monthly

    First result
    1 quarter
    Cost
    Legal review
    Owner
    Head of Data Partnerships

    Smallest test

    Use the template in the next evaluation.

    Why it works

    Publish a short, standard evaluation agreement (scope, time-limited access, no training on the evaluation data, deletion at the end) and a two-week evaluation plan: kickoff, a data room with the cohort and quality card, a mid-point call with the computational biologist and a decision meeting. Mention "two-week evaluation" on /data.

    A clear, short evaluation lowers the internal effort for the buyer's legal and science teams and gives the partnerships lead a predictable sales cycle.

    Risks and how to handle them

    • Large pharma insists on its own paper

      Offer the template as a starting point and accept theirs when faster.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
  • 17

    Origin keeps producing samples, and the partnerships lead prices multi-year terms. A licence that only covers a fixed snapshot gives the buyer no reason to renew, while a stream of new samples does.

    High impactLow effort First result in 2 quarters Head of Data Partnerships

    First action

    Write the release note template.

    How to measure it

    Licences renewed or expanded

    Now
    –
    Target
    80%
    By
    4 quarters

    Steps

    1. 1

      Set the cadence

      Quarterly drop dates.

      CTO 1 day

    2. 2

      Release note

      Template: new samples, QC, annotations, one analysis.

      Computational Biologist, Spatial Transcriptomics 2 days

    3. 3

      Deliver

      Data room update and email.

      Computational Biologist, Spatial Transcriptomics quarterly

    4. 4

      Review call

      30 minutes per licensee.

      Head of Data Partnerships quarterly

    5. 5

      Log requests

      Feed into cohort planning.

      Head of Data Partnerships quarterly

    6. 6

      Renew early

      Open renewal talks two drops before expiry.

      Head of Data Partnerships ongoing

    First result
    2 quarters
    Cost
    Staff time only
    Owner
    Head of Data Partnerships

    Smallest test

    Send one release note to the first licensee.

    Why it works

    Structure licences around quarterly drops: new samples, updated QC cards, new annotations and a short release note with what changed and one example analysis. Hold a 30-minute review call with each licensee per drop to capture what they want next.

    Regular, visible value keeps the data in use inside the buyer's team, and the calls give Origin the demand signal it needs for the next cohort.

    Risks and how to handle them

    • Drops slip

      Ship what is ready, with a note on what follows.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
  • 18

    The partnerships lead is asked to be Origin's voice at conferences. Translational and computational oncology buyers gather at a few meetings a year, and a poster with real Origin data gives a reason to talk that a booth alone does not.

    Medium impactMedium effort First result in 2 quarters Head of Data Partnerships

    First action

    List the next three meetings and their abstract deadlines.

    How to measure it

    Buyer meetings per conference

    Now
    –
    Target
    12
    By
    2 quarters

    Steps

    1. 1

      Pick the meeting

      Oncology or spatial biology, by buyer attendance.

      Head of Data Partnerships 1 day

    2. 2

      Write the abstract

      Quality card or responder case.

      Computational Biologist, Spatial Transcriptomics 2 weeks

    3. 3

      Make the poster

      With a QR code to the cohort spec.

      CTO 1 week

    4. 4

      Pre-book

      Email target teams with slots.

      Head of Data Partnerships 3 weeks

    5. 5

      Attend

      Poster session and meetings.

      Head of Data Partnerships 1 week

    6. 6

      Follow up

      Within 48 hours, with the spec sheet.

      Head of Data Partnerships 2 days

    First result
    2 quarters
    Cost
    Registration and travel
    Owner
    Head of Data Partnerships

    Smallest test

    Pre-book meetings at one meeting already on the calendar.

    Why it works

    Submit an abstract built on the quality card or the responder case study to the next suitable oncology or spatial biology meeting. Before the meeting, email target teams with the poster number and offer 20-minute slots; afterwards, post the poster on the Research page.

    Scientific buyers trust peer-facing work, and pre-booked meetings turn an expensive trip into pipeline.

    Risks and how to handle them

    • Abstract rejected

      Attend anyway with pre-booked meetings and a printed one-pager.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • HiringComputational Biologist, Spatial Transcriptomics (San Francisco or remote, $140K-$170K): owns post-run QC and secondary analysis of Origin's Xenium data (transcript detection, segmentation, artefacts, run-to-run variation); aligns H&E whole-slide images to Xenium DAPI images; brings pathologist annotations into the spatial data; analyses cell states, neighbourhoods and tumour-immune interactions; turns analyses into reproducible tasks for evaluating AI models.Source

Big bets 4

Larger plays that can change the trajectory.

  • 19

    Origin wants to sell AI evaluation benchmarks to AI labs, and its computational biologist is hired to turn analyses into reproducible tasks for evaluating AI models. The homepage says the data can train models to "use morphology alone to predict spatial gene expression". A public leaderboard on a held-out set makes Origin the reference for that task.

    High impactHigh effort First result in 2 quarters CTO

    First action

    Choose the first task and size the hidden test set.

    How to measure it

    Unique submitting teams

    Now
    –
    Target
    25
    By
    3 quarters

    Steps

    1. 1

      Define tasks

      H&E to expression, H&E to cell type.

      Computational Biologist, Spatial Transcriptomics 2 weeks

    2. 2

      Split data

      Public validation, hidden test, no patient overlap.

      Computational Biologist, Spatial Transcriptomics 2 weeks

    3. 3

      Build scoring

      Submission and evaluation service.

      CTO 1 month

    4. 4

      Baseline

      Origin's own model score.

      CTO 2 weeks

    5. 5

      Launch

      Leaderboard, paper-style write-up, licence.

      CEO 1 week

    6. 6

      Sell the full set

      Offer private tasks to labs that submit.

      Head of Data Partnerships ongoing

    First result
    2 quarters
    Cost
    Engineering time and hosting
    Owner
    CTO

    Smallest test

    Run the task privately with two friendly groups.

    Why it works

    Release a benchmark: public H&E inputs and a small public validation set, a hidden test set of matched spatial expression that Origin scores, and tasks such as predicting expression and cell types from H&E. Publish a leaderboard and a baseline. The full private set and harder tasks become the paid offer for AI labs.

    AI labs pay attention to benchmarks that others compete on. The hidden test set keeps the data protected, and every submission is a relationship with a potential buyer.

    Risks and how to handle them

    • Test labels leak

      Score server-side only and limit submissions per day.

    • Little uptake

      Launch with two academic groups already committed to submit.

    Builds on

    Evidence

    • HiringComputational Biologist, Spatial Transcriptomics (San Francisco or remote, $140K-$170K): owns post-run QC and secondary analysis of Origin's Xenium data (transcript detection, segmentation, artefacts, run-to-run variation); aligns H&E whole-slide images to Xenium DAPI images; brings pathologist annotations into the spatial data; analyses cell states, neighbourhoods and tumour-immune interactions; turns analyses into reproducible tasks for evaluating AI models.Source
    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
  • 20

    Profiling tumours is capital-intensive, and the partnerships lead is asked to feed buyer needs back into what Origin produces next and to price early access. Today Origin chooses cohorts first and looks for buyers afterwards.

    High impactHigh effort First result in 2 quarters CEO

    First action

    Choose three candidate cohorts and a commitment threshold for each.

    How to measure it

    Pre-committed contract value

    Now
    –
    Target
    $300K
    By
    3 quarters

    Steps

    1. 1

      Pick candidates

      Three indications with tissue access.

      CEO 2 weeks

    2. 2

      Set thresholds

      Commitment needed to start each.

      CEO 1 week

    3. 3

      Draft terms

      Deposit, head start, exclusivity window.

      Head of Data Partnerships 2 weeks

    4. 4

      Publish the roadmap

      On /data, without prices.

      CTO 1 day

    5. 5

      Sell commitments

      To shortlisted buyer teams.

      Head of Data Partnerships 1 quarter

    6. 6

      Start profiling

      First cohort to cross its threshold.

      CTO on threshold

    First result
    2 quarters
    Cost
    Staff and legal time
    Owner
    CEO

    Smallest test

    Ask three current prospects which cohort they would pay to accelerate.

    Why it works

    Publish a cohort roadmap (the next three indications under consideration) and let buyers reserve early access with a deposit or a signed commitment. Profile the cohort that reaches its commitment threshold first, and give committing buyers a time-limited head start or exclusivity.

    Pre-commitments fund the lab work, lower risk and make sure every cohort has buyers, the same logic as pre-sales for any expensive product.

    Risks and how to handle them

    • No cohort reaches its threshold

      Set modest thresholds and combine a commitment with existing archives.

    Builds on

    Evidence

    • HiringHead of Data Partnerships (San Francisco, $200K-$250K, 1-1.5% equity): "This hire will be our first commercial lead." Closes the first data licensing deals with pharma, biotech and AI labs (translational science, computational biology, AI/ML and external innovation teams); runs NDA, data evaluation agreement and licence, with contracts from $100K; packages cohort specs, sample datasets and a data room; prices exclusivity, early access and multi-year terms; feeds buyer needs into what Origin produces next; represents Origin at conferences and on LinkedIn; helps sell AI evaluation benchmarks and RL environments to AI labs.Source
    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
  • 21

    Origin says enough matched data lets it "use morphology alone to predict spatial gene expression, predict drug response". Sponsors already hold H&E slides from their own completed trials, with known outcomes. A model trained on Origin's matched data could read those slides without any new tissue.

    High impactHigh effort First result in 3 quarters CTO

    First action

    Measure how well current models predict spatial features from H&E on held-out samples.

    How to measure it

    Signed studies

    Now
    –
    Target
    2
    By
    4 quarters

    Steps

    1. 1

      Validate the model

      Held-out accuracy by feature.

      CTO 1 quarter

    2. 2

      Define the study

      Inputs, outputs, timeline, limits.

      Computational Biologist, Spatial Transcriptomics 2 weeks

    3. 3

      Pilot on public data

      A public trial cohort with slides and outcomes, if available.

      Computational Biologist, Spatial Transcriptomics 1 month

    4. 4

      Price it

      Per study plus optional spatial confirmation.

      CEO 1 week

    5. 5

      Sell the first study

      To a founding licensee.

      Head of Data Partnerships 1 quarter

    6. 6

      Publish with consent

      A case study of the result.

      CEO after study

    First result
    3 quarters
    Cost
    Model development and compute
    Owner
    CTO

    Smallest test

    Score one public H&E cohort and share the result with a prospect.

    Why it works

    Offer a service: the sponsor shares de-identified H&E images and outcomes from a completed trial, Origin infers spatial features from morphology and tests which features separate responders, then reports candidate enrichment markers for the next trial. Price per study, with an option to profile a subset of blocks with spatial transcriptomics to confirm.

    It turns the homepage promise into something a sponsor can buy on data it already owns, with a fast answer, and it creates demand for Origin's spatial profiling to validate the result.

    Risks and how to handle them

    • Morphology predictions are too weak

      Offer only features that pass a validation threshold and confirm with spatial profiling.

    Builds on

    Evidence

    • Your websiteData engine: 01 "It starts with a real tumor" from "hospital and biobank partners"; 03 "Matched spatial transcriptomics, H&E imaging and clinical data for each patient tumor sample"; 05 "We process retrospective and prospective cohorts" to "leverage existing archives, while addressing bespoke data requirements"; 06 "Use morphology alone to predict spatial gene expression, predict drug response, identify interpretable biomarkers"; 07 "Optimized trials cost less and reach more patients."Source
    • Your websiteDrug table: seven approved cancer drugs (Keytruda, Opdivo, Cabometyx, Tecentriq, Yervoy, Erbitux, Stivarga) with marker, eligible group, response and 2025 revenue, each linked to its pivotal trial; response rates within trial-eligible groups range from 1.0% to 34.3%; bringing a cancer drug to market costs "a median of $2.8 billion" (JAMA).Source
  • 22

    Origin lists three academic advisors and works with hospital and biobank partners, but its public research so far is blog posts about regulatory DNA. Rivals cite published studies, and pharma scientists check the literature before trusting a new data source.

    Medium impactHigh effort First result in 3 quarters CEO

    First action

    Agree authorship and target journal with one tissue partner.

    How to measure it

    Preprint live

    Now
    –
    Target
    1 preprint
    By
    3 quarters

    Steps

    1. 1

      Agree authorship

      Partners and advisors.

      CEO 2 weeks

    2. 2

      Write methods

      From the QC and processing pipeline.

      Computational Biologist, Spatial Transcriptomics 1 month

    3. 3

      Example analyses

      Two findings that show value.

      Computational Biologist, Spatial Transcriptomics 1 month

    4. 4

      Internal review

      Advisors read the draft.

      CEO 3 weeks

    5. 5

      Post the preprint

      Then submit to a journal.

      CEO 1 week

    6. 6

      Promote

      LinkedIn, Research page, spec sheets.

      Head of Data Partnerships 2 weeks

    First result
    3 quarters
    Cost
    Staff time and publication fees
    Owner
    CEO

    Smallest test

    A short methods preprint on the QC workflow.

    Why it works

    Write a dataset descriptor paper for the first tumour cohort: sample collection, Xenium and H&E processing, QC, annotations, example findings and access terms. Co-author with tissue partners and advisors, post a preprint first, and link it from /data and every spec sheet.

    A citable paper is lasting proof of method and quality, rewards the institutions that supply tissue, and is something buyers can forward to sceptical colleagues.

    Risks and how to handle them

    • Paper reveals too much

      Describe methods and a subset; keep full data under licence.

    Builds on

    Evidence

    • Your websiteHomepage (checked November 9, 2026): page title "What if · Origin Bio"; H1 "What if we made as many cancer drugs next year as we have in the last decade?"; a six-slide argument on trial costs and response rates comes before the data engine; nav Workspace, Data engine, Research; CTAs "Get Access", "Get in touch" and "See the platform"; a live workspace showing one lung sample ("57,252 predicted cells"); three named academic advisors; investors Y Combinator, Topology, Glasswing and 15 individuals; LinkedIn and X links.Source
    • Your websiteResearch section: three posts, all on regulatory DNA: "Introducing Axis" (October 8, 2025), "Does Muon improve regulatory DNA learning?" (March 5, 2026) and "10,000 AI-designed regulatory DNA sequences" (March 13, 2026). None covers the spatial tumour data the rest of the homepage sells.Source
    • HiringComputational Biologist, Spatial Transcriptomics (San Francisco or remote, $140K-$170K): owns post-run QC and secondary analysis of Origin's Xenium data (transcript detection, segmentation, artefacts, run-to-run variation); aligns H&E whole-slide images to Xenium DAPI images; brings pathologist annotations into the spatial data; analyses cell states, neighbourhoods and tumour-immune interactions; turns analyses into reproducible tasks for evaluating AI models.Source

Wildcards 1

Unexpected experiments, each with a cheap way to test it.

  • 23

    Origin's homepage already contains the core of a citable dataset: approved cancer drugs with their markers, eligible groups, response rates (1.0% to 34.3%) and revenue, each linked to its pivotal trial. Expanded to every approved oncology drug, it becomes a public reference that journalists, investors and pharma strategy teams would cite.

    Medium impactMedium effort First result in 1 quarter CEO

    First action

    Extend the current seven-row table to 30 drugs as a pilot.

    How to measure it

    Referring domains

    Now
    –
    Target
    30
    By
    2 quarters

    Steps

    1. 1

      Define methods

      Which trials, which endpoints, how revenue is sourced.

      CEO 1 week

    2. 2

      Compile

      All approved oncology drugs with sources.

      Computational Biologist, Spatial Transcriptomics 1 month

    3. 3

      Check

      Advisor review of method and outliers.

      CEO 2 weeks

    4. 4

      Publish

      Page, table download and charts.

      CTO 1 week

    5. 5

      Pitch

      Trade press and analyst newsletters.

      Head of Data Partnerships 2 weeks

    6. 6

      Update

      Each year.

      CEO annual

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    CEO

    Smallest test

    Publish the 30-drug pilot table on LinkedIn.

    Why it works

    Compile the response rate within the eligible group for all approved oncology drugs from pivotal trial publications, and estimate the revenue spent on treating non-responders. Publish it as an annual index with methods, a downloadable table and three charts, and update it each year.

    Original, sourced numbers earn links and press in a way product pages do not, and every mention restates the problem Origin solves.

    Risks and how to handle them

    • Response rates are not comparable across trials

      State endpoints clearly and group by endpoint type.

    Evidence

    • Your websiteDrug table: seven approved cancer drugs (Keytruda, Opdivo, Cabometyx, Tecentriq, Yervoy, Erbitux, Stivarga) with marker, eligible group, response and 2025 revenue, each linked to its pivotal trial; response rates within trial-eligible groups range from 1.0% to 34.3%; bringing a cancer drug to market costs "a median of $2.8 billion" (JAMA).Source

03

How Programmable can help

Where an idea in this playbook is something Programmable builds or runs, it's listed here with what we'd do. Everything else is yours to run.

ProgrammableInfo

Turn the research into a list you can act on

These ideas start with finding the right companies. Info finds them and the right person at each, researches every one and drafts the outreach, then asks before it sends.

  • Build this list in Info for idea #13 (opens programmable.info in a new tab)

New accounts: 100,000 free credits with code BETA100K

Find the companies (opens programmable.info in a new tab)
ProgrammableDigital

Show it, don't just describe it

These ideas need something to show: a demo, a video, an animation, graphics or a deck. Digital turns screenshots or the site into all of them, in the company's own brand.

Make a demo (opens programmable.digital in a new tab)
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Appfire growth strategyAppfire builds apps that extend Atlassian (and Salesforce and Microsoft) tools: workflow automation (JMWE, JSU, Power Scripts), portfolio…21 growth ideas
  • Vela growth strategyAn AI scheduling agent that recruiters and search partners CC on email or text. It books interviews across candidates, clients and panels…23 growth ideas
  • Darkroom growth strategyAn AI-native growth marketing agency for consumer brands: paid media, performance creative, Amazon and TikTok Shop, retention, AI search…23 growth ideas
  • Telnyx growth strategyA licensed carrier and cloud platform that owns its network, edge compute and GPUs: Voice AI agents, inference, voice and SIP, messaging…21 growth ideas
  • ShipStation Global (formerly Auctane) growth strategyA shipping and logistics group formed in 2026 from Auctane and WWEX Group, running ShipStation, Stamps.com, Worldwide Express…22 growth ideas
  • Titan Advanced Energy Solutions growth strategyBuilds IonSight, an ultrasound and industrial-AI inspection platform that scans every lithium-ion battery cell in line or at inbound…21 growth ideas
  • Daily Dump growth strategyA Bengaluru design company that makes home and community composters (terracotta Khamba, Gobble, Gobble Mini and Max, bulk converters)…21 growth ideas
  • Scott AI growth strategyThe planning layer for teams that build with coding agents: a local plugin syncs agent sessions, planspaces turn them into shared design…23 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 10, 2026, and cites them throughout. Origin Bio did not commission, review or endorse it, and Programmable is not affiliated with Origin Bio. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Origin Bio runs its business.

Something wrong, or do you represent Origin Bio? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 306 ideas

    Make the offer readable: data catalogue, qualifying form, page title, tumour-focused research section, a plain summary for AI search, and a weekly spatial map post.

  2. Days 31 to 6010 ideas

    Build the evaluation path: a public sample slice, a data quality card, a responder case study, the biobank page, custom cohorts, founding licensees, registry outbound, the buyer's guide, a guided tour and a standard evaluation agreement.

  3. Days 61 to 907 ideas

    Retention and the bets: quarterly cohort drops, a conference poster, the public benchmark, buyer-funded cohorts, morphology-only prediction on archived slides, a dataset paper and the responder gap index.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value matched spatial, H&E and clinical data from the same tumour. The most cited problem: low response rates inside trial-eligible groups.

  • Data produced in Origin's own lab rather than aggregated
  • Expensive trials that enrol likely non-responders
  • In their words: “responders”

Market

Competitive landscape

3 competitors compared: Enable Medicine, Owkin and Noetik.

  • Gap to exploit: Published quality metrics for spatial tumour data
  • Enable Medicine: Broad across diseases, less oncology-specific
  • Owkin: Large platform, less focused on matched spatial data as a product

Hiring signals

Origin is a team of 8 hiring its first commercial lead, a Head of Data Partnerships who will close the first $100K+ data licences with pharma, biotech and AI labs, package cohort specs, sample data and a data room…

  • Head of Data Partnerships: First commercial lead; $100K+ data licences; benchmarks for AI labs. Ideas 1, 3, 12, 16, 20.
  • Computational Biologist, Spatial Transcriptomics: Xenium QC, H&E registration and AI evaluation tasks. Ideas 8, 19, 21.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks first (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence