<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Tulparus blog</title>
    <link>https://tulparus.ai/blog/</link>
    <atom:link href="https://tulparus.ai/blog/feed.xml" rel="self" type="application/rss+xml" />
    <description>Articles for clinical genetics teams: what the literature says about unresolved cases, where practice loses the evidence, and how Tulparus handles it.</description>
    <language>en</language>
    <lastBuildDate>Sun, 15 Nov 2026 09:00:00 GMT</lastBuildDate>
    <item>
      <title>The cheapest test is a better phenotype.</title>
      <link>https://tulparus.ai/blog/the-cheapest-test-is-a-better-phenotype/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-cheapest-test-is-a-better-phenotype/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Yield follows the quality of the phenotype more than the depth of sequencing. Recording what was examined, what was absent and what was never looked at.</description>
    </item>
    <item>
      <title>The negative finding nobody wrote down.</title>
      <link>https://tulparus.ai/blog/the-negative-finding-nobody-wrote-down/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-negative-finding-nobody-wrote-down/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>An examined and normal finding argues against a whole class of candidates. Most request forms have nowhere to put it, so the laboratory never sees it.</description>
    </item>
    <item>
      <title>The clinical question belongs in the case record.</title>
      <link>https://tulparus.ai/blog/the-clinical-question-belongs-in-the-case-record/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-clinical-question-belongs-in-the-case-record/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A test is ordered to answer a question. When it is not written down, the laboratory answers a different one, and nobody can tell when the case is done.</description>
    </item>
    <item>
      <title>Why possible hypotonia should stay possible.</title>
      <link>https://tulparus.ai/blog/why-possible-hypotonia-should-stay-possible/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/why-possible-hypotonia-should-stay-possible/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Clinicians hedge for good reasons. A record that strips the hedge and stores a firm finding feeds the ranking something the examiner never claimed.</description>
    </item>
    <item>
      <title>Age of onset narrows the search.</title>
      <link>https://tulparus.ai/blog/age-of-onset-narrows-the-search/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/age-of-onset-narrows-the-search/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>When a feature first appeared separates conditions that share the same terms. Forms keep the feature and lose the when, and the ranking cannot use it.</description>
    </item>
    <item>
      <title>When two diagnoses explain one patient.</title>
      <link>https://tulparus.ai/blog/when-two-diagnoses-explain-one-patient/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/when-two-diagnoses-explain-one-patient/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A patient can have two genetic conditions at once, and the blend fits neither. An analysis that stops at the first good answer misses the second.</description>
    </item>
    <item>
      <title>What the referral letter leaves out.</title>
      <link>https://tulparus.ai/blog/what-the-referral-letter-leaves-out/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/what-the-referral-letter-leaves-out/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A request form carries a line or two. The clinic letters carry the patient. A search run on the form alone uses a fraction of what the hospital knows.</description>
    </item>
    <item>
      <title>Kinship checks before anything else.</title>
      <link>https://tulparus.ai/blog/kinship-checks-before-anything-else/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/kinship-checks-before-anything-else/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A trio analysis assumes the three samples are who the form says they are. Checking relatedness from the data first takes minutes and saves cases.</description>
    </item>
    <item>
      <title>The trio you cannot get.</title>
      <link>https://tulparus.ai/blog/the-trio-you-cannot-get/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-trio-you-cannot-get/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Much of a trio's value is knowing what is de novo. With one parent missing, a case can still borrow some of it, if it records what is absent.</description>
    </item>
    <item>
      <title>Reduced penetrance has to be argued.</title>
      <link>https://tulparus.ai/blog/reduced-penetrance-has-to-be-argued/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/reduced-penetrance-has-to-be-argued/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>An unaffected parent who carries the candidate variant is evidence against it. Reduced penetrance can answer that, if argued from the gene's own data.</description>
    </item>
    <item>
      <title>Consanguinity changes the search.</title>
      <link>https://tulparus.ai/blog/consanguinity-changes-the-search/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/consanguinity-changes-the-search/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Related parents shift the odds towards recessive disease and leave a signature in the genome. Both should shape the analysis. Neither settles it.</description>
    </item>
    <item>
      <title>Mosaicism hides in the allele fraction.</title>
      <link>https://tulparus.ai/blog/mosaicism-hides-in-the-allele-fraction/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/mosaicism-hides-in-the-allele-fraction/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A variant present in a fraction of cells shows in a fraction of reads, and a pipeline built for germline calls drops it. The number was there all along.</description>
    </item>
    <item>
      <title>Compound heterozygosity needs phasing.</title>
      <link>https://tulparus.ai/blog/compound-heterozygosity-needs-phasing/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/compound-heterozygosity-needs-phasing/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Two heterozygous variants in a recessive gene explain a case only if they sit on different copies. Without phase, the report is guessing.</description>
    </item>
    <item>
      <title>Beyond the small variant.</title>
      <link>https://tulparus.ai/blog/beyond-the-small-variant/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/beyond-the-small-variant/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>An exome pipeline tuned for small variants is blind to deletions, duplications and repeat expansions. Some negative cases were never looked at properly.</description>
    </item>
    <item>
      <title>Splice variants and the limits of prediction.</title>
      <link>https://tulparus.ai/blog/splice-variants-and-the-limits-of-prediction/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/splice-variants-and-the-limits-of-prediction/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A variant near an exon boundary may wreck splicing or do nothing. Predictors narrow the question. RNA answers it, and the case should say which it has.</description>
    </item>
    <item>
      <title>Synonymous does not mean silent.</title>
      <link>https://tulparus.ai/blog/synonymous-does-not-mean-silent/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/synonymous-does-not-mean-silent/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Variants that leave the protein unchanged are filtered out of most analyses by default. A few alter splicing or expression, and some cause disease.</description>
    </item>
    <item>
      <title>The non-coding genome in a clinical exome era.</title>
      <link>https://tulparus.ai/blog/the-non-coding-genome-in-a-clinical-exome-era/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-non-coding-genome-in-a-clinical-exome-era/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Most of the genome lies outside the exons, and a growing list of diagnoses sits there. A case should record what its test could not see.</description>
    </item>
    <item>
      <title>Reclassifying a VUS is a process.</title>
      <link>https://tulparus.ai/blog/reclassifying-a-vus-is-a-process/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/reclassifying-a-vus-is-a-process/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A variant of uncertain significance is an open question with a list of things that would answer it. Most are filed as answers and never worked on again.</description>
    </item>
    <item>
      <title>One ranking channel is one point of failure.</title>
      <link>https://tulparus.ai/blog/one-ranking-channel-is-one-point-of-failure/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/one-ranking-channel-is-one-point-of-failure/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Every prioritisation method has blind spots set by its inputs. A case ranked by one method inherits them all. Several, fused and shown apart, do not.</description>
    </item>
    <item>
      <title>A novel gene needs a different standard.</title>
      <link>https://tulparus.ai/blog/a-novel-gene-needs-a-different-standard/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/a-novel-gene-needs-a-different-standard/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>When the best candidate has no disease association, the classification rules do not apply. A case can still pursue it, if the standard is written first.</description>
    </item>
    <item>
      <title>Expression tells you where to look.</title>
      <link>https://tulparus.ai/blog/expression-tells-you-where-to-look/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/expression-tells-you-where-to-look/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A gene never switched on in the affected tissue is a weak candidate however rare its variant. Expression belongs in the ranking, with its limits stated.</description>
    </item>
    <item>
      <title>Model organism knockouts in a human differential.</title>
      <link>https://tulparus.ai/blog/model-organism-knockouts-in-a-human-differential/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/model-organism-knockouts-in-a-human-differential/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A mouse with the patient's features is evidence for a gene no human study has reached. It belongs in the ranking, scored as what it is.</description>
    </item>
    <item>
      <title>Similar patients, without sharing patients.</title>
      <link>https://tulparus.ai/blog/similar-patients-without-sharing-patients/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/similar-patients-without-sharing-patients/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The strongest evidence for a rare diagnosis is often another patient like this one. Finding them can be done without moving anyone's record.</description>
    </item>
    <item>
      <title>Similarity earns a look. Then the work begins.</title>
      <link>https://tulparus.ai/blog/similarity-earns-a-look-then-the-work-begins/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/similarity-earns-a-look-then-the-work-begins/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A high similarity score says a gene's phenotype resembles the patient's. It says nothing about the variant, and a number shaped like a verdict is believed.</description>
    </item>
    <item>
      <title>The paralogue and the gene next door.</title>
      <link>https://tulparus.ai/blog/the-paralogue-and-the-gene-next-door/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-paralogue-and-the-gene-next-door/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>When the phenotype points at a gene and the gene is clean, the answer is sometimes in its family. A known-gene filter removes the relatives first.</description>
    </item>
    <item>
      <title>The reranker should explain itself.</title>
      <link>https://tulparus.ai/blog/the-reranker-should-explain-itself/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-reranker-should-explain-itself/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A learned model moved a gene from fourteenth to second. If the only answer to why is that it weighed many features, it has no place in a clinical record.</description>
    </item>
    <item>
      <title>Which ClinVar did you look at?</title>
      <link>https://tulparus.ai/blog/which-clinvar-did-you-look-at/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/which-clinvar-did-you-look-at/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>ClinVar changes with every release. A classification quoted without its date and submitter cannot be checked, and the case cannot be reanalysed.</description>
    </item>
    <item>
      <title>Provenance is a clinical safety feature.</title>
      <link>https://tulparus.ai/blog/provenance-is-a-clinical-safety-feature/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/provenance-is-a-clinical-safety-feature/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Every fact in an interpretation came from somewhere, at some version. Keeping the where and the when is what lets a conclusion be checked when it matters.</description>
    </item>
    <item>
      <title>Immutable evidence, versioned interpretation.</title>
      <link>https://tulparus.ai/blog/immutable-evidence-versioned-interpretation/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/immutable-evidence-versioned-interpretation/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The facts in a case do not change. What they mean does. A system that edits records in place loses the first and cannot track the second.</description>
    </item>
    <item>
      <title>The literature search nobody can repeat.</title>
      <link>https://tulparus.ai/blog/the-literature-search-nobody-can-repeat/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-literature-search-nobody-can-repeat/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Every interpretation leans on a literature search done once, in a browser, and never recorded. The next scientist starts again and may find less.</description>
    </item>
    <item>
      <title>Gene-disease validity comes first.</title>
      <link>https://tulparus.ai/blog/gene-disease-validity-comes-first/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/gene-disease-validity-comes-first/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Before a variant can be pathogenic for a condition, the gene has to be established as causing it. That step is skipped more often than any report admits.</description>
    </item>
    <item>
      <title>When two laboratories disagree about a variant.</title>
      <link>https://tulparus.ai/blog/when-two-laboratories-disagree-about-a-variant/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/when-two-laboratories-disagree-about-a-variant/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The same variant is pathogenic in one submission and uncertain in another. The disagreement is information, and one label in its place throws it away.</description>
    </item>
    <item>
      <title>The ontology changed under the case.</title>
      <link>https://tulparus.ai/blog/the-ontology-changed-under-the-case/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-ontology-changed-under-the-case/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A phenotype is coded in terms, and the terms belong to a release. When the release moves, a case coded three years ago means something slightly different.</description>
    </item>
    <item>
      <title>The unsolved exome ages well.</title>
      <link>https://tulparus.ai/blog/the-unsolved-exome-ages-well/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-unsolved-exome-ages-well/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A negative exome from three years ago is worth more today than when it was reported. What a reanalysis has to carry to be more than a re-run.</description>
    </item>
    <item>
      <title>Let the knowledge trigger the reanalysis.</title>
      <link>https://tulparus.ai/blog/let-the-knowledge-trigger-the-reanalysis/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/let-the-knowledge-trigger-the-reanalysis/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A fixed reanalysis schedule reopens most cases for nothing and misses the one whose answer was published last week. Let the change itself be the trigger.</description>
    </item>
    <item>
      <title>Who tells the family about a reclassification?</title>
      <link>https://tulparus.ai/blog/who-tells-the-family-about-a-reclassification/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/who-tells-the-family-about-a-reclassification/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A classification changes years after the report. The laboratory knows, the clinic has moved on, and the family is making decisions on the old answer.</description>
    </item>
    <item>
      <title>A new gene is a patient waiting.</title>
      <link>https://tulparus.ai/blog/a-new-gene-is-a-patient-waiting/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/a-new-gene-is-a-patient-waiting/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The literature names new genes every month, and many explain patients already sequenced. Most laboratories learn of them by chance, and the patients wait.</description>
    </item>
    <item>
      <title>Write the hypothesis down before you test it.</title>
      <link>https://tulparus.ai/blog/write-the-hypothesis-down-before-you-test-it/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/write-the-hypothesis-down-before-you-test-it/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Unresolved cases drift because the candidates live in people's heads. How to keep competing explanations, and the evidence against them, on record.</description>
    </item>
    <item>
      <title>Render the report from the record.</title>
      <link>https://tulparus.ai/blog/render-the-report-from-the-record/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/render-the-report-from-the-record/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A report written by hand becomes the only account of a case, and the reasoning is lost with it. One rendered from the case record stays true to it.</description>
    </item>
    <item>
      <title>Sign-off is a decision. Record it as one.</title>
      <link>https://tulparus.ai/blog/sign-off-is-a-decision-record-it-as-one/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/sign-off-is-a-decision-record-it-as-one/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A signature on a genomic report is a clinical judgement made under uncertainty. Most systems store it as a timestamp and lose what was judged.</description>
    </item>
    <item>
      <title>What is the next most informative test?</title>
      <link>https://tulparus.ai/blog/what-is-the-next-most-informative-test/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/what-is-the-next-most-informative-test/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>After a negative result the usual next step is the bigger test. The better question is which test would most change the hypotheses already on the table.</description>
    </item>
    <item>
      <title>Knowing what the case does not know.</title>
      <link>https://tulparus.ai/blog/knowing-what-the-case-does-not-know/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/knowing-what-the-case-does-not-know/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A negative result is only as good as the list of what was looked at. Most case records keep the findings and lose the gaps, so the next step is a guess.</description>
    </item>
    <item>
      <title>Turnaround time is set by the slowest handover.</title>
      <link>https://tulparus.ai/blog/turnaround-time-is-set-by-the-slowest-handover/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/turnaround-time-is-set-by-the-slowest-handover/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Sequencing takes days. Cases take months. The difference is queues between people, and the rapid programmes show what happens when the queues are removed.</description>
    </item>
    <item>
      <title>Tenant isolation is a clinical requirement.</title>
      <link>https://tulparus.ai/blog/tenant-isolation-is-a-clinical-requirement/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/tenant-isolation-is-a-clinical-requirement/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A genome identifies a person for life. A platform holding many services' cases has to keep each one's data apart at every layer, and be able to prove it.</description>
    </item>
    <item>
      <title>The case lives in email and spreadsheets.</title>
      <link>https://tulparus.ai/blog/the-case-lives-in-email-and-spreadsheets/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-case-lives-in-email-and-spreadsheets/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The laboratory system holds the sample and the report. Everything between them, candidates, queries and decisions, lives in files owned by individuals.</description>
    </item>
    <item>
      <title>The cost of a case that stays open.</title>
      <link>https://tulparus.ai/blog/the-cost-of-a-case-that-stays-open/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/the-cost-of-a-case-that-stays-open/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>An unresolved case is charged to no budget line, so it looks free. The family, the clinic and the laboratory each pay for it every year.</description>
    </item>
    <item>
      <title>Decision support, and where the decision stays.</title>
      <link>https://tulparus.ai/blog/decision-support-and-where-the-decision-stays/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/decision-support-and-where-the-decision-stays/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>A tool that proposes and a person who decides works only if the record shows which was which. Most systems blur the line, and always in one direction.</description>
    </item>
    <item>
      <title>Patient data trains nobody's model by default.</title>
      <link>https://tulparus.ai/blog/patient-data-trains-nobodys-model-by-default/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/patient-data-trains-nobodys-model-by-default/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>The terms say de-identified data may improve the service. For genomic data that clause is larger than it looks, and the default should run the other way.</description>
    </item>
    <item>
      <title>Benchmarks you can audit.</title>
      <link>https://tulparus.ai/blog/benchmarks-you-can-audit/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/benchmarks-you-can-audit/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Two vendors claim a yield and neither names the set it was measured on. A figure without its evaluation set, sample size and interval is only a sentence.</description>
    </item>
    <item>
      <title>From pilot to practice.</title>
      <link>https://tulparus.ai/blog/from-pilot-to-practice/</link>
      <guid isPermaLink="true">https://tulparus.ai/blog/from-pilot-to-practice/</guid>
      <pubDate>Sun, 15 Nov 2026 09:00:00 GMT</pubDate>
      <description>Most clinical software pilots succeed and most are never adopted. The gap is rarely the tool. It is the governance and the exit nobody wrote down first.</description>
    </item>
  </channel>
</rss>
