2 February 2026

Here’s a concise, blog-style briefing you could drop almost straight into a daily post.

---

1. Major AI updates in the last day or so

• A new research model called Leviathan is getting attention for squeezing more “capacity” out of fewer parameters using a continuous embedding generator, making smaller models perform like much larger ones and improving sample efficiency over long training runs.[quantumzeitgeist]​

• Google continues to push Gemini 3.x deeper into products (Gmail, search, and creative tools), including AI-summarized inbox views and drafting assistance in Gmail that can condense long threads and propose replies.fortune+1

• Meta leadership is teasing a major 2026 rollout of agentic AI tools—AI “agents” that can act on your behalf in commerce and messaging—which hints at more automation around customer contact, content distribution, and possibly ad management on Facebook and Instagram.[techcrunch]​

• Google DeepMind’s “Project Genie” has started rolling out to some Google AI Ultra subscribers; it provides AI-generated “world building” for creators, which is essentially structured scenario and setting generation for visual and narrative projects.[mediapost]​

• January roundups emphasize a shift from just “bigger models” toward deployment and integration—Gemini more tightly embedded across Google’s ecosystem, healthcare-focused platforms from OpenAI and Anthropic, and broader “AI in every device” ambitions from major players.riskinfo+2

• Newsletter and industry briefs (e.g., “Brain Pulse” for Feb. 1) highlight continued iteration on frontier models (GPT‑5.x, Gemini 3, Claude 4.x, Grok 3), but the real story for practitioners is more polished tools for long-context reasoning, multimodal input (text + images), and domain-tuned assistants.mean+2

For a working genealogist, the takeaway is: (a) long-context models are getting better at holding an entire research problem in mind, (b) multimodal tools are improving at reading images and handwriting, and (c) generative features are creeping into everyday tools like email and document editors.many-roads+3

---

2. Twenty-plus concrete AI uses for genealogists

These are phrased as “immediately try this” ideas for research, analysis, writing, teaching, and blogging.

1. AI-assisted transcription of handwriting

   • Use an AI OCR or handwriting model to transcribe wills, deeds, church registers, or letters, then manually correct and annotate; this builds on the same approach Ancestry used at scale for the 1950 U.S. census.[many-roads]​

2. Batch indexing of small collections

   • Feed scans of family or local history book or cemetery register page-by-page to an AI to extract names, dates, and places into a spreadsheet you can then proof and publish as a index.[many-roads]​

3. Record linkage suggestions

   • Paste details from a census entry, a baptism record, and a burial record into a chat model and ask it to propose whether they refer to the same individual, and why, making sure you evaluate its reasoning against the Genealogical Proof Standard.denyseallen.substack+1

4. Hypothesis brainstorming for brick walls

   • Present a brick-wall summary (e.g., a man vanishing between censuses) and ask an AI to list plausible hypotheses and targeted record types (tax lists, land partitions, guardianship records) to test each one.emptybranchesonthefamilytree+1

5. Research-plan drafting

   • Give the AI a concise research question plus what you’ve already checked; ask for a prioritized research plan with repositories, record types, and time frames, then adapt it into your own formal plan.denyseallen.substack+1

6. Source survey for a locality

   • Ask the model for an overview of typical record sets and substitutes for a specific county or parish and time period, using this as a rough checklist to refine with your own locality knowledge.emptybranchesonthefamilytree+1

7. Timeline construction and gap spotting

   • Paste all known events for a person or family and ask the AI to generate a chronological timeline and highlight gaps, inconsistencies, and periods that need more evidence.denyseallen.substack+1

8. Place and boundary context summaries

   • Have the AI summarize jurisdictional changes, parish splits, or county boundary shifts for your research area and time period, then check those against gazetteers or local histories.many-roads+1

9. Interpreting archaic occupations and terms

   • Paste unfamiliar occupations or legal terms from records (e.g., “cordwainer”, “husbandman”, “yeoman”, or Latin legal phrases) and get brief, era-specific explanations you can quote or adapt in your notes.[many-roads]​

10. Language assistance for foreign records

    • Use AI translation to get draft readings of records in German, Latin, Polish, etc., then refine them with dictionaries and paleography guides, especially when dealing with standard formulaic entries.[many-roads]​

11. Narrative drafts for ancestor biographies

    • Provide a timeline and key facts and have the model draft a short biographical sketch; treat it as a first-pass narrative to revise for accuracy, nuance, and proper citation.[many-roads]​

12. Contextual historical background paragraphs

    • Ask AI to write a paragraph about a specific event or context your ancestor lived through (e.g., a local civil war campaign or an epidemic) to weave into blog posts or family history chapters.[many-roads]​

13. Writing prompts for oral history interviews

    • Generate era- and place-appropriate interview questions for living relatives—similar to StoryCorps’ AI-driven prompts that helped families surface stories they’d never discussed before.[many-roads]​

14. Teaching aids for classes and workshops

    • Create simple examples, case studies, and “mystery ancestor” exercises using AI-generated composite families, focusing on record analysis and accuracy rather than real people.[youtube]​[many-roads]​

15. Ethics and standards discussion scenarios

    • Ask AI to generate scenarios involving DNA surprises, sensitive discoveries, or privacy dilemmas, then use them to teach responsible AI and genealogy practices in classes or blog posts, echoing RootsTech-style ethics discussions.[youtube]​

16. Accessibility improvements to materials

    • Use AI to produce simplified summaries of dense research write-ups, alternative text for images, and audio versions of blog posts to support readers with disabilities, echoing the Accessible Genealogy Project’s goals.[many-roads]​

17. Idea generation for blog series

    • Have AI propose multi-part blog series structures (e.g., “Five records that changed this line” or “Following one farm through four censuses”) tailored to your existing content and audience.facebook+1

18. Drafting social media posts promoting new articles

    • Paste your latest post and ask for several short, platform-specific blurbs (FB group, X, Instagram caption, newsletter teaser) to promote it, then edit for tone and accuracy.[facebook]​

19. Citation scaffolding

    • Give the AI a description of a source (e.g., a baptismal record entry with repository and call number) and ask it to suggest a citation in your preferred style; then adjust it to conform to Evidence Explained or your house style.

20. Experimental narrative formats (letters, prayers, sermons)

    • For faith-centered family histories, have AI draft creative formats—like a “letter” from an ancestor based only on documented facts or a short reflection tied to a family story—and then rewrite to ensure factual soundness.

21. Curriculum outlines for genealogy classes

    • Ask AI to build a 4–6 week syllabus for a society or community genealogy course, including learning outcomes, homework ideas, and ethical touchpoints where appropriate.[youtube]​[many-roads]​

22. Genealogy / AI policy drafts for your practice

    • Have it generate a short “AI use” statement for your blog or client work (how you use AI, what you never delegate to it, and how you protect privacy), guided by emerging conversations about responsible AI in genealogy.[youtube]​[many-roads]​

Each of these can be trialed in a single sitting: pick one research problem or one piece of writing you’re already working on, give the AI a tightly framed task, and then evaluate its output as you would a junior assistant’s work—helpful, but always checked and documented.