16 May 2026

 

AI over the last day has focused on efficiency, new reasoning models, and steady tool updates, while genealogists are using these systems to plan research, analyze messy evidence, and draft polished stories faster than ever. Below is a concise “today’s briefing” plus a toolbox of concrete use cases you can try immediately in your own workflow.

A. Named releases & features (last ~72 hours)

• Anthropic – Claude “dreaming” in Managed Agents (research preview)
  New background-review feature that lets Claude agents scan past sessions for patterns, extend their own “memory,” and improve at handling long, multi-step tasks with minimal steering.

• Anthropic – Outcomes & multi‑agent orchestration for Managed Agents (public beta)
  Lets you define concrete goals (“outcomes”) and have multiple Claude agents coordinate via webhooks and tools to complete complex workflows, such as multi‑stage research projects.

• Anthropic – Expanded Managed Agents webhooks
  Adds richer webhook support so Claude agents can trigger external systems and tools more reliably, enabling tighter integration with research infrastructure and automations.

• Anthropic – 20+ new MCP connectors (legal focus)
  While marketed to law firms, these new connectors show Anthropic’s push toward domain‑specific workflows and hint at similar patterns genealogists can mirror using MCP for archives, catalogs, and databases.

• Anthropic – Managed Agents access update (May 13, 2026)
  Managed Agents capabilities (including dreaming, outcomes, and orchestration) advanced to research preview/public beta, making sophisticated “AI coworker” setups more widely accessible to power users.

• OpenAI – GPT‑5.5 Instant as the default ChatGPT model (rolled out, still propagating)
  GPT‑5.5 Instant is now the standard ChatGPT foundation model, with better reliability, lower hallucinations, and improved coding/knowledge work while staying fast and responsive.

• OpenAI – GPT‑5.5 and GPT‑5.5 Pro in the API
  GPT‑5.5 family models are available to developers with stronger capabilities for research, data analysis, and multi‑step reasoning, enabling more robust custom genealogy tools and scripts.

• OpenAI – New real‑time voice models (live translation & speech‑to‑text)
  Three new voice models support voice reasoning, live speech‑to‑speech translation in 70+ languages, and improved live transcription, helpful for oral history work and interviews.

• Google – Gemini “Personal Intelligence” expansion
  Gemini’s Personal Intelligence, which connects to Gmail, Drive, Photos, and more, is expanding globally (with some regional exclusions), strengthening personalized, cross‑document assistance.

• Google – Gemini Notebooks inside Gemini app
  NotebookLM‑style “Notebooks” now live in the Gemini app, giving you a structured place to organize chats, research notes, and source excerpts as a single, AI‑aware project.

• Google – Gemini native macOS app
  A dedicated Gemini app for Mac now lets genealogists access AI help directly from the desktop, better for long‑form writing, multi‑window research, and file‑based workflows.

• Google – Visual explanations and interactive visuals in Gemini
  Gemini can now turn complex questions into interactive visuals directly in chat, making it easier to map out timelines, migration paths, or record types visually.

• Perplexity – “Personal Computer” feature
  Perplexity’s new “personal computer” mode allows the assistant to work across your local environment and apps (beyond just the browser), pushing toward a true research assistant that can help manage files and context.

• Perplexity – Assistant continues multi‑app, multi‑modal support
  The Perplexity Assistant can act across multiple apps and modalities (including camera and on‑screen content), providing context‑aware help for whatever is on your device.

• Google – Gemini 3.1 for Google Home ecosystem
  Gemini 3.1, with improved reasoning, is rolling into Google Home and related devices, making voice‑driven queries and reminders smarter for at‑home, hands‑free genealogy tasks.

• Anthropic – Claude for Small Business (recent announcement context)
  While not genealogy‑specific, Claude for Small Business packages Claude capabilities with higher limits and collaboration features that small research practices or genealogy shops can leverage.

• Open‑weight models – Ongoing “reasoning” models trend
  The broader ecosystem continues to chase “reasoning”‑style models (inspired by o‑series and others), emphasizing step‑by‑step thinking and chain‑of‑thought, which benefits complex source evaluation and correlation.

• Genealogy‑oriented AI tools – Goldie May AI assistant updates (recent RootsTech‑era context)
  Goldie May’s AI now helps with research objectives, source‑gap analysis, pedigree reviews, and timeline “subway map” visualizations integrated with FamilySearch and desktop genealogy software.

---

B. Implications for genealogists this week

The headline this week is agents and memory: Anthropic’s Managed Agents “dreaming” and orchestration, plus OpenAI’s more capable GPT‑5.5 Instant default, make it more realistic to treat AI as an ongoing research coworker instead of a one‑off Q&A box. This is especially relevant for long, multi‑month projects like multi‑ancestor studies, cluster research, or locality guides where continuity and pattern‑spotting matter.

The second theme is desktop‑integrated, project‑centric workflows. Google’s Gemini Notebooks, the new Mac app, and Perplexity’s “personal computer” feature all push toward AI that sits alongside your documents, scans, and software rather than isolated in a browser tab. For genealogists, this means more natural ways to connect PDFs, images, spreadsheets, and narrative drafts inside a single AI‑assisted research environment.

Finally, voice and multimedia get more serious. OpenAI’s new real‑time voice models, Gemini’s visual explanations, and evolving open‑weight reasoning models give you richer tools for oral histories, multilingual materials, and complex evidence correlation. In practice, that translates into better support for interviews, foreign‑language records, and “explain this pile of records to me like a research mentor” style sessions.

20+ concrete AI use cases for genealogists

Each item is something you could drop into today’s research session or lesson prep.

Planning and research design

1. Turn a problem into a research plan
   Paste a focused problem statement (e.g., “Identify parents of John Smith, born about 1810, living in X County, 1840–1870”) and have AI propose prioritized record sets (civil, church, land, probate, local histories) and repositories, then you annotate with specific collections and call numbers.familysearch+1

2. Generate locality and record‑type guides
   Ask AI for an overview of a county or parish—jurisdictions over time, boundary changes, vital, land, probate, and church record coverage, plus major repositories—then you correct, localize, and add citations before using it as a handout or blog post.last24zotero.blogspot+1

3. Create step‑by‑step workflows
   Describe how you search a specific database (e.g., full‑text newspaper search or FamilySearch image browsing), and let AI turn it into a numbered workflow or checklist you can save, teach from, or publish.familyhistorystorytelling.wordpress+2

4. Build annual or quarterly research roadmaps
   Provide your research priorities for the year and have AI slot them into monthly or weekly tasks, including record types to tackle, repositories to contact, and times to review DNA matches, giving you a living research calendar.familysearch+1

5. Set up “watch lists” for new collections
   Use AI to periodically scan announcement feeds from major sites (FamilySearch, Ancestry, state archives) and extract only new or updated collections relevant to your surnames or localities into a short list with links and dates.last24zotero.blogspot+1

Working with documents and data

6. Transcribe and clean text from images
   Use multimodal/OCR‑capable models to transcribe typed documents and increasingly decent-quality historical handwriting (deeds, probate packets, parish registers), then have AI normalize spacing, expand abbreviations, and flag segments where the handwriting is uncertain.youtubefamilysearch

7. Summarize long records into abstracts
   Paste a long pension file, probate inventory, or land case transcript and ask AI for a structured abstract listing key people, dates, places, relationships, and a brief summary of each major document within the file.denyseallen.substack+2

8. Turn messy notes into a research log
   Dump raw notes from a research session (screenshots, clipped text, quick jottings) into AI and ask it to reorganize into a table or structured log with date, repository/site, collection, search terms, results, and next steps.familyhistorystorytelling.wordpress+1

9. Standardize place names and dates
   Provide a list of inconsistent place and date formats pulled from your database export and ask AI to normalize them (e.g., modern county/state names, standardized date formats), while retaining original spellings in a notes column.familysearch+1

10. Extract entities from narrative sources
    Paste a county history sketch or obituary and have AI pull out all individuals, relationships, places, occupations, and time periods into a bullet list or table you can then verify and enter into your software.denyseallen.substack+2

Evidence analysis and problem‑solving

11. Compare conflicting evidence in plain language
    Give AI multiple transcriptions or abstracts that disagree (ages, birthplaces, parents, migration paths) and ask it to lay out the conflicts side by side and narrate where they agree and where they diverge—without reaching a conclusion for you.last24zotero.blogspot

12. Identify gaps and “negative evidence” opportunities
    Provide a summary of what you’ve checked for a person or couple; AI can list obvious record types or time periods you have not yet examined, helping you plan additional searches.denyseallen.substack+1

13. Generate research hypotheses and alternative explanations
    After you describe a thorny problem (e.g., two men of the same name in the same county), AI can suggest multiple plausible scenarios and point out which records might discriminate between them, such as land chains or tax lists.familysearch+2

14. Timeline construction for identity and migration
    Paste information about multiple individuals with similar names; AI can build side‑by‑side timelines of events, locations, and associates to help you distinguish identity and track migration paths.last24zotero.blogspot+1

15. Clarify jurisdictional changes
    Ask AI to outline how a county or town’s jurisdiction changed over a certain time span and which offices held which records when, so you know whether to look at the parent county, a territorial jurisdiction, or a later state‑level archive.familysearch+1

Writing, teaching, and publishing

16. Draft readable family‑history narratives
    From structured notes (problem, evidence, conclusion), have AI generate a narrative aimed at lay family members, focusing on story, context, and clear explanations; you then revise, add citations, and integrate images.familyhistoryfanatics+3

17. Polish proof arguments and reports
    Paste a draft report or proof argument and use AI as a line editor to tighten sentences, improve transitions, and flag areas where your reasoning leaps too quickly or lacks explicit support.familyhistorystorytelling.wordpress+1

18. Create parallel versions for different audiences
    From one master narrative, ask AI to produce a “technical” version for peers (with more methodology) and a shorter, story‑focused version for relatives or blog readers, while preserving factual consistency.familyhistorystorytelling.wordpress+1

19. Design lesson plans and handouts
    Give AI your learning objectives for a class (e.g., “Using AI safely with census records” or “Planning research with AI without copying errors”) and ask for an outline, timing suggestions, examples, and a draft one‑page handout.familyhistorystorytelling.wordpress+2

20. Convert workflows into tutorials
    Once you have a successful pattern for working with a database or set of records, describe what you do and let AI rephrase it into a step‑by‑step tutorial with headings, tips, and “common pitfalls” boxes suitable for students.familyhistoryfanatics+2

21. Draft and refine blog posts and newsletters
    Paste topic bullets or a rough draft, and have AI expand them into a concise blog post or email newsletter, keeping paragraphs short and suggesting subheadings; then you inject your own voice, stories, and exact citations.familyhistoryfanatics+2

22. Generate illustrative examples using public figures or sample data
    For teaching, ask AI for generic sample families or public‑domain examples (e.g., 19th‑century migration scenarios) and use those to demonstrate research methods in class or in a blog post while protecting your clients’ privacy.familysearch+1

Working with online platforms and AI‑enhanced records

23. Leverage AI‑indexed and hinting systems
    Major platforms now use AI to index records, match data across datasets, and suggest potential relatives or attached records; you can treat these suggestions as leads, then independently verify them with your own evaluation and citations.familysearch

24. Ask AI about record types you haven’t used before
    Use conversational queries such as “What kinds of information might I find in Oklahoma probate packets for the 1890s?” or “How do city directories help resolve identity conflicts between same‑name men?” to discover fresh record types to add to your toolkit.familyhistoryfanatics+3

25. Document privacy, ethics, and AI settings for clients
    Ask AI to help you draft a one‑page explanation of how you use AI in your research and writing, what you do and do not send to external systems (e.g., avoiding living persons’ details), and how you protect client data, then adapt it to your practice.familyhistorystorytelling.wordpress+1