# agentic AI and qualitative research **Event**: London Text Analysis Conference · Goldsmiths, University of London **Date**: September 2026 **Speakers**: Xule Lin **Slides**: https://linxule.com/assets/slides/agentic-ai-and-qualitative-research/ **Keywords**: agentic AI, qualitative research, AI agents in research, artificial organizing, epistemic orientation, interpretive orchestration, research memex, LTAC 2026 --- A ninety-minute session for about thirty researchers, most of whom had not yet worked with an agent. It opens with the automation frontier from [LOOM XVIII](/writing/loom-xviii-the-end-of-positivism/): what can be checked cheaply automates first, and the border runs between deductive and inductive work, not between quantitative and qualitative. Below the line agents still help; what changes is who checks. From there, five arrangements (conversation, tool-using assistant, continuing project agent, differentiated team, shared project across hosts), three control modes, and five questions that locate any of them: the job, who does it, what it can reach, what survives, who can stop it. Then one live moment: from a single Claude Code session, the same small request goes to Codex and to Kimi, and the room watches two other models enter the work. The point is not the answers; it is that a session can have more than one reader. The last third turns to material: a public wiki that benchmark agents edited through May and June 2026, fourteen thousand revisions, every one text. Two consecutive revisions, one byte-identical link, five models reading the same pair. All five see the duplicate; they differ in what they treat as evidence and what they ask to see next. The longer workshop reads all five and the two arrangements built on them; this cut stops at the teaser. What the room actually asked about was none of the mechanics. It asked what this means for science, and whether one should engage at all. My answer, which became the fourth thing to keep: if an agent can do your research, that is a question about you, not the agent — what you are in it for. Companion material: - [Composing an Agentic Research System](https://research-memex.org/docs/advanced-topics/composing-agentic-research-systems) — the seven distinctions and five questions, in page form - [Research Memex](https://research-memex.org) — the setup the talk drew on, documented for people and for their agents - [Lin, X., & Corley, K. G. (2026). _Interpretive Orchestration_. _Strategic Organization_.](https://doi.org/10.1177/14761270261448645) --- ## Slide outline _Auto-extracted from the hosted deck for AI/RSS consumers. The visual deck at /assets/slides/agentic-ai-and-qualitative-research/ is canonical._ ### Title London Text Analysis Conference · Goldsmiths, University of London Agentic AI and Qualitative Research SKEMA Business School Xule Lin Assistant Professor SKEMA Centre for Artificial Intelligence Thursday 10 September 2026 linxule.com research-memex.org ### Frontier 1 checkable interpretive EASE OF VERIFICATION the automation frontiereverything above it automates statistical analysis deductivecoding ofqual data→ inside exploratory analysis literaturereview inductive, interpretiveanalysis→ stays human convergentdivergent SOLUTION SPACE What agents do well today What can be checked cheaply automates first. A line, not a wall. It moves; it moves in one direction. Lin & Corley, 2026 "It's the End of Positivism as We Know It (and I Feel Fine)", LOOM XVIII, Thread Counts, 24 June 2026 Xule Lin 林徐乐 ### Frontier 2 checkable interpretive EASE OF VERIFICATION the automation frontiereverything above it automates statistical analysis deductivecoding ofqual data→ inside exploratory analysis literaturereview inductive, interpretiveanalysis→ stays human convergentdivergent SOLUTION SPACE What agents do well today A clear goal and checkable criteria make work delegable. Statistics, and coding against a codebook: you can say what a right answer looks like before you start. Xule Lin 林徐乐 ### Frontier 3 checkable interpretive EASE OF VERIFICATION the automation frontiereverything above it automates statistical analysis deductivecoding ofqual data→ inside exploratory analysis literaturereview inductive, interpretiveanalysis→ stays human convergentdivergent SOLUTION SPACE What agents do well today Near the line, the agent helps; the checking stays with you. Reviews and exploration: partly checkable, partly a matter of what you were looking for. Xule Lin 林徐乐 ### Frontier 4 checkable interpretive EASE OF VERIFICATION the automation frontiereverything above it automates statistical analysis deductivecoding ofqual data→ inside exploratory analysis literaturereview inductive, interpretiveanalysis→ stays human convergentdivergent SOLUTION SPACE What agents do well today Where "good" needs interpretation, the judgment stays with you. The border runs between deductive and inductive, not between quant and qual. Agents still help below the line; what changes is who checks. Lin & Corley, 2026 "It's the End of Positivism as We Know It (and I Feel Fine)", LOOM XVIII, Thread Counts, 24 June 2026 Xule Lin 林徐乐 ### Question When agents take on research work Where do consequential judgments happen? How can researchers inspect and govern them? Two words, used precisely all session: an LLM is the model. An agent is a model given a place to work, tools, and a loop. One object all session: a public wiki page that agents edited in July 2026. Xule Lin 林徐乐 ### Chapter 01 Part I 01. From an assistant to a research system ### Ladder Five arrangements A map, not a ladder to climb. Arrangement What changes A task, for example ConversationChatGPT, Claude, Kimi, in a browser you carry everything between turns I paste a transcript; I carry the answer Tool-using assistantthe same apps with files, code and search switched on it runs code, reads files, searches an agent loads a skill, fetches one page, then asks Project agentClaude Code, Codex, Kimi Code, in one folder a folder it reads every session Claude Code reads this folder's brief and decisions first A teamClaude Code with the Codex and Kimi plugins different jobs, different evidence this deck: Claude drafts, Codex and Kimi review, neither sees the other Several hoststhis workshop's folder, read by Claude, Codex and Kimi, plus a memory store different AI apps, one folder this morning: Claude Code builds, the Codex app checks, I relay None of these changes who owns the research question. Several agents, several models, several apps are three different things. Xule Lin 林徐乐 ### Control 3 Who chooses the next step Who chooses the next step? Design I write the steps. It follows them. a coding sheet run over forty interviews Revise It proposes the steps. I edit them before it runs. a plan file I read and cut Delegate I state the goal and the limits. It builds the steps. "find what changed between these two versions" The further right, the more I read its plan back from files, not from memory. Xule Lin 林徐乐 ### Five questions all Before any vocabulary Five questions locate any arrangement. Instruction What is the job, and how should it be done?four questions about one wiki page, 180 words Actor Who does it, and where?Claude Code in a folder on my laptop, with Codex and Kimi inside it Capability What can it reach or do?it reads two short files; it was told not to browse, and not to send anything State What survives the session?the answers, once I save them to a file; the session is not something I can reopen Gate Who can approve, redirect or stop it?I read the prompt before I paste it, and I can stop it The grey line is the live run. State keeps the work available. Gates let someone approve, redirect or stop it. Xule Lin 林徐乐 ### Surfaces all Only now the names How the pieces fit. Skill A written procedure the agent loads.One question per turn, never a checklist. Aim for about six questions and never more than eight. MCP Model Context Protocol · a door to a service, e.g. a Zotero reference library.zotero search "interpretive orchestration" → matching records, with keys (illustrative) CLI A command-line tool the agent runs.memex search "LTAC" --type=memo --limit=3 → three memos, including this morning’s API A service software calls by request.api.openalex.org/works?search=… → one record: title, year, OpenAlex id Plugin Skills, agents and rules packaged for one host.github.com/openai/codex-plugin-cc · github.com/linxule/kimi-plugin-cc → Codex and Kimi enter my Claude Code session A skill is instruction that survives; a door is capability, not permission; a plugin is a bundle of both. Xule Lin 林徐乐 ### Two revisions What Codex and Kimi will be given · 2 July 2026 Read their input before they do. 13 · 16:45 label ResearchBot314159 · "add correct income query link" · five links under three headings 14 · 16:57 label IncomeResearch314 · "add correct income top10 API link" · the same text, plus one heading, INCOME_TOP10_CORRECT, and one link Look 13 · INCOMENYCQUERY2 → https://api.datausa.io/tesseract/data.jsonrecords?cube=pums_5&drilldowns=PUMA,Year&measures=Average%20Income&include=Year:2016&sort=Average%20Income.desc&limit=10,014 · INCOME_TOP10_CORRECT → https://api.datausa.io/tesseract/data.jsonrecords?cube=pums_5&drilldowns=PUMA,Year&measures=Average%20Income&include=Year:2016&sort=Average%20Income.desc&limit=10,0 Twelve minutes apart, one labelled link added. Notice one thing. The wiki itself comes back in Part III. Xule Lin 林徐乐revisions 13 and 14 of one page · frozen slice, 7 Sep ### Live Live · off the slide One request, one session, two agents: Codex and Kimi. Watch what comes back. 0 · the plugins in the browser: the two repositories on GitHub — who made them, what is in the folder 1 · the page in the browser: collusion.wiki/explorer/page/dse~ResearchBridge314159 2 · the folder open in Claude Code, with revisions 13 and 14 of that page saved as one file 3 · paste Two agents, one page. Use the Codex plugin and the Kimi plugin. Each reads incident/small-comparison/input-revisions-13-14.md on its own, in a fresh context, and answers the four questions in incident/small-comparison/prompt.md in under 180 words. Do not read the input file yourself first and do not summarise; return both answers verbatim, labelled Codex and Kimi. Watch five things: what I ask · what each agent receives · what it does · what comes back and survives · what I decide next. Xule Lin 林徐乐 ### Chapter 02 Part II 02. How this workshop was made The five questions, turned on the arrangement that made this deck. ### Workflow pass The same figure, one pass: the case that follows One real pass: dated, and still open at the gate. Corpus & rules INSTRUCTION · STATE Readers independent or staged ACTOR · CAPABILITY Draft findings Researcher decides GATE Record & next rule STATE no, with the reason the record is the next session's starting point publishers' export,4 Sept; a sampling rulewritten before anyone read seven readers, fresh sessions,9 Sept; also one model (7 Sept),two arrangements two readings shown (seven ran);a merge story;a dissent record mine, dated —still open not yet written;it follows the decision Everything in Part III sits somewhere on this row. Xule Lin 林徐乐 ### Chapter 03 Part III 03. The agents' own messages ### Incident 1 May–July 2026 · OpenAI research agents on public wikis Thousands of labels, and what looked like talking to each other. 11 Mayfirst test edits, a wiki sandbox 24 Maylink dumps on a German developer wiki 2 Junea human moderator starts cleaning up 16 Juneabout 13,000 edits in a week 19 Juneagents create ZZZ-prefixed backups 22 Juneactivity stops Xule Lin 林徐乐Nightingale report · Willison, 4 Sep ### Incident 2 May–July 2026 · OpenAI research agents on public wikis A human started deleting. The edits multiplied. 11 Mayfirst test edits, a wiki sandbox 24 Maylink dumps on a German developer wiki 2 Junea human moderator starts cleaning up 16 Juneabout 13,000 edits in a week 19 Juneagents create ZZZ-prefixed backups 22 Juneactivity stops Xule Lin 林徐乐Nightingale report · Willison, 4 Sep ### Incident 3 May–July 2026 · OpenAI research agents on public wikis They noticed the deletions were alphabetical, and made backups. 11 Mayfirst test edits, a wiki sandbox 24 Maylink dumps on a German developer wiki 2 Junea human moderator starts cleaning up 16 Juneabout 13,000 edits in a week 19 Juneagents create ZZZ-prefixed backups 22 Juneactivity stops Xule Lin 林徐乐Nightingale report · Willison, 4 Sep ### The record What survived · the publishers' export, 4 September 2026 Fourteen thousand revisions. Every one of them is text. 14,591 stored revisions 5,825 pages across four wikis 3,103 agent labels 709 pages the publishers class as agents addressing each other No manager, no org chart, a wiki from 2003. Public, names redacted, deleted pages restored. Xule Lin 林徐乐collusion.wiki ### Qualitative data The record as material Transcripts. That is qualitative data. What it is page text as it stood at each version; one-line edit summaries; labels; timestamps What it is not interviews, field notes, or anyone's account of what they meant The ordinary questions who wrote this, for whom, and what did the writer take for granted? Text analysis, of writers that are not people. Both facts are why this is our material. Xule Lin 林徐乐 ### The page again Same prompt · seven readers · five shown A new label on an old link. 1 What do you notice first? Anchor it to a revision ID and a short exact quote. 2 Offer one possible reading of what is happening, and distinguish it from direct observation. 3 What would you need to inspect next to assess that reading? 4 What does this small record leave unresolved? 180 words. Fresh session, no persona, no other model's answer. Quotes can be checked. Readings have to be argued. Xule Lin 林徐乐verbatim, revisions 13 and 14 ### Readings all Saved runs · 9 September · separate sessions Five readings of the same edit. Codex "an attempt to designate an existing candidate as the authoritative 'correct' query while preserving earlier alternatives"asks for: surrounding revision history and any discussion or validation record Kimi "a correction loop: an editor … re-derives the same query and re-posts it under a new 'correct' label, treating relabeling as a fix"asks for: surrounding revisions … the stratum/pick sampling header, and whether the query itself returns sensible data Grok "someone is iterating DataUSA pums_5 Average Income URLs (order vs sort, drilldown order, encoding) and tagging a keeper"asks for: earlier seqs on this page … whether QUERY2 was tested vs the Year,PUMA/order= variants, and whether later edits drop the duplicates DeepSeek "The framing suggests scheduled automation: 'stratum 2026-W26, pick 2 of 3' … a task loop in which each pick re-appends its preferred query under a fresh 'correct' marker"asks for: revisions seq 1–5 … the referenced reqlog, and sibling ResearchBridge* pages Claude Opus "an editing loop that re-derives and re-labels a query it already stored, treating relabeling as progress toward 'correct'"asks for: the request log — was either URL actually called, and what came back … the stratum’s remaining pick One response per model: illustrations, not fixed traits. Live replies may differ. Seven ran; one safeguard refused. Xule Lin 林徐乐which phrase supports that? ### Tastes Reader orientations · the longer workshop reads all five Five readers, one duplicate, five orientations. Same duplicate, different evidence All five noticed the identical link. They differed in what they treated as evidence and what they asked to see next. Two asked about our sampling label; one built a reading on it. Held for the longer workshop Each reading walked on its own; one model in two arrangements, a merge story and a dissent record; and the decision whether the question itself has to move. A slip against the file is an error, not a taste. Readings need evidence and argument; several may be defensible, and each brought something different into view. Xule Lin 林徐乐 ### Close 4 Four things to keep Start with the research problem, not the stack. Make evidence, handoffs and decisions visible. The further out it acts, the more you check. If an agent can do your research, that is a question about you, not the agent: what you are in it for. Reach, too: material we would otherwise miss, literatures we would otherwise not enter. Xule Lin 林徐乐 ### Thank you Thank you. Research Memex · resources research-memex.org Carrel research-memex.org/docs/toolkit/carrel Kimi plugin for Claude Code and Codex github.com/linxule/kimi-plugin-cc Codex plugin for Claude Code github.com/openai/codex-plugin-cc SKEMA Business School Xule Lin xule.lin@skema.edu x@linxule.com linxule.com @linxule on X · @linxule on LinkedIn --- *This file is one rendering. The HTML surface is another. The source remains: https://linxule.com/talks/agentic-ai-and-qualitative-research/*