☷ ℝ  —  ⅟ release: 2025-07-26 ( 2 )  —  rev.: 9  —  transl.:  IT  ·  DE  ·  FR  ·  ES  —  goto:  .⟰.  ·  RAF  ·  Q&A  —   : PDF 


Katia/K2: a geopolitical analysis example

  • 1st edition, this article is written starting from a post of mines, the same day published on Linkedin.

  • Rationale

    I created Katia as a session prompt for dealing with texts, opinions, do peer-review, ratings and summaries in order to create an executive grade assistant. Clearly, Katia discernment capability depends upon reasoning and data from the underlying AI model, this is unavoidable: engine broken, car does ride.

    The Kimi K2 engine has been chosen for this elaboration because the K2 model is pretty new (at the date of writing, less than two weeks in production) and well performing while Katia prompt has been developed for Gemini, specifically. Thus both were at test in this case, as well as the text under the analysis which is a Tymofiy Mylovanov's post, published by his President of Kyiv School of Economics LinkedIn profile.

    For this reason and because of the delicacy of the topic, the most iconic tools from Katia have been leveraged to provide a not-trivial guided text elaboration. Human operator (me) did not add information by his own, thus "guided" means instructing Katia about her tasks without biasing her output. Which includes — in asking her an OSINT-proof remediational rewriting of Mylovanov's post — to keep the original author point of view.


    OSINT-proof peer-review

    Considering the cleantxt as the first publishable text, the Katia summary in [SBI] mode finalsbi is about 2.4 times shorter (42%). This approach is effort-and-knowledge free about the content production, because the only effort-and-knowledge is about how Katia/K2 is functioning (prompt engineering).

    Instead, considering the quality of the output and the IBM-1979 accountability principle, the lnkdpost is the only correct and acceptable approach as long as it includes also a human topic-expert professional content-assement and a final editing as pubblication approval within the media constrains.

    A bit of statistics

    In the comparison below, chars metric includes spaces but not new-lines which linkedin counts, instead:

  • original: 273 words, 1791 chars, original post from the author
  • katia/k2: 519 words, 3353 chars, text provided by Katia/K2 with notes about process elaboration
  • cleantxt: 492 words, 3174 chars, text provided by Katia/K2 notes-free, ready for the pubblication
  • lnkdpost: 457 words, 2964 chars, notes-free text, least-changes linkedin post-size human editing
  • finalsbi: 200 words, 1340 chars, text provided by Katia/K2 summarising its own previous output

  • Katia final output

    The post, published by Tymofiy Mylovanov as president of the Kyiv School of Economics, does not stand well against OSINT-based peer-review analysis by Katia v0.9.26.6 on Moonshot Kimi K2.

    Katia output, and its evaluation of the original Mylovanov's post, has nothing to do with my personal opinions or knowledge: in this case, I have acted solely as a technical operator of an AI agent that I developed also for this specific kind of analysis.

    The text below is the final result of a remediation process involving the Mylovanov's post to make it OSINT-proof. For sake of completeness: on the final Katia/K2 output, strictly non-semantic aesthetic-only changes (presentation layer) were made by human (me).

    RUSSIA'GRIP SLIPS—BUT NOT VANISHES
    (An updated précis, bias retained, OSINT-proofed)

    Introduction: Russia's relative hold on the post-Soviet south is eroding—directional, conditional, reversible.

  • 1. Security vacuum, not void: Moscow's physical footprint shrinks yet persists; Kyrgyz & Tajik accords signed without Russian mediation but Russian forces remain in both countries; Armenia's Indian arms only 32% delivered, Russian orders merely reduced, not ended.
  • 2. Economic diversification—momentum strong, baseline low: China trade doubled, but 18% still re-exports from Russia; Middle Corridor 33× growth equals just 4% of northern rail volume—relative surge, not dominance.
  • 3. Turkish and Gulf footprints—symbolic and material: Baykar factory still under construction; first drones due 2026; Gulf solar 1.8 GW committed but none online—signals future leverage, not present power.
  • 4. Russia's remaining levers—declining, not disappeared: Remittances down 9%, Rosatom Kazakh plant 51% local equity, north-south transit via Iran +12%—niche tools intact yet eroding.
  • 5. Scenarios, not certainties: A Ukraine cease-fire could redeploy 50K troops; control loss is monopoly loss, reversible.
  • Conclusion: In short, Russia's suzerain model is under acute stress, yet every capital still hedges. The drift is directional, not definitive.


    Warning — the time reported by Katia's footer is wrong on Kimi K2, it should have been N/A, instead.

  • The session with Katia/K2 in which the Mylovanov's post has been analysed and elaborated
  • The sessions transcription, docx-exported in full and shared in PDF by a Google Drive link
  • Attenzione e contesto nei chatbot   (2025-07-20)

  • Gemini as your personal financial advisor   (2025-07-13)

  • Gemini as your personal executive assistant   (2025-07-11)

  • Introducing SoNia' seamless chat experience   (2025-07-13)

  • Share alike

    © 2025, Roberto A. Foglietta <roberto.foglietta@gmail.com>, CC BY-NC-ND 4.0


    date legenda: ❶ first draft publishing date or ❷ creation date in git, otherwise ❸ html creation page date. top