Google Cloud Yazindua Open Knowledge Format ili Kuwezesha AI Agents

Google Cloud imezindua Open Knowledge Format (OKF), maelezo mapya yaliyoundwa ili kuweka maarifa ya shirika katika mfumo mmoja kupitia faili za Markdown zinazoweza kubebeka. Kwa kubadilisha data zilizotawanyika kuwa mfumo unaoweza kutumika na mifumo mbalimbali, OKF inalenga kutatua pengo kubwa la muktadha ambalo kwa sasa linazuia AI agents kufanya kazi kwa ufanisi katika mifumo tata ya mashirika.

Kutatua Tatizo la Utawanyiko katika Mitindo ya Kazi ya AI Agents

Kadiri AI agents zinavyozidi kuwa na uwezo wa kujitegemea, zinakabiliwa na kikwazo kikubwa: "utawanyiko wa muktadha" (context fragmentation). Kwa sasa, taarifa muhimu zimetawanyika katika katalogi za metadata, wikis za ndani, maelezo ya kodi (code comments), na seli za Jupyter notebook. Wakati AI agent inapojaribu kutekeleza kazi—kama vile kuandika SQL query sahihi kwa seti fulani ya data—inapaswa kupambana kuunganisha vipande hivi tofauti.

Google Cloud inabainisha kuwa hali ya sasa ni mchanganyiko wa suluhisho za kipekee na zilizojitenga. Watengenezaji kwa sasa wanatengeneza suluhisho za muktadha za kipekee, kuanzia Obsidian Vaults hadi faili za kanuni kama AGENTS.md au CLAUDE.md. Ingawa mifumo hii ya "metadata kama kodi" (metadata as code) inafanya kazi vizuri, inakosa uwezo wa kutumika na mifumo tofauti (interoperability). Maarifa yanabaki yaliyofungwa ndani ya mfumo au ghala (repository) maalum lililoyatengeneza, jambo linalozuia mtiririko wa taarifa usio na kikwazo kati ya zana na mifumo (frameworks) tofauti.

Maelezo ya OKF: Rahisi na Yanayoweza Kutumika na Mifumo Tofauti

Open Knowledge Format (v0.1) inachukua dhana ya "LLM wiki" na kuifanya kuwa kiwango cha ulimwengu. Katika msingi wake, kifurushi cha OKF ni directory ya faili za Markdown zinazotumia YAML frontmatter. Maelezo haya yameundwa kuwa rahisi kwa makusudi ili kuhimiza matumizi; uwanja (field) pekee wa lazima ni "type," ingawa watengenezaji wanaweza kujumuisha uwanja wa hiari kama vile kichwa cha habari (title), maelezo (description), rasilimali (resource), lebo (tags), na muda (timestamps).

Kwa sababu inategemea Markdown ya kawaida, grafu ya maarifa (knowledge graph) inaunda kupitia viungo vya Markdown vya kawaida, ikounganisha dhana kwa asili. Muundo huu unahakikisha kuwa OKF ina uwezo mkubwa wa kubebeka: kifurushi cha OKF kinaweza kusomwa katika programu yoyote ya kuhariri maandishi (text editor), kuonyeshwa moja kwa moja kwenye GitHub, na kuorodheshwa na zana yoyote ya utafutaji iliyopo. Muhimu zaidi, maelezo haya yanatenganisha watengenezaji na watumiaji, ikimaanisha kuwa hati iliyoandikwa na binadamu inaweza kuchakatwa na AI agent, na kifurushi kilichotengenezwa na mashine kinaweza kuonekana kwa urahisi na binadamu.

Ujumuishaji wa Mifumo na Mifano ya Marejeleo

To ensure the spec moves beyond theory, Google Cloud is providing several reference implementations and tools. This includes an enrichment agent capable of crawling BigQuery datasets to automatically generate OKF documents for every table. Google has also released a static HTML visualizer and provided sample bundles for complex datasets, including GA4 e-commerce, Stack Overflow, and Bitcoin data.

Crucially, Google Cloud has updated its own Knowledge Catalog to ingest OKF, allowing the format to be served directly to AI agents. By making the spec and code available on GitHub, Google is positioning OKF as a foundational layer for the next generation of agentic workflows, where knowledge is treated as a standardized, portable asset rather than a locked data silo.

Key Takeaways

• Standardized Context: OKF converts scattered documentation into a unified directory of Markdown files, allowing AI agents to navigate complex knowledge graphs using standard links.
• Minimalist Design: By requiring only a single "type" field in YAML frontmatter, the specification ensures high portability and low friction for developers.
• Interoperability: The format bridges the gap between human-readable wikis and machine-readable metadata, working across any cloud provider, database, or agent framework.