Financial AI adoption is constrained not by model quality but by cognition friction: users repeatedly restate fragmented context, historical judgments, and risk preferences. Existing financial agents remain turn-based and workflow-disposable.</p>\n<p>We propose InKH, a financial-agent architecture combining: an event-stream view of user/market/tool updates; a bounded working context buffer assembled by passive injection rather than agent-driven search; a temporal knowledge graph for low-latency retrieval; a wiki audit surface; and background extraction with maturity, decay, and write-time invalidation.</p>\n<p>We report a reproducible benchmark with 24 seeds, 4 rounds, 80 episodes per round, and 6 baselines. InKH achieves 0.815 task quality at 900ms mean latency. Relative to agent-driven wiki-walk memory: 82.95% lower latency, 82.29% lower token cost, 96.58% less stale-knowledge usage, +0.108 task quality, +0.461 traceability.</p>\n","updatedAt":"2026-06-05T12:26:19.189Z","author":{"_id":"6a1f0c797e841a290a08be86","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/GO89MXvjQJrJu-D6JOELJ.jpeg","fullname":"Sofia","name":"pidturkina","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8753438591957092},"editors":["pidturkina"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/GO89MXvjQJrJu-D6JOELJ.jpeg"],"reactions":[],"isReport":false}},{"id":"6a237b62ed8b6eeac4e0f455","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":362,"isUserFollowing":false},"createdAt":"2026-06-06T01:44:02.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"This is an automated message from the [Librarian Bot](https://huggingface.co/librarian-bots). I found the following papers similar to this paper. \n\nThe following papers were recommended by the Semantic Scholar API \n\n* [Eywa: Provenance-Grounded Long-Term Memory for AI Agents](https://huggingface.co/papers/2605.30771) (2026)\n* [FundaPod: A Multi-Persona Agent Pod Platform with Knowledge Graph Memory for AI-Assisted Fundamental Investment Research](https://huggingface.co/papers/2605.27864) (2026)\n* [Representation Signatures and Risk-Feedback Alignment in LLM Trading Agents](https://huggingface.co/papers/2605.28850) (2026)\n* [Stateless Decision Memory for Enterprise AI Agents](https://huggingface.co/papers/2604.20158) (2026)\n* [SkillsVote: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution](https://huggingface.co/papers/2605.18401) (2026)\n* [Hierarchical Long-Term Semantic Memory for LinkedIn's Hiring Agent](https://huggingface.co/papers/2604.26197) (2026)\n* [Herculean: An Agentic Benchmark for Financial Intelligence](https://huggingface.co/papers/2605.14355) (2026)\n\n\n Please give a thumbs up to this comment if you found it helpful!\n\n If you want recommendations for any Paper on Hugging Face checkout [this](https://huggingface.co/spaces/librarian-bots/recommend_similar_papers) Space\n\n You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: `@librarian-bot recommend`","html":"<p>This is an automated message from the <a href=\"https://huggingface.co/librarian-bots\">Librarian Bot</a>. I found the following papers similar to this paper. </p>\n<p>The following papers were recommended by the Semantic Scholar API </p>\n<ul>\n<li><a href=\"https://huggingface.co/papers/2605.30771\">Eywa: Provenance-Grounded Long-Term Memory for AI Agents</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2605.27864\">FundaPod: A Multi-Persona Agent Pod Platform with Knowledge Graph Memory for AI-Assisted Fundamental Investment Research</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2605.28850\">Representation Signatures and Risk-Feedback Alignment in LLM Trading Agents</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2604.20158\">Stateless Decision Memory for Enterprise AI Agents</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2605.18401\">SkillsVote: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2604.26197\">Hierarchical Long-Term Semantic Memory for LinkedIn's Hiring Agent</a> (2026)</li>\n<li><a href=\"https://huggingface.co/papers/2605.14355\">Herculean: An Agentic Benchmark for Financial Intelligence</a> (2026)</li>\n</ul>\n<p> Please give a thumbs up to this comment if you found it helpful!</p>\n<p> If you want recommendations for any Paper on Hugging Face checkout <a href=\"https://huggingface.co/spaces/librarian-bots/recommend_similar_papers\">this</a> Space</p>\n<p> You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: <code><span class=\"SVELTE_PARTIAL_HYDRATER contents\" data-target=\"UserMention\" data-props=\"{"user":"librarian-bot"}\"><span class=\"inline-block\"><span class=\"contents\"><a href=\"/librarian-bot\">@<span class=\"underline\">librarian-bot</span></a></span> </span></span> recommend</code></p>\n","updatedAt":"2026-06-06T01:44:02.375Z","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":362,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7194701433181763},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.01886","authors":[{"_id":"6a1f149fe292c1c78ecb1173","name":"Ailiya Borjigin","hidden":false},{"_id":"6a1f149fe292c1c78ecb1174","name":"Igor Stadnyk","hidden":false},{"_id":"6a1f149fe292c1c78ecb1175","name":"Ben Bilski","hidden":false},{"_id":"6a1f149fe292c1c78ecb1176","name":"Maksym Chikita","hidden":false},{"_id":"6a1f149fe292c1c78ecb1177","name":"Dmytro Kyrylenko","hidden":false},{"_id":"6a1f149fe292c1c78ecb1178","user":{"_id":"6a1f0c797e841a290a08be86","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/GO89MXvjQJrJu-D6JOELJ.jpeg","isPro":false,"fullname":"Sofia","user":"pidturkina","type":"user","name":"pidturkina"},"name":"Sofiia Pidturkina","status":"claimed_verified","statusLastChangedAt":"2026-06-04T12:42:58.582Z","hidden":false},{"_id":"6a1f149fe292c1c78ecb1179","name":"Julia Stadnyk","hidden":false}],"publishedAt":"2026-06-01T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"Absorbing Complexity: An Interaction-Native Knowledge Harness for Financial LLM Agents","submittedOnDailyBy":{"_id":"6a1f0c797e841a290a08be86","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/GO89MXvjQJrJu-D6JOELJ.jpeg","isPro":false,"fullname":"Sofia","user":"pidturkina","type":"user","name":"pidturkina"},"summary":"Financial AI agents often fail for a simple reason: they make users carry the complexity. A user must repeatedly restate goals, risk preferences, portfolio context, past judgments, and shifting market assumptions, while the agent answers, retrieves, acts, and forgets. In finance, this is not just inconvenient. In tasks such as market analysis, copy-trading review, and trade preparation, forgotten context and stale memory can create latency, repeated errors, weak auditability, and unsafe decisions.\n We propose the interaction-native knowledge harness (InKH), an architecture for financial LLM agents that absorbs complexity into the system. InKH converts user, market, portfolio, and tool events into structured operational knowledge. It uses passive knowledge injection to assemble a bounded working context buffer before the main model step, temporal graph memory for low-latency retrieval, a wiki audit surface for human-readable governance, and background extraction with maturity, decay, and write-time invalidation.\n We evaluate InKH on a reproducible controlled synthetic benchmark with 24 random seeds, 4 rounds, 80 episodes per round, and 6 baselines, producing 46,080 baseline-conditioned evaluations. InKH achieves mean task quality of 0.815 at 900 ms latency. Compared with agent-driven wiki-walk memory, it reduces latency by 82.95 percent, token cost by 82.29 percent, and stale-knowledge usage by 96.58 percent, while improving quality by 0.108 and traceability by 0.461. Compared with a temporal-graph system without invalidation, it improves quality by 0.050 and reduces stale-memory usage by 96.58 percent with comparable serving cost.\n The results support a design thesis for financial AI: adoption happens when complexity is absorbed by the system rather than transferred to the user. The benchmark validates architecture-level behavior, not live trading performance.","upvotes":1,"discussionId":"6a1f149fe292c1c78ecb117f","ai_summary":"Financial AI agents struggle with user complexity, but a new architecture called InKH addresses this by embedding complexity into the system through structured knowledge management and temporal memory mechanisms.","ai_keywords":["financial LLM agents","knowledge harness","passive knowledge injection","temporal graph memory","wiki audit surface","background extraction","maturity","decay","write-time invalidation"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"6a1f0d0f6ffb1e2b7b3440e7","name":"inc4-net","fullname":"INC4","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/ypVEU9TVn5KW5jswT5udq.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6a1f0c797e841a290a08be86","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/GO89MXvjQJrJu-D6JOELJ.jpeg","isPro":false,"fullname":"Sofia","user":"pidturkina","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6a1f0d0f6ffb1e2b7b3440e7","name":"inc4-net","fullname":"INC4","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6a1f0c797e841a290a08be86/ypVEU9TVn5KW5jswT5udq.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.01886.md"}">
Absorbing Complexity: An Interaction-Native Knowledge Harness for Financial LLM Agents
Published on Jun 1
· Submitted by Sofia on Jun 5 Abstract
Financial AI agents struggle with user complexity, but a new architecture called InKH addresses this by embedding complexity into the system through structured knowledge management and temporal memory mechanisms.
Financial AI agents often fail for a simple reason: they make users carry the complexity. A user must repeatedly restate goals, risk preferences, portfolio context, past judgments, and shifting market assumptions, while the agent answers, retrieves, acts, and forgets. In finance, this is not just inconvenient. In tasks such as market analysis, copy-trading review, and trade preparation, forgotten context and stale memory can create latency, repeated errors, weak auditability, and unsafe decisions.
We propose the interaction-native knowledge harness (InKH), an architecture for financial LLM agents that absorbs complexity into the system. InKH converts user, market, portfolio, and tool events into structured operational knowledge. It uses passive knowledge injection to assemble a bounded working context buffer before the main model step, temporal graph memory for low-latency retrieval, a wiki audit surface for human-readable governance, and background extraction with maturity, decay, and write-time invalidation.
We evaluate InKH on a reproducible controlled synthetic benchmark with 24 random seeds, 4 rounds, 80 episodes per round, and 6 baselines, producing 46,080 baseline-conditioned evaluations. InKH achieves mean task quality of 0.815 at 900 ms latency. Compared with agent-driven wiki-walk memory, it reduces latency by 82.95 percent, token cost by 82.29 percent, and stale-knowledge usage by 96.58 percent, while improving quality by 0.108 and traceability by 0.461. Compared with a temporal-graph system without invalidation, it improves quality by 0.050 and reduces stale-memory usage by 96.58 percent with comparable serving cost.
The results support a design thesis for financial AI: adoption happens when complexity is absorbed by the system rather than transferred to the user. The benchmark validates architecture-level behavior, not live trading performance.
Community
Financial AI adoption is constrained not by model quality but by cognition friction: users repeatedly restate fragmented context, historical judgments, and risk preferences. Existing financial agents remain turn-based and workflow-disposable.
We propose InKH, a financial-agent architecture combining: an event-stream view of user/market/tool updates; a bounded working context buffer assembled by passive injection rather than agent-driven search; a temporal knowledge graph for low-latency retrieval; a wiki audit surface; and background extraction with maturity, decay, and write-time invalidation.
We report a reproducible benchmark with 24 seeds, 4 rounds, 80 episodes per round, and 6 baselines. InKH achieves 0.815 task quality at 900ms mean latency. Relative to agent-driven wiki-walk memory: 82.95% lower latency, 82.29% lower token cost, 96.58% less stale-knowledge usage, +0.108 task quality, +0.461 traceability.
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.01886 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.01886 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.01886 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.