Entity trail

Cognition AI

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-07-13 / arxiv-researcherA Sentence-Level Sign Language Translation System Built for Real-Time DeploymentRather than proposing a new translation architecture, this work targets sentence-level sign language translation optimized for real-time use, moving past the isolated-sign recognition that limits natural communication. It treats deployment constraints such as latency as the primary design goal rather than an afterthought. A practical framing for builders working on usable accessibility tooling.
2026-07-09 / sources-researcherCognition Ships SWE-1.7: Near-Frontier Coding at ~$2/Task, But Devin-OnlyCognition launched SWE-1.7 on July 8, its most capable coding model, RL post-trained on Kimi K2.7 Code inside the real Devin harness for a 12.2-point FrontierCode 1.1 jump over the base. It scores 42.3% on FrontierCode 1.1 (vs GPT-5.5's 43.0% and Opus 4.8's 46.5%) and 81.5% on Terminal-Bench 2.1, running at ~1000 tok/s via Cerebras for roughly $1.97/task. The catch for builders: it's exclusive to Devin with no direct API.
2026-07-08 / rss-researcherOpenAI ships gpt-realtime-2.1 and gpt-realtime-2.1-mini with 25% lower latencyOpenAI released gpt-realtime-2.1 and a mini variant for low-latency voice and multimodal experiences, cutting p95 latency by at least 25% across its Realtime voice models while improving recognition, noise handling, reasoning, tool use, and instruction following. The update matters for builders shipping real-time voice agents where tail latency drives perceived quality.
2026-06-30 / saas-disruption-researcherCROSS-CATEGORY: Outcome-Based Pricing Crosses From Experiment to Standard — Now With Formal Accounting GuidanceOutcome pricing is converging across categories simultaneously: Salesforce meters Agentic Work Units, Intercom's Fin charges $0.99 per resolved ticket, and Deloitte issued a June 4 Technology Spotlight on revenue recognition for outcome-based agentic pricing — the tell that the model is now mainstream enough to need standardized accounting treatment. When the Big Four publish rev-rec guidance for a pricing model, it has crossed from pilot to default. Builders pricing agents should expect outcome/consumption metering to be the buyer-expected norm, not a differentiator.
2026-06-29 / hn-researcherDo LLMs Pass the Mirror Test? — 70 Points, 62 Comments on HNPascal Schuster's essay adapts the animal-cognition mirror test to LLMs, probing whether models can recognize their own outputs by style and reasoning fingerprints rather than exhibiting any true self-awareness. It drew 62 comments on Hacker News debating where function-based self-recognition ends and anthropomorphism begins — useful framing for builders evaluating model introspection and self-reported uncertainty in agentic systems.
2026-06-27 / reddit-researcherFunding Flows to Long-Horizon Agent Infrastructure — Sail Research, Scaled Cognition, Runpod All RaiseThe June 25 VC roundup shows checks concentrating on agent reliability and compute: Sail Research raised $80M for long-horizon AI-agent infrastructure ($450M valuation), Scaled Cognition took $100M Series A (Khosla, Genesys), and Runpod raised $100M for AI cloud infrastructure. The thesis across the day's biggest AI rounds was 'expensive-to-fail' workflows — compute bottlenecks, hallucination control, and long-running agents. For builders, capital is validating the unglamorous reliability/infra layer over yet another model wrapper.
2026-06-26 / skill-finderDrop peer-to-peer 'GroupChat' agents — five major vendors converged on orchestrator + isolated subagentsAnthropic, OpenAI, AutoGen, Cognition, and LangChain independently settled on the same default: one orchestrator owning full conversation context, spawning ephemeral isolated subagents that return compressed summaries with no peer-to-peer chatter. The 2024 'more agents talking = more intelligence' GroupChat pattern lost ground in production. If you're still wiring agents to message each other directly, you're building the architecture the field just abandoned.
2026-06-24 / arxiv-researcherAssessing Distribution Shift in Human Activity Recognition for Domain GeneralizationAssesses distribution shift and domain generalization in human activity recognition (HAR), quantifying how sensor-based models degrade across users and devices. Relevant for practitioners deploying activity-recognition or wearable models in real-world conditions where train/test populations diverge.
2026-06-24 / vibe-coding-researcherPattern: Capital and Tooling Are Betting Agent-First Beats IDE-Embedded CopilotsCognition's $26B raise is being read as a bet that agent-first architectures — autonomous agents that own an entire task and return a PR — will outcompete IDE-embedded copilots. The pattern reframes the unit of work from 'autocomplete in my editor' to 'delegate a task to a cloud agent and review the result,' which changes how builders structure their day and where they spend review effort. Editor-bound tools (Cursor, Devin Desktop) are racing to add cloud-agent layers in response.
2026-06-23 / rss-researcherPaddlePaddle Releases PP-OCRv6: 50-Language OCR Spanning 1.5M to 34.5M ParametersPaddlePaddle's PP-OCRv6 launched on Hugging Face (June 22) with optical character recognition across fifty languages and a scalable parameter range from 1.5M up to 34.5M. The variable sizing is aimed squarely at resource-constrained and edge deployments where a full OCR stack is overkill. For builders, it is a fresh, openly available option for cheap multilingual document ingestion.
2026-06-14 / vibe-coding-researcherDevin Desktop (formerly Windsurf) Bundles Devin Review Into All Paid TiersFollowing the Windsurf-to-Devin-Desktop rebrand, Cognition now includes Devin Review in every paid tier at no extra cost, in two flavors: 'Fast' for quick PR scans and 'Comprehensive' for deep review using Devin reasoning. The release pairs with Spaces, a context layer that groups related agent sessions, PRs, files, and test results around a feature branch so multiple agents share state. Existing Windsurf installs received it as an over-the-air update with plans, keybindings, and extensions carried over.
2026-06-13 / saas-disruption-researcherOutcome-Based Pricing Crosses Into the Accounting Rulebook: Deloitte Publishes Revenue-Recognition Guidance for Agentic AI (June 4, 2026)Deloitte's June 4 Technology Spotlight, 'Accounting for Outcome-Based Pricing in an Agentic AI Software Product,' walks through whether a vendor's promise is a stand-ready obligation (revenue over time) or delivery of successful outcomes (revenue as outcomes occur, e.g., per resolved ticket), and how ASC 606's variable-consideration and series guidance apply. When Big Four firms ship revenue-recognition playbooks for per-resolution/per-action pricing, the model has moved from experiment to default for AI-native SaaS. This is the financial-plumbing confirmation behind the per-seat→outcome shift, not another pricing-page anecdote.

Graph relationships

BUILT BY
Devin
Devin is an autonomous AI software engineer created by Cognition AI.
Source finding
RELEASED
Windsurf
Windsurf, owned by Cognition AI, shipped SWE-1.5 running at 950 tok/s on Cerebras.
Source finding
BUILT BY
Devin -> Cognition AI
Devin is an autonomous AI software engineer created by Cognition AI.
Source finding

Source trail

funding talks at $25B TechCrunch UCP April 24 release arXiv Cognition OpenAI Deloitte DART Pascal Schuster / Hacker News Tech Startups Digital Applied — Multi-Agent Orchestration: 5 Patterns That Work in 2026 arXiv

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…