GenAI PMCognition
The company behind Devin, referenced for providing credits to Ryan Carson. It is mentioned in the context of scaling use of autonomous coding agents.
Key Highlights
- Cognition is best known as the company behind Devin, a leading autonomous software engineering agent platform.
- Its recent launches span Outposts, Slack Code, stacked PR support, and sandboxed deployment infrastructure.
- Cognition’s updates are useful to AI PMs studying enterprise readiness, agent orchestration, and cost-performance tradeoffs.
- Real-world usage examples, including Ryan Carson’s parallel Devin workflows, show how agentic development is being operationalized.
- The company also participates in adjacent conversations around benchmarks, trustworthiness, and secure AI deployment.
Cognition
Overview
Cognition is the company behind Devin, an autonomous software engineering agent platform that has become a recurring reference point in discussions about scaling AI-driven development work. In the newsletter corpus, Cognition appears primarily through product launches, benchmark updates, infrastructure announcements, and real-world usage stories—especially around how founders and teams are operationalizing multiple parallel coding agents. It is also referenced as the provider of substantial usage credits to Ryan Carson, illustrating both the cost profile and the emerging vendor-led adoption playbook for agentic software development.For AI Product Managers, Cognition matters because it represents a leading edge of the shift from “copilot” workflows to managed, cloud-based, autonomous engineering workflows. Its launches around Outposts, Slack integration, stacked PRs, isolated sandboxes, and evaluation benchmarks show how the category is moving beyond raw code generation into orchestration, governance, collaboration, deployment, and cost/performance optimization. Cognition is therefore useful both as a product case study and as a signal of where enterprise-ready software agents are heading.
Key Developments
- 2026-07-22 — Cognition launched Devin Outposts, enabling Devin to run on any machine, including Mac minis, private-network VMs, GPU boxes, and Kubernetes clusters near internal services.
- 2026-07-24 — Cognition introduced DeepWiki, a free repo-indexing tool covering more than 500,000 repositories to help developers understand unfamiliar codebases and libraries.
- 2026-07-25 — In a Ryan Carson workflow example, Cognition’s cloud VMs were used to run five to ten parallel agent sessions, contributing to reported output of 22–40 PRs per day.
- 2026-07-28 — Cognition was cited in security-related coverage connected to the Open Secure AI Alliance, reinforcing its presence in conversations about evaluation and secure agent deployment.
- 2026-07-31 — Cognition added native support for GitHub Stacked PRs in Devin, allowing large changes to be split into smaller reviewable diffs with automated rebasing and comment resolution across the stack.
- 2026-08-01 — Cognition formally launched Outposts as a Devin capability for running and testing apps on any computer.
- 2026-08-05 — Cognition announced that improvements to the harness and models made Devin Fusion 4% more intelligent and 27% less expensive on FrontierCode 1.1.
- 2026-08-06 — Cognition shared that Devin Outposts can run on Vercel Sandbox, using isolated microVMs with Docker support, VPN-based private-network access, and filesystem snapshots.
- 2026-08-21 — Cognition announced Slack Code in Devin, enabling Devin to proactively create dedicated code channels and improving how agent work is organized in Slack.
- 2026-08-25 — Ryan Carson said he had moved most engineering work to cloud-based Devin agents, reporting prior spending of roughly $5,000/month, a peak of $20,000 in one month, and then receiving $20,000/month in Cognition credits.
Relevance to AI PMs
1. Designing multi-agent product workflows Cognition shows how AI coding agents are evolving from single-task assistants into parallel work systems. AI PMs can study workflows like bug triage, priority queues, PR landing, and account monitoring to design agent-native product operations rather than just adding “AI features.”2. Evaluating enterprise readiness
Features like Outposts, Vercel Sandbox support, VPN access, snapshots, and Slack/GitHub integrations highlight the operational requirements needed for enterprise deployment. AI PMs can use Cognition as a benchmark for what secure, auditable, infrastructure-aware agent products should support.
3. Managing unit economics and performance tradeoffs
Cognition’s updates around FrontierCode, Devin Fusion cost reductions, and real-world monthly spend figures are directly relevant for PMs building business cases for agent adoption. The lesson is that agent products must be judged on throughput, review burden, cost per completed task, and infrastructure fit—not just model quality.
Related
- Devin — Cognition’s flagship autonomous coding agent and the main reason the company is repeatedly referenced.
- Devin Outposts — A deployment model that extends Devin onto local, private, or enterprise-controlled infrastructure.
- DeepWiki — A Cognition-built repo exploration tool that complements coding-agent workflows by helping users understand unfamiliar codebases.
- FrontierCode / SWE-Check / trustworthiness-evaluation — Related evaluation and benchmarking efforts that connect Cognition to measurable agent performance and reliability.
- Slack, GitHub, Vercel Sandbox — Key ecosystem integrations showing how Cognition is embedding agents into real developer workflows.
- Ryan Carson — A prominent practitioner example cited in connection with heavy Devin usage and Cognition credits.
- Open Secure AI Alliance — Relevant to Cognition’s proximity to secure deployment and evaluation discussions.
- Devon / software-agents / ai-agents — Broader category neighbors that place Cognition within the competitive and conceptual landscape of autonomous engineering tools.
Newsletter Mentions (55)
“Ryan Carson moved most engineering work to cloud-based Devin agents; he previously spent about $5,000 per month on Devin and reached $20,000 in one month before receiving $20,000 per month in Cognition credits.”
▶️ How I manage 15 AI agents 24/7 as a solo founder | Ryan Carson How I AI Podcast Ryan Carson manages roughly 10–15 concurrent Devin threads by sorting them into Bugs and P0/P1/P2 folders, using a handwritten weekly-priorities list to keep P0 business work visible, and using Devin playbooks for account monitoring and PR landing. Ryan Carson moved most engineering work to cloud-based Devin agents; he previously spent about $5,000 per month on Devin and reached $20,000 in one month before receiving $20,000 per month in Cognition credits.
“Cognition announced Slack Code in Devin, which enables Devin to proactively create dedicated code channels to keep work focused.”
#5 𝕏 Cognition announced Slack Code in Devin, which enables Devin to proactively create dedicated code channels to keep work focused. Cognition also shared details on tuning Devin’s personality in Slack.
“Cognition shared that Devin Outposts can run on Vercel Sandbox to build and test apps in an isolated microVM, with support for Docker, private-network access via VPN, and filesystem snapshots that preserve repositories, dependencies, and build state.”
#4 𝕏 Cognition shared that Devin Outposts can run on Vercel Sandbox to build and test apps in an isolated microVM, with support for Docker, private-network access via VPN, and filesystem snapshots that preserve repositories, dependencies, and build state. #5 in Guillermo Rauch shared that one line of code in AI SDK saves 90% or more in DeepSeek v4 Flash AI Gateway tokens.
“Cognition announced that improvements to the harness and models made Devin Fusion 4% more intelligent and 27% less expensive on FrontierCode 1.1.”
#7 𝕏 Cognition announced that improvements to the harness and models made Devin Fusion 4% more intelligent and 27% less expensive on FrontierCode 1.1.
“Cognition launched Outposts, a new Devin feature that lets you natively run and test apps on any computer.”
#8 𝕏 Cognition launched Outposts, a new Devin feature that lets you natively run and test apps on any computer.
“Cognition : Devin now natively supports GitHub Stacked PRs, breaking large changes into smaller, reviewable diffs, addressing and fixing comments across stacks, and automatically rebasing downstream changes.”
#16 𝕏 Cognition : Devin now natively supports GitHub Stacked PRs, breaking large changes into smaller, reviewable diffs, addressing and fixing comments across stacks, and automatically rebasing downstream changes. #17 𝕏 Cognition updated FrontierCode 1.1 with new discounts for GPT-5.6 Terra and Luna, positioning the GPT-5.6 series on the Pareto curve for price/performance efficiency.
“Also covered by: @Cognition #6 𝕏 NVIDIA AI contributed open models, weights, data and research to the Open Secure AI Alliance, including its NVIDIA Labs Object-Oriented Agents (NOOA) open agent harness.”
GenAI PM Daily July 28, 2026. Cognition is cited as a related source on security and alliance coverage and later as a launcher of an evaluation.
“#5 ▶️ Most Valuable Skill of 2026: Managing AI Agents Greg Isenberg Ryan Carson uses Cognition’s Devon cloud VMs to run five to ten parallel AI agent sessions, shipping 22–40 PRs per day (about 50% from his iPhone), and automates QA with a $60 thrice-weekly end-to-end browser-based signup test.”
#5 ▶️ Most Valuable Skill of 2026: Managing AI Agents Greg Isenberg Ryan Carson uses Cognition’s Devon cloud VMs to run five to ten parallel AI agent sessions, shipping 22–40 PRs per day (about 50% from his iPhone), and automates QA with a $60 thrice-weekly end-to-end browser-based signup test. Runs five to ten Cognition Devon cloud VM sessions in parallel, reporting a 50× jump in output and shipping 22–40 PRs per day, with roughly 50% of tasks completed via his iPhone.
“Cognition built DeepWiki, a free tool indexing over 500,000 repos to help you explore unfamiliar libraries, and will reveal its under-the-hood design in an upcoming LangChain + @jacobtpl meetup.”
#13 𝕏 Cognition built DeepWiki, a free tool indexing over 500,000 repos to help you explore unfamiliar libraries, and will reveal its under-the-hood design in an upcoming LangChain + @jacobtpl meetup. #14 𝕏 Guillermo Rauch announces that AI Gateway now supports real-time streaming transcription via the new streamTranscribe endpoint in the AI SDK, enabling developers to build live voice agents with unprecedented speed.
“Cognition launched Devin Outposts, so you can deploy Devin on any machine—whether it’s your Mac mini, a GPU box in your lab, a private-network VM, or a Kubernetes cluster alongside your internal services.”
Cognition launched Devin Outposts, so you can deploy Devin on any machine—whether it’s your Mac mini, a GPU box in your lab, a private-network VM, or a Kubernetes cluster alongside your internal services.
Related
A creator/curator in the AI PM space who shared the ai-evals-course repository. He is mentioned as a source for practical AI eval resources.
A developer platform company mentioned as the home of Vercel AI Gateway and the company of Guillermo Rauch. It is discussed in relation to AI gateway growth and model pricing.
A framework for building LLM applications and agents. In this newsletter it appears in the story about the founders’ attempt to automate dropshipping.
An autonomous coding agent used by a solo founder to manage engineering work and PR workflows. The newsletter highlights extensive threaded usage, playbooks, and review automation around it.
A workplace messaging platform used here as an operational surface for AI agents. PMs may care because agent integrations increasingly extend into team communication workflows.
Autonomous or semi-autonomous AI systems that use tools, manage context, and complete tasks on behalf of users. The newsletter discusses common blockers such as tool quality, context overload, and system verification.
Linear is a product and issue-tracking company whose team shared practical guidance for building production agents.
A frontier model release referenced as improving price-performance for developers. It is discussed as being available in Kiro for more cost-effective application development.
A Claude model variant being updated with stronger biology safeguards to reduce false positives while still routing dual-use biology requests to higher-safety fallback behavior. Relevant for PMs considering safety tradeoffs and product-surface-specific policy tuning.
A software development platform used here as the source and sync target for repositories. It is central to AI coding workflows, plugin distribution, and agent automation.
A product or version referenced in comparison with GStack. The newsletter uses it as a marker for a workflow change in Claude Code usage.
An unspecified system or capability referenced by Boris Cherny as being used unchanged by his group. The newsletter provides little detail beyond its use in cybersecurity refusal work.
A reimagined code review interface from Cognition that groups related changes and flags issues by confidence and severity. Useful as an example of AI-native developer workflow design.
A systems programming language mentioned in the context of a Rust-based Bun port embedded in Claude Code. It is part of an implementation-level investigation.
A model released on Windsurf with a limited-time launch discount. It is relevant as another model option available to developers.
The process of updating legacy COBOL systems, often for enterprise migration and maintenance. AI agents are increasingly positioned as tools to accelerate this high-friction modernization work.
A legacy programming language often targeted for modernization and migration efforts. For PMs, it represents enterprise technical debt and transformation risk.
A product access offering mentioned in the context of pricing tiers and credits. It appears to be part of a broader AI product subscription structure.
Stay updated on Cognition
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free