# PostHog

PostHog's RSS Feed

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## 6 things I've learned in 6 years at PostHog

DevFeed: [6 things I've learned in 6 years at PostHog](<https://devfeed.tech/articles/6-things-i-ve-learned-in-6-years-at-posthog-88777.md>)

Original publisher: [Read original article](<https://posthog.com/blog/6-years-at-posthog>)

Author: Charles Cook

Published: 2026-10-16T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [blog](<https://devfeed.tech/tags/blog.md>), [blog-post](<https://devfeed.tech/tags/blog-post.md>), [brand](<https://devfeed.tech/tags/brand.md>), [culture](<https://devfeed.tech/tags/culture.md>), [curiosity](<https://devfeed.tech/tags/curiosity.md>), [data](<https://devfeed.tech/tags/data.md>), [data-warehouse](<https://devfeed.tech/tags/data-warehouse.md>), [desktop](<https://devfeed.tech/tags/desktop.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [error-tracking](<https://devfeed.tech/tags/error-tracking.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [hiring](<https://devfeed.tech/tags/hiring.md>), [manager](<https://devfeed.tech/tags/manager.md>), [share](<https://devfeed.tech/tags/share.md>)

### AI overview

The author's six lessons are that strong hiring makes management easier, company memes can communicate strategy, public writing improves with practice, and growth need not require abandoning a startup's core practices. They advise managers to set a high hiring bar, give capable people context and room to work, and reserve hands-on intervention for occasional issues. At PostHog, names and memes helped make priorities memorable, while humor had to be genuinely funny rather than merely cute. The author also argues for small engineering teams, regular user conversations, and retaining a distinct brand and feature-focused selling as the company moves upmarket. For merchandise, they recommend useful products, a dedicated owner and budget, and pricing that reflects quality; they cite Micromerch as a provider they recommend. For writing, they suggest using plain language, stating views directly with reasons, and making time for thoughtful responses. The text ends with a product description of PostHog's developer tools and data warehouse/CDP.

### Source excerpt

After 6 years at PostHog, I am delighted to share that I have learned exactly 6 things. 80% of being a good manager is good hiring The other 20% is...

## We built our own cloud agents runtime. Here's what we learned

DevFeed: [We built our own cloud agents runtime. Here's what we learned](<https://devfeed.tech/articles/we-built-our-own-cloud-agents-runtime-here-s-what-we-learned-80494.md>)

Original publisher: [Read original article](<https://posthog.com/blog/cloud-agents-runtime>)

Author: Cleo Lant

Published: 2026-10-08T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [long-running](<https://devfeed.tech/topics/long-running.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [virtual machines](<https://devfeed.tech/topics/virtual-machines.md>), [Langgraph](<https://devfeed.tech/topics/langgraph.md>), [Cloud](<https://devfeed.tech/topics/cloud.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [api](<https://devfeed.tech/tags/api.md>), [cloud-agents](<https://devfeed.tech/tags/cloud-agents.md>), [cloud-run](<https://devfeed.tech/tags/cloud-run.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [runtime](<https://devfeed.tech/tags/runtime.md>)

### AI overview

PostHog's cloud agent runs are managed by a durable Temporal workflow so a run can survive worker deploys and sandbox loss: it snapshots the working tree, saves the conversation log, and can rebuild a VM and resume. Slack follow-ups are queued in arrival order, with duplicate webhook messages ignored; supported mid-turn redirects can join the active turn. The runtime moved from gVisor containers to Modal VMs with a real Linux kernel and Docker, plus network allowlists enforced at the network layer. For custom images, repository setup commands run without tokens after a build-only secret is removed. PostHog also describes a warming pool and project-specific images. The excerpt says these coding settings are in closed alpha, planned for open beta, and gives brief feature comparisons with Cursor, Codex, Claude Code, and Vercel.

### Source excerpt

You've killed an agent mid-run by closing your laptop. Or watched three of them fight over your CPU while the fans took off (same). A cloud agent...

## WTF is context engineering? (with real examples)

DevFeed: [WTF is context engineering? (with real examples)](<https://devfeed.tech/articles/wtf-is-context-engineering-with-real-examples-80696.md>)

Original publisher: [Read original article](<https://posthog.com/blog/wtf-is-context-engineering>)

Author: Jina Yoon

Published: 2026-10-07T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [context window](<https://devfeed.tech/topics/context-window.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [knowledge-engineering](<https://devfeed.tech/topics/knowledge-engineering.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [2025](<https://devfeed.tech/tags/2025.md>), [2026](<https://devfeed.tech/tags/2026.md>), [accessible](<https://devfeed.tech/tags/accessible.md>), [activation](<https://devfeed.tech/tags/activation.md>), [addy-osmani](<https://devfeed.tech/tags/addy-osmani.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-harness](<https://devfeed.tech/tags/agent-harness.md>), [agent-orchestration](<https://devfeed.tech/tags/agent-orchestration.md>), [agents](<https://devfeed.tech/tags/agents.md>), [agents-md](<https://devfeed.tech/tags/agents-md.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [ai-assisted](<https://devfeed.tech/tags/ai-assisted.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [context](<https://devfeed.tech/tags/context.md>), [context-engineering](<https://devfeed.tech/tags/context-engineering.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [examples](<https://devfeed.tech/tags/examples.md>)

### AI overview

Context engineering designs what an agent receives on each turn--such as instructions, tools, retrieved data, examples, and memory--and when it receives it. The article distinguishes this broader system from prompt engineering, which organizes a single request, and describes RAG as one retrieval technique within it. PostHog's Wizard uses a separate "context-mill" pipeline to gather docs, 40+ example apps, and hand-written instructions, package them as versioned skills, and provide them to client agents through MCP. The article also describes maintaining agent context through feedback from runs, model benchmarks, scope rules, and CI evaluations. Its guidance is to focus on source quality and information design, keep context lean as models improve, and separate context from the agent harness so each can be updated independently.

### Source excerpt

Say you're deploying an AI assistant that processes online order returns. For it to work, it would need access to your store's purchase policy, item...

## To build or buy feature flags: Non-obvious things to know

DevFeed: [To build or buy feature flags: Non-obvious things to know](<https://devfeed.tech/articles/to-build-or-buy-feature-flags-non-obvious-things-to-know-80477.md>)

Original publisher: [Read original article](<https://posthog.com/blog/build-or-buy-feature-flags>)

Author: Ian Vanagas

Published: 2026-09-16T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [build vs buy components](<https://devfeed.tech/topics/build-vs-buy-components.md>), [internal developer portal](<https://devfeed.tech/topics/internal-developer-portal.md>), [Tool](<https://devfeed.tech/topics/tool.md>)

Tags: [access-controls](<https://devfeed.tech/tags/access-controls.md>), [adoption](<https://devfeed.tech/tags/adoption.md>), [agent-tools](<https://devfeed.tech/tags/agent-tools.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [arena](<https://devfeed.tech/tags/arena.md>), [audit](<https://devfeed.tech/tags/audit.md>), [basic](<https://devfeed.tech/tags/basic.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [bugs](<https://devfeed.tech/tags/bugs.md>), [building](<https://devfeed.tech/tags/building.md>), [cases](<https://devfeed.tech/tags/cases.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [feature-flags](<https://devfeed.tech/tags/feature-flags.md>), [has](<https://devfeed.tech/tags/has.md>), [teams](<https://devfeed.tech/tags/teams.md>), [tools](<https://devfeed.tech/tags/tools.md>)

### AI overview

A feature flag can start as a key-value lookup, but reliable targeting and evaluation may depend on user identity, analytics, changing cohorts, and warehouse data. Integrations and queries add complexity and can create performance or data-safety problems for connected services. Because flags sit on the critical path, teams that build them also take responsibility for production failures, future feature changes, language support, migrations, and maintenance. Agent use adds tools and access controls to support, plus risks from unauthorized edits and malformed calls; the excerpt cites audit logs and MCP performance visibility as ways PostHog addressed those issues. Building may fit when flags are a product differentiator, an in-house team can maintain them as a product, or there are strong reasons to distrust third-party vendors. The excerpt describes PostHog's own experience and recommendation, so its vendor-performance figures and claims are specific to that account.

### Source excerpt

AI agents are causing teams to consider whether they should build their own version of tools they pay for. Feature flags have long been a prime...

## 6-7 loops we use every day to make PostHog self-driving

DevFeed: [6-7 loops we use every day to make PostHog self-driving](<https://devfeed.tech/articles/6-7-loops-we-use-every-day-to-make-posthog-self-driving-80600.md>)

Original publisher: [Read original article](<https://posthog.com/blog/self-driving-loops>)

Author: Andy Maguire

Published: 2026-09-15T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [workflow automation](<https://devfeed.tech/topics/workflow-automation.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [alert](<https://devfeed.tech/tags/alert.md>), [alerts](<https://devfeed.tech/tags/alerts.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [apis](<https://devfeed.tech/tags/apis.md>), [bug](<https://devfeed.tech/tags/bug.md>), [bug-reports](<https://devfeed.tech/tags/bug-reports.md>), [build](<https://devfeed.tech/tags/build.md>), [code-review](<https://devfeed.tech/tags/code-review.md>), [codex](<https://devfeed.tech/tags/codex.md>), [compatibility](<https://devfeed.tech/tags/compatibility.md>), [content](<https://devfeed.tech/tags/content.md>), [context-layer](<https://devfeed.tech/tags/context-layer.md>), [data](<https://devfeed.tech/tags/data.md>), [data-warehouse](<https://devfeed.tech/tags/data-warehouse.md>), [loops](<https://devfeed.tech/tags/loops.md>), [self-driving](<https://devfeed.tech/tags/self-driving.md>)

### AI overview

PostHog describes six self-driving loops that turn product and team signals into investigated reports and, often, reviewed fixes. Scouts group and route structured MCP feedback, monitor Slack threads and GitHub issue specs, investigate metric alerts and runtime logs, and use Replay Vision to spot user friction recordings can reveal beyond telemetry. Ready findings can start tasks and PRs; humans still write specs, steer investigations, review diagnoses and code, and decide what merges. Scouts can deduplicate reports, append evidence across runs, and recheck whether fixes hold. In one reported MCP case, a missing tool was reported and a PR opened within minutes, with the author reviewing before merge. Autonomy varies: one error-rate spike required substantial human root-cause guidance, and another follow-up PR remained open. The author reports that about one in ten scout runs included self-validation during the prior month.

### Source excerpt

I've mostly given up keeping up with agent trends. Every few months, I ignore all of it and ask what I'm actually getting use out of. Three things...

## Multimodal models need video. We froze time to give it to them

DevFeed: [Multimodal models need video. We froze time to give it to them](<https://devfeed.tech/articles/multimodal-models-need-video-we-froze-time-to-give-it-to-them-80568.md>)

Original publisher: [Read original article](<https://posthog.com/blog/multimodal-models-need-video>)

Author: Tue Haulund

Published: 2026-09-09T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [FFmpeg (Fast Forward Moving Picture Experts Group)](<https://devfeed.tech/topics/ffmpeg.md>), [Tool](<https://devfeed.tech/topics/tool.md>), [Event-Streaming](<https://devfeed.tech/topics/event-streaming.md>)

Tags: [it](<https://devfeed.tech/tags/it.md>), [models](<https://devfeed.tech/tags/models.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [multimodal-models](<https://devfeed.tech/tags/multimodal-models.md>), [time](<https://devfeed.tech/tags/time.md>), [video](<https://devfeed.tech/tags/video.md>)

### AI overview

Session Replay recordings are stored as rrweb DOM snapshots and change events, not as video, so multimodal models cannot directly understand them. PostHog built a rasterizer that plays each stream in headless Chromium, captures frames through Chrome's DevTools Protocol, and encodes them as MP4. Its first version produced freezes or skipped sections under CPU contention because rendering and capture advanced independently. The fix combines Chrome's `HeadlessExperimental.beginFrame`, which makes rendering wait for each capture, with JavaScript virtual-time shims that advance the rrweb player by exactly one frame per tick. Contention can then slow a job without changing its output. The article says this system has run in production since March 2026; it covers rasterization, while choosing video content for the model is left for a later post.

### Source excerpt

Almost none of the session recordings we ingest are watched . The majority are stored, never opened, and expire without receiving any attention. It's...

## Mind the context gap

DevFeed: [Mind the context gap](<https://devfeed.tech/articles/mind-the-context-gap-80563.md>)

Original publisher: [Read original article](<https://posthog.com/blog/mind-the-context-gap>)

Author: Lizzie Epton

Published: 2026-08-24T00:00:00Z

Content type: opinion

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [context](<https://devfeed.tech/topics/context.md>), [etl](<https://devfeed.tech/topics/etl.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [apps](<https://devfeed.tech/tags/apps.md>), [better](<https://devfeed.tech/tags/better.md>), [build](<https://devfeed.tech/tags/build.md>), [business](<https://devfeed.tech/tags/business.md>), [business-context](<https://devfeed.tech/tags/business-context.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [central](<https://devfeed.tech/tags/central.md>), [challenge](<https://devfeed.tech/tags/challenge.md>), [complex](<https://devfeed.tech/tags/complex.md>), [connect](<https://devfeed.tech/tags/connect.md>), [context](<https://devfeed.tech/tags/context.md>), [context-layer](<https://devfeed.tech/tags/context-layer.md>), [cost](<https://devfeed.tech/tags/cost.md>), [data](<https://devfeed.tech/tags/data.md>), [gap](<https://devfeed.tech/tags/gap.md>), [product](<https://devfeed.tech/tags/product.md>)

### AI overview

The author argues that AI product work stalls when agents lack trustworthy business context, while teams face fragmented data, fragile pipelines, and specialized tools. The article cites 56% of teams naming data quality as a frequent challenge, two-thirds of data professionals not fully trusting their AI data, and 42% reporting that more than half their AI projects have stalled. It defines data readiness as all data being in one place and reachable by people and agents, and proposes a "context warehouse" that combines ingestion, modeling, storage, and querying. In PostHog's implementation, product data is captured directly, external sources sync into the same warehouse, and product tools and agents query shared data; the Semantic Layer provides fixed definitions. These are the author's claims and product description, not comparative evidence that this approach resolves the cited problems.

### Source excerpt

Part of my job as a product marketer at PostHog is keeping tabs on what every other company in the data space is talking about. So I read a lot of...

## Building a semantic layer: What it is and how we did it at PostHog

DevFeed: [Building a semantic layer: What it is and how we did it at PostHog](<https://devfeed.tech/articles/building-a-semantic-layer-what-it-is-and-how-we-did-it-at-posthog-80602.md>)

Original publisher: [Read original article](<https://posthog.com/blog/semantic-layer>)

Author: Thiago Rocha Salvatore

Published: 2026-08-19T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [semantic-layer](<https://devfeed.tech/topics/semantic-layer.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [activation](<https://devfeed.tech/tags/activation.md>), [added](<https://devfeed.tech/tags/added.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [amplify](<https://devfeed.tech/tags/amplify.md>), [building](<https://devfeed.tech/tags/building.md>), [claude](<https://devfeed.tech/tags/claude.md>), [cursor](<https://devfeed.tech/tags/cursor.md>), [layer](<https://devfeed.tech/tags/layer.md>), [queries](<https://devfeed.tech/tags/queries.md>), [question](<https://devfeed.tech/tags/question.md>), [semantic](<https://devfeed.tech/tags/semantic.md>), [semantic-layer](<https://devfeed.tech/tags/semantic-layer.md>), [you-ll](<https://devfeed.tech/tags/you-ll.md>)

### AI overview

PostHog's semantic layer gives people and agents a governed catalog of metric definitions, trusted tables, and data joins on top of existing warehouse data. Its catalog is exposed as ordinary SQL tables, so agents can discover definitions through SQL rather than a separate catalog API. Agent-created entries start as proposed; humans approve them. Approved metrics return to proposed if edited, while metrics based on changed insights are flagged as drifted. Only approved, non-drifted definitions are canonical. Metrics can use SQL or snapshot an insight's query so funnels and trends run through the same engine as their dashboards; PostHog checks for changes to the source insight when a metric is read. SQL metrics use saved views. The article says the feature is generally available through MCP tools, the UI, and SQL, and describes an 80% target on fixed golden questions as one measure of success.

### Source excerpt

Ask Claude, Cursor, and PostHog AI the same question: "what was our MRR last month?", and you'll get three different queries and three different...

## The most common questions about developer marketing, answered

DevFeed: [The most common questions about developer marketing, answered](<https://devfeed.tech/articles/the-most-common-questions-about-developer-marketing-answered-80501.md>)

Original publisher: [Read original article](<https://posthog.com/blog/developer-marketing-questions>)

Author: Joe Black

Published: 2026-08-14T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [developer-relations](<https://devfeed.tech/topics/developer-relations.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>)

Tags: [advice](<https://devfeed.tech/tags/advice.md>), [announcement](<https://devfeed.tech/tags/announcement.md>), [approach](<https://devfeed.tech/tags/approach.md>), [automation](<https://devfeed.tech/tags/automation.md>), [awareness](<https://devfeed.tech/tags/awareness.md>), [channel](<https://devfeed.tech/tags/channel.md>), [community](<https://devfeed.tech/tags/community.md>), [content](<https://devfeed.tech/tags/content.md>), [culture](<https://devfeed.tech/tags/culture.md>), [data](<https://devfeed.tech/tags/data.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developer-marketing](<https://devfeed.tech/tags/developer-marketing.md>), [marketing](<https://devfeed.tech/tags/marketing.md>), [questions](<https://devfeed.tech/tags/questions.md>)

### AI overview

The author's main advice is to treat marketing as ongoing work shaped by a team's strengths and culture, rather than follow a universal formula. Extend launches across channels and into pre- and post-launch work; choose channels the team can use well and authentically. To help users discover more of a product, invest in onboarding, targeted in-app cross-selling, and regular shipped-feature updates, while reserving dedicated emails for changes that materially affect users. For betas, PostHog uses a public roadmap, waitlists, and automations for access and feedback, but avoids paid promotion because beta products and messaging can change substantially. The author tracks sign-ups but also judges efforts by user response and whether the team would repeat them, warning that rigid ROI metrics can create poor incentives. A failed $10,000 event with six attendees taught them to define whether a community is more than a support forum and to use event expertise. They see billboards as awareness-building, assessed by whether people notice and discuss them, and merch as a source of joy rather than profit. The advice is presented as personal experience, not a general marketing theory.

### Source excerpt

One of the baffling things about leading the marketing team at PostHog is how often I get asked for advice. People seem to look at PostHog and say...

## I bribe customers to talk to me

DevFeed: [I bribe customers to talk to me](<https://devfeed.tech/articles/i-bribe-customers-to-talk-to-me-80476.md>)

Original publisher: [Read original article](<https://posthog.com/blog/bribing-customers-to-talk-to-me>)

Author: Jake Sciotto

Published: 2026-08-13T00:00:00Z

Content type: opinion

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [X (Twitter)](<https://devfeed.tech/topics/twitter.md>), [Support](<https://devfeed.tech/topics/support.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [attention](<https://devfeed.tech/tags/attention.md>), [book](<https://devfeed.tech/tags/book.md>), [bribe](<https://devfeed.tech/tags/bribe.md>), [calendar](<https://devfeed.tech/tags/calendar.md>), [career](<https://devfeed.tech/tags/career.md>), [carrier-pigeon](<https://devfeed.tech/tags/carrier-pigeon.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [code](<https://devfeed.tech/tags/code.md>), [communication](<https://devfeed.tech/tags/communication.md>), [company](<https://devfeed.tech/tags/company.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-success](<https://devfeed.tech/tags/customer-success.md>), [customers](<https://devfeed.tech/tags/customers.md>), [i](<https://devfeed.tech/tags/i.md>), [me](<https://devfeed.tech/tags/me.md>), [talk](<https://devfeed.tech/tags/talk.md>), [time](<https://devfeed.tech/tags/time.md>)

### AI overview

The author says playful, short, personalized outreach helps PostHog customer success managers get steady-state customers to respond. Examples include offering merch for a 30-minute conversation, sharing a dashboard built for a customer, and noting a user's unusually high MCP activity to invite feedback. These messages led to quick calls, product feedback, and more users joining Slack; one follow-up clarified a confusing leaderboard message. The author believes the humor and departure from corporate-sounding outreach matter more than the merch itself. For a company that continued to ignore messages, the author tried sending a care package with a potato bearing their face and a prompt to check email.

### Source excerpt

Customer success managers at PostHog spend a lot of time getting customers to talk to them. This is weird because in my past post-sales roles, I sat...

## Are agents really killing UI?

DevFeed: [Are agents really killing UI?](<https://devfeed.tech/articles/are-agents-really-killing-ui-80457.md>)

Original publisher: [Read original article](<https://posthog.com/blog/agents-killing-ui>)

Author: Cleo Lant

Published: 2026-08-06T00:00:00Z

Content type: opinion

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI Bots](<https://devfeed.tech/topics/ai-bots.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>)

Tags: [aeo](<https://devfeed.tech/tags/aeo.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-experience](<https://devfeed.tech/tags/agent-experience.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-chat](<https://devfeed.tech/tags/ai-chat.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [app](<https://devfeed.tech/tags/app.md>), [auth](<https://devfeed.tech/tags/auth.md>), [automated](<https://devfeed.tech/tags/automated.md>), [ax](<https://devfeed.tech/tags/ax.md>), [figma](<https://devfeed.tech/tags/figma.md>), [software](<https://devfeed.tech/tags/software.md>), [ui](<https://devfeed.tech/tags/ui.md>), [users](<https://devfeed.tech/tags/users.md>)

### AI overview

The article argues that agents add a headless access layer to software rather than making human interfaces obsolete. It recommends agent-completable onboarding, discoverable and parseable docs, and safe authentication, alongside human screens that show setup progress and next steps. MCPs let agents act through product tools; product teams should instrument agent calls and make every action inspectable by a person. Examples include PostHog's terminal setup and MCP analytics, Vercel's deployment MCP with dashboard logs and rollback, and AgentMail's human email verification. The article says agent-focused screens should support approval, diff review, undo, attribution, orchestration, and interruption. It presents UI as especially valuable for rapid state recognition and human oversight, while acknowledging that some agents can operate existing interfaces and that MCP reach alone does not guarantee a good experience.

### Source excerpt

The UI is dead. Or so I keep hearing: "Agents are your users now, software is losing its head , and everyone who learned Figma should start learning...

## We spend $45,000 on doing more weird every month

DevFeed: [We spend $45,000 on doing more weird every month](<https://devfeed.tech/articles/we-spend-45-000-on-doing-more-weird-every-month-80571.md>)

Original publisher: [Read original article](<https://posthog.com/blog/on-doing-more-weird>)

Author: Charles Cook

Published: 2026-08-05T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [intrinsic motivation](<https://devfeed.tech/topics/intrinsic-motivation.md>), [bestpractice](<https://devfeed.tech/topics/bestpractice.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [brand](<https://devfeed.tech/tags/brand.md>), [budget](<https://devfeed.tech/tags/budget.md>), [campaign](<https://devfeed.tech/tags/campaign.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [channel](<https://devfeed.tech/tags/channel.md>), [company](<https://devfeed.tech/tags/company.md>), [context-layer](<https://devfeed.tech/tags/context-layer.md>), [cost](<https://devfeed.tech/tags/cost.md>), [culture](<https://devfeed.tech/tags/culture.md>), [desktop](<https://devfeed.tech/tags/desktop.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [editor](<https://devfeed.tech/tags/editor.md>), [effort](<https://devfeed.tech/tags/effort.md>), [executive](<https://devfeed.tech/tags/executive.md>), [ideas](<https://devfeed.tech/tags/ideas.md>)

### AI overview

PostHog sets aside $45,000 per month for employee-proposed experiments intended to help the company stand out and make work more fun. Its process uses a Slack channel for open submissions, then an automated monthly review that weighs likely cost, effort and team enthusiasm; a small "Council of Weird" decides which proposals receive funding. The text suggests starting with roughly 5% of the marketing budget, while encouraging ideas that need no money. It advises keeping experiments coherent with the brand, testing them with people outside the company, and accepting that many will fail. Broad participation, cultural safety and an executive sponsor help ideas move forward; the budget and review process are PostHog's own approach.

### Source excerpt

"Do more weird" is an actual PostHog company value , and we have a discretionary $45k/mo budget that anyone can use to bring weird ideas to life...

## Replay Vision watches the session recordings you never will

DevFeed: [Replay Vision watches the session recordings you never will](<https://devfeed.tech/articles/replay-vision-watches-the-session-recordings-you-never-will-80593.md>)

Original publisher: [Read original article](<https://posthog.com/blog/replay-vision>)

Author: Sara Miteva

Published: 2026-07-31T00:00:00Z

Content type: release

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [recording](<https://devfeed.tech/topics/recording.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>), [Business, Industry, Usability, User Experience, User Research, Web Strategy](<https://devfeed.tech/topics/business-industry-usability-user-experience-user-research-web-strategy.md>), [product-experimentation](<https://devfeed.tech/topics/product-experimentation.md>), [Error Handling](<https://devfeed.tech/topics/error-handling.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [background](<https://devfeed.tech/tags/background.md>), [but](<https://devfeed.tech/tags/but.md>), [button](<https://devfeed.tech/tags/button.md>), [error](<https://devfeed.tech/tags/error.md>), [recordings](<https://devfeed.tech/tags/recordings.md>), [user](<https://devfeed.tech/tags/user.md>), [vision](<https://devfeed.tech/tags/vision.md>), [watches](<https://devfeed.tech/tags/watches.md>)

### AI overview

Replay Vision scans session recordings for user struggles, including issues that may not produce errors. Teams configure monitors for yes/no questions, scorers to rank frustration, classifiers to tag user intent, or summarizers to describe sessions; scanners run on demand or on a schedule, and each finding links to the relevant recording moment. In PostHog's examples, scans surfaced a misleading installation prompt and a throttled refresh loop, and a transient empty-result cache that hid existing events. A background agent investigated findings and drafted pull requests for human review and merge. Scanners can be set up or queried through Slack, PostHog Desktop, or an editor via the PostHog MCP.

### Source excerpt

Somewhere in your session recordings is the moment a user rageclicked a dead button and walked out. There was no error thrown, and nobody filed a...

## How to get hired to PostHog's marketing teams

DevFeed: [How to get hired to PostHog's marketing teams](<https://devfeed.tech/articles/how-to-get-hired-to-posthog-s-marketing-teams-80537.md>)

Original publisher: [Read original article](<https://posthog.com/blog/how-to-get-hired-marketing>)

Author: Joe Black

Published: 2026-07-30T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Tool](<https://devfeed.tech/topics/tool.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [art](<https://devfeed.tech/tags/art.md>), [b2b](<https://devfeed.tech/tags/b2b.md>), [b2b-saas](<https://devfeed.tech/tags/b2b-saas.md>), [brand](<https://devfeed.tech/tags/brand.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [creative](<https://devfeed.tech/tags/creative.md>), [data](<https://devfeed.tech/tags/data.md>), [data-stack](<https://devfeed.tech/tags/data-stack.md>), [data-warehouse](<https://devfeed.tech/tags/data-warehouse.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [devtools](<https://devfeed.tech/tags/devtools.md>), [error-tracking](<https://devfeed.tech/tags/error-tracking.md>), [hiring](<https://devfeed.tech/tags/hiring.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [linkedin](<https://devfeed.tech/tags/linkedin.md>), [marketing](<https://devfeed.tech/tags/marketing.md>), [teams](<https://devfeed.tech/tags/teams.md>)

### AI overview

PostHog's marketing hiring favors candidates who execute as well as plan, bring distinct opinions and one standout skill, adapt quickly, show genuine enthusiasm, and accept feedback without ego. The examples include a marketer building a tool to solve a recurring problem and a fast learner succeeding in an unfamiliar product area. To improve an application, the author recommends getting an employee referral, writing a brief personal cover letter (AI-generated letters are discouraged), using PostHog and offering concrete product feedback, and showing creative initiative. Direct messages asking for an informal chat can work against candidates; specific role questions are acceptable, but the hiring manager prefers the established process. The excerpt presents these as the author's hiring experience, not universal rules.

### Source excerpt

Marketing roles at PostHog are competitive to a degree that surprises people. I posted on LinkedIn that we're hiring another Developer Marketer and...

## 10,000 PRs a month is easy: How devex is evolving at PostHog

DevFeed: [10,000 PRs a month is easy: How devex is evolving at PostHog](<https://devfeed.tech/articles/10-000-prs-a-month-is-easy-how-devex-is-evolving-at-posthog-80447.md>)

Original publisher: [Read original article](<https://posthog.com/blog/10k-prs-a-month>)

Author: Paul D'Ambra

Published: 2026-07-28T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Developer experience](<https://devfeed.tech/topics/developer-experience.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [about](<https://devfeed.tech/tags/about.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerated](<https://devfeed.tech/tags/accelerated.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [application](<https://devfeed.tech/tags/application.md>), [cadence](<https://devfeed.tech/tags/cadence.md>), [devex](<https://devfeed.tech/tags/devex.md>), [easy](<https://devfeed.tech/tags/easy.md>), [june](<https://devfeed.tech/tags/june.md>), [shipping](<https://devfeed.tech/tags/shipping.md>)

### AI overview

PostHog reports that monthly pull requests rose from 1,441 in January to 4,725 in June with engineering headcount up 10%, and attributes the pace to greater AI agent use and automation. Agents now open about 70% of monorepo PRs, while self-driving workflows handle roughly 4% and are growing. To keep work manageable, PostHog is developing cloud dev machines and tools for local validation, CI management, and review; it also uses Depot for CI and is improving test selection, flaky test handling, merge queues, and job speed. Reliability measures include extensive automated tests, required PR approvals, StampHog's deterministic checks and LLM review (approving 20% of PRs), and human escalation for showstoppers. The company says PR type and revert rates have stayed broadly stable, while noting that agents can produce unusually large PRs. A `qa-frontend` skill is an early effort to have agents test changes through the browser and surface screenshots or video for review.

### Source excerpt

Shipping cadence is accelerating at PostHog. In the last 6 months, we've gone from shipping 1,441 PRs in January to 4,725 PRs in June while increasing...

## Repositioning PostHog

DevFeed: [Repositioning PostHog](<https://devfeed.tech/articles/repositioning-posthog-80594.md>)

Original publisher: [Read original article](<https://posthog.com/blog/repositioning-posthog>)

Author: Charles Cook

Published: 2026-07-23T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Product Management](<https://devfeed.tech/topics/product-management.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [build](<https://devfeed.tech/tags/build.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [cli](<https://devfeed.tech/tags/cli.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [context](<https://devfeed.tech/tags/context.md>), [context-layer](<https://devfeed.tech/tags/context-layer.md>), [copy](<https://devfeed.tech/tags/copy.md>), [data-warehouse](<https://devfeed.tech/tags/data-warehouse.md>), [desktop](<https://devfeed.tech/tags/desktop.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [developers](<https://devfeed.tech/tags/developers.md>), [error-tracking](<https://devfeed.tech/tags/error-tracking.md>), [errors](<https://devfeed.tech/tags/errors.md>), [self-driving](<https://devfeed.tech/tags/self-driving.md>)

### AI overview

PostHog is repositioning itself as the context layer that makes products "self-driving," and says the change requires a coordinated, sustained effort across its messaging and channels. Its framing separates access surfaces such as the web app, Slack, MCP, CLI and Desktop from tools like analytics, replay, feature flags and error tracking; shared context includes product telemetry and external business data, with a context warehouse covering storage and ingestion. The company presents the alternative as fragmented manual work--triaging errors, reviewing replays, writing fixes and reconnecting data across tools--and expects customers to build with PostHog from the start. It acknowledges that this emerging category needs repeated explanation and may temporarily confuse users or reduce signups. Early reactions from some customers are promising, but the shift is only weeks old and most mentions still associate PostHog with analytics; the company says it will assess progress through user conversations and adjust if needed.

### Source excerpt

As PostHog enters its next chapter , we have spent the last 6 weeks going all in repositioning as the company that makes your product self-driving...

## What is a context warehouse?

DevFeed: [What is a context warehouse?](<https://devfeed.tech/articles/what-is-a-context-warehouse-80676.md>)

Original publisher: [Read original article](<https://posthog.com/blog/what-is-a-context-warehouse>)

Author: Lizzie Epton

Published: 2026-07-21T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [Data pipelines](<https://devfeed.tech/topics/data-pipelines.md>), [context](<https://devfeed.tech/topics/context.md>), [business-intelligence](<https://devfeed.tech/topics/business-intelligence.md>), [etl](<https://devfeed.tech/topics/etl.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [accessible](<https://devfeed.tech/tags/accessible.md>), [advanced](<https://devfeed.tech/tags/advanced.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [bi](<https://devfeed.tech/tags/bi.md>), [business](<https://devfeed.tech/tags/business.md>), [business-intelligence](<https://devfeed.tech/tags/business-intelligence.md>), [context](<https://devfeed.tech/tags/context.md>), [data](<https://devfeed.tech/tags/data.md>), [data-engineers](<https://devfeed.tech/tags/data-engineers.md>), [intelligence](<https://devfeed.tech/tags/intelligence.md>), [like](<https://devfeed.tech/tags/like.md>), [look](<https://devfeed.tech/tags/look.md>), [optimized](<https://devfeed.tech/tags/optimized.md>), [warehouse](<https://devfeed.tech/tags/warehouse.md>), [what-is](<https://devfeed.tech/tags/what-is.md>)

### AI overview

A context warehouse combines storage, ingestion, modeling, and direct agent access so product events and business data can be queried together without teams maintaining a separate sync pipeline. PostHog says it mirrors product events to organization-partitioned S3 storage and brings in external data through 200+ Warehouse Sources integrations. Each organization gets an isolated DuckDB instance in a Firecracker MicroVM; idle workers sleep and wake on query, and access uses a standard Postgres wire protocol or MCP and PostHog interfaces. DuckHog supports local analysis of data subsets. The article presents this as a fit for agent-driven product teams and teams starting a data stack, while noting that organizations with advanced dbt modeling needs may still need dedicated tools; the platform is deliberately not a full data stack.

### Source excerpt

Data warehouses are optimized for analytics, business intelligence (BI), and data engineers. What would they look like if they optimized for AI agents...

## The clues were there all along. Tracing is now in beta

DevFeed: [The clues were there all along. Tracing is now in beta](<https://devfeed.tech/articles/the-clues-were-there-all-along-tracing-is-now-in-beta-80666.md>)

Original publisher: [Read original article](<https://posthog.com/blog/traces-beta>)

Author: Sara Miteva

Published: 2026-07-16T00:00:00Z

Content type: release

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [tracing](<https://devfeed.tech/topics/tracing.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>), [ai observability](<https://devfeed.tech/topics/ai-observability.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [api](<https://devfeed.tech/tags/api.md>), [app](<https://devfeed.tech/tags/app.md>), [bearer](<https://devfeed.tech/tags/bearer.md>), [bearer-token](<https://devfeed.tech/tags/bearer-token.md>), [beta](<https://devfeed.tech/tags/beta.md>), [bug](<https://devfeed.tech/tags/bug.md>), [cdp](<https://devfeed.tech/tags/cdp.md>), [checkout](<https://devfeed.tech/tags/checkout.md>), [context](<https://devfeed.tech/tags/context.md>), [data](<https://devfeed.tech/tags/data.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [distributed-tracing](<https://devfeed.tech/tags/distributed-tracing.md>), [logs](<https://devfeed.tech/tags/logs.md>), [tracing](<https://devfeed.tech/tags/tracing.md>), [user](<https://devfeed.tech/tags/user.md>)

### AI overview

PostHog tracing is in beta and accepts traces from an existing OpenTelemetry exporter through an OTLP/HTTP endpoint with a bearer token, without requiring a new SDK or rewritten instrumentation. In PostHog, users can inspect traffic volume, filter root spans by service and other span attributes, and open traces as flamegraphs. The article says traces share a warehouse and project with PostHog Logs, errors, replays, and user events, enabling investigation that links a user's spans to their session replay. It presents this shared context as useful for people and automated agents diagnosing performance problems. A setup wizard is offered, or already-instrumented users can configure their exporter directly. The text describes the feature as beta.

### Source excerpt

Distributed tracing is now in beta. If you've been using PostHog Logs , you probably saw this coming. PostHog starts with your user data and...

## Our session replays are opening their own pull requests

DevFeed: [Our session replays are opening their own pull requests](<https://devfeed.tech/articles/our-session-replays-are-opening-their-own-pull-requests-80605.md>)

Original publisher: [Read original article](<https://posthog.com/blog/session-replays-opening-pull-requests>)

Author: Sara Miteva

Published: 2026-07-10T00:00:00Z

Content type: release

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [session replay](<https://devfeed.tech/topics/session-replay.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [all](<https://devfeed.tech/tags/all.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [background](<https://devfeed.tech/tags/background.md>), [bug](<https://devfeed.tech/tags/bug.md>), [data](<https://devfeed.tech/tags/data.md>), [errors](<https://devfeed.tech/tags/errors.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [product](<https://devfeed.tech/tags/product.md>), [pull-requests](<https://devfeed.tech/tags/pull-requests.md>), [stack](<https://devfeed.tech/tags/stack.md>), [that-s](<https://devfeed.tech/tags/that-s.md>), [trace](<https://devfeed.tech/tags/trace.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

PostHog's Replay Vision scans every session selected by a scanner and classifies success or struggle; struggle signals enter the same Inbox pipeline used for errors, where agents group reports, investigate, and draft fixes. In its first two weeks, the classifier read about 3,400 sessions and emitted about 1,850 signals, a rate the article says still needs evaluation. The article gives three examples: a popover that closed immediately, a dropdown that ignored clicks across two app locations, and a cohort search that missed a result on another tab. Agents opened draft pull requests for the first two; the third fix was merged the same afternoon after product data showed the issue affected many users. The pipeline can research and write fixes without human intervention, but it opens drafts rather than merging; a human reviews before shipping.

### Source excerpt

Errors are the easiest data to act on in our product improvement pipeline . They arrive with a stack trace that's groupable and says exactly what...

## How to build a PostHog integration with the provisioning API

DevFeed: [How to build a PostHog integration with the provisioning API](<https://devfeed.tech/articles/how-to-build-a-posthog-integration-with-the-provisioning-api-80509.md>)

Original publisher: [Read original article](<https://posthog.com/blog/embedded-analytics-provisioning-api>)

Author: Matt Brooker

Published: 2026-07-07T00:00:00Z

Content type: tutorial

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Provisioning](<https://devfeed.tech/topics/provisioning.md>), [self-service](<https://devfeed.tech/topics/self-service.md>), [API Governance](<https://devfeed.tech/topics/api-governance.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [access-tokens](<https://devfeed.tech/tags/access-tokens.md>), [account](<https://devfeed.tech/tags/account.md>), [added](<https://devfeed.tech/tags/added.md>), [api](<https://devfeed.tech/tags/api.md>), [arts](<https://devfeed.tech/tags/arts.md>), [better](<https://devfeed.tech/tags/better.md>), [build](<https://devfeed.tech/tags/build.md>), [farm](<https://devfeed.tech/tags/farm.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [i](<https://devfeed.tech/tags/i.md>), [integration](<https://devfeed.tech/tags/integration.md>), [little](<https://devfeed.tech/tags/little.md>), [live](<https://devfeed.tech/tags/live.md>), [my](<https://devfeed.tech/tags/my.md>), [plants](<https://devfeed.tech/tags/plants.md>), [provisioning](<https://devfeed.tech/tags/provisioning.md>), [shares](<https://devfeed.tech/tags/shares.md>), [website](<https://devfeed.tech/tags/website.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

HogFarm uses PostHog's provisioning API to create a project for each farm and show analytics inside its own dashboard. The integration registers a public Client ID Metadata Document (CIMD); its URL must match the client_id and be reachable, and its scopes cap what tokens may request. Since the client has no secret, token exchange uses PKCE. Account creation returns a code directly for a new email, but existing PostHog users must consent in a browser; API calls pin version 0.1d. The project response provides an API key and team ID. HogFarm publishes saved Endpoints for recurring dashboard queries; their default result caching can preserve an empty first response, so the example forces refresh for newly provisioned projects. Session recording also requires enabling it at project level. The example queries recordings and enables public sharing for an iframe, noting that anyone with the link can view a shared recording. For demo events with backdated timestamps, the regular capture pipeline makes them queryable within seconds, while historical_migration can take many minutes.

### Source excerpt

I live on a little farm and recently built a website to sell shares of hogs. My fellow farmers are better versed in the subtle arts of soil, plants...

## Why we rebuilt our data warehouse on DuckDB over ClickHouse

DevFeed: [Why we rebuilt our data warehouse on DuckDB over ClickHouse](<https://devfeed.tech/articles/why-we-rebuilt-our-data-warehouse-on-duckdb-over-clickhouse-80692.md>)

Original publisher: [Read original article](<https://posthog.com/blog/why-we-rebuilt-our-data-warehouse>)

Author: Eric Duong

Published: 2026-06-29T00:00:00Z

Content type: comparison

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [DuckDB](<https://devfeed.tech/topics/duckdb.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [data lake](<https://devfeed.tech/topics/data-lake.md>), [Data Management](<https://devfeed.tech/topics/data-management.md>), [copy](<https://devfeed.tech/topics/copy.md>), [olap](<https://devfeed.tech/topics/olap.md>), [llamaindex](<https://devfeed.tech/topics/llamaindex.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [behaviour](<https://devfeed.tech/tags/behaviour.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [beta](<https://devfeed.tech/tags/beta.md>), [bi](<https://devfeed.tech/tags/bi.md>), [business](<https://devfeed.tech/tags/business.md>), [business-context](<https://devfeed.tech/tags/business-context.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [clickhouse](<https://devfeed.tech/tags/clickhouse.md>), [data](<https://devfeed.tech/tags/data.md>), [data-engineer](<https://devfeed.tech/tags/data-engineer.md>), [data-warehouse](<https://devfeed.tech/tags/data-warehouse.md>), [duckdb](<https://devfeed.tech/tags/duckdb.md>), [warehouse](<https://devfeed.tech/tags/warehouse.md>)

### AI overview

PostHog rebuilt its warehouse to support longer-running, flexible modeling queries without placing unbounded warehouse workloads on its shared ClickHouse cluster. The company says ClickHouse remains its choice for fast event analytics, but cites limitations for general warehousing: no cost-based optimizer, delayed maturity of catalog support for S3 and table formats, and query results that can vary across versions or settings. Its replacement gives each organization a single-tenant DuckDB instance that sleeps when idle and wakes for queries, with a Postgres Wire endpoint for common data tools. DuckLake stores data in S3 separately from compute, while mirrored PostHog events and synced external sources are available through the same endpoint. DuckHog lets users work on local subsets with DuckDB or dataframe tools and write results back. The excerpt presents this as a way to make data tooling and business context more accessible; the warehouse was described as a beta waitlist offering.

### Source excerpt

PostHog's original data warehouse had a single goal: delay the moment a company needs to hire their first data engineer. It did that job for a 2...

## What is a Scout? A technical deep dive

DevFeed: [What is a Scout? A technical deep dive](<https://devfeed.tech/articles/what-is-a-scout-a-technical-deep-dive-80679.md>)

Original publisher: [Read original article](<https://posthog.com/blog/what-is-a-scout>)

Author: Andy Maguire

Published: 2026-06-23T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Tool](<https://devfeed.tech/topics/tool.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-skill](<https://devfeed.tech/tags/agent-skill.md>), [agentic-loop](<https://devfeed.tech/tags/agentic-loop.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [data](<https://devfeed.tech/tags/data.md>), [deep-dive](<https://devfeed.tech/tags/deep-dive.md>), [technical](<https://devfeed.tech/tags/technical.md>), [technical-deep-dive](<https://devfeed.tech/tags/technical-deep-dive.md>), [what-is](<https://devfeed.tech/tags/what-is.md>)

### AI overview

PostHog scouts are scheduled agents whose business logic lives mainly in a readable Agent Skill, with a thin harness handling runs, memory, and signal emission. They inspect data available in PostHog, apply skill-defined criteria and restraint, and emit evidence-backed signals that the self-driving pipeline groups into inbox reports for human or automated action. Skills can range from plain markdown to reference files and scripts; users can run them through agents with PostHog MCP, while scheduled sandboxed runs can update a scratchpad and emit signals. Examples show an MCP feedback scout escalating a previously known defect after it blocked an agent, and a brand-mentions scout holding isolated or low-reach complaints below its threshold while surfacing corroborated themes. The system's tuning relies on written criteria, disqualifiers, deduplication, severity, and auditable scratchpad memory. The article says scouts were gradually rolling out and invite-only at the time.

### Source excerpt

A scout is a small agent (don't overthink that word) that looks at, wrangles, and makes sense of any data available in PostHog. It's a core component...

## What if your product built itself?

DevFeed: [What if your product built itself?](<https://devfeed.tech/articles/what-if-your-product-built-itself-80675.md>)

Original publisher: [Read original article](<https://posthog.com/blog/what-if-your-product-built-itself>)

Author: Josh Snyder

Published: 2026-06-22T00:00:00Z

Content type: article

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [AI-enhanced product development](<https://devfeed.tech/topics/ai-enhanced-product-development.md>), [Analytics Pipeline](<https://devfeed.tech/topics/analytics-pipeline.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [product analytics](<https://devfeed.tech/topics/product-analytics.md>), [Linear](<https://devfeed.tech/topics/linear.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-sdk](<https://devfeed.tech/tags/agent-sdk.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [all](<https://devfeed.tech/tags/all.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [data](<https://devfeed.tech/tags/data.md>), [error](<https://devfeed.tech/tags/error.md>), [error-tracking](<https://devfeed.tech/tags/error-tracking.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [product](<https://devfeed.tech/tags/product.md>), [product-analytics](<https://devfeed.tech/tags/product-analytics.md>), [session-replay](<https://devfeed.tech/tags/session-replay.md>), [tracking](<https://devfeed.tech/tags/tracking.md>), [web](<https://devfeed.tech/tags/web.md>)

### AI overview

PostHog describes a pipeline that turns product signals into proposed code fixes: it filters malicious signals, normalizes and groups them into reports, then uses research and actionability steps to decide whether to gather more evidence, ask for human input, or generate a PR. The team found that clustering raw signal embeddings grouped items by format rather than shared meaning; asking an LLM to describe each signal with short queries improved grouping across unlike sources. A sandboxed Claude Agent SDK researches promoted reports using product data, code context, and external MCPs, then can create a PR and resume work after CI failures or review comments. Specific errors tend to be more actionable than broad Slack reports or session replays, which may need more evidence or human judgment. The post also stresses representative production testing and suggests optimizing agent costs only after learning repeated behavior from production traces.

### Source excerpt

PostHog collects an enormous amount of data about your product ( analytics , session replay , web analytics , error tracking , experiments ) and...

## Everything (and everyone) is build mode now

DevFeed: [Everything (and everyone) is build mode now](<https://devfeed.tech/articles/everything-and-everyone-is-build-mode-now-80513.md>)

Original publisher: [Read original article](<https://posthog.com/blog/everything-is-build-mode>)

Author: Cleo Lant

Published: 2026-06-15T00:00:00Z

Content type: opinion

Language: en

Sources: [PostHog](<https://devfeed.tech/sources/posthog.md>)

Topics: [Figma AI](<https://devfeed.tech/topics/figma-ai.md>), [Product Management](<https://devfeed.tech/topics/product-management.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [account](<https://devfeed.tech/tags/account.md>), [activation](<https://devfeed.tech/tags/activation.md>), [adopting](<https://devfeed.tech/tags/adopting.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-observability](<https://devfeed.tech/tags/ai-observability.md>), [ai-products](<https://devfeed.tech/tags/ai-products.md>), [all](<https://devfeed.tech/tags/all.md>), [build](<https://devfeed.tech/tags/build.md>), [function](<https://devfeed.tech/tags/function.md>), [future](<https://devfeed.tech/tags/future.md>), [idea](<https://devfeed.tech/tags/idea.md>), [isn-t](<https://devfeed.tech/tags/isn-t.md>), [job](<https://devfeed.tech/tags/job.md>), [means](<https://devfeed.tech/tags/means.md>), [one](<https://devfeed.tech/tags/one.md>), [ship](<https://devfeed.tech/tags/ship.md>), [version](<https://devfeed.tech/tags/version.md>)

### AI overview

"Build mode" means enabling people across product, marketing, sales, and support to use AI with live product context to investigate questions and ship changes. The article describes PostHog's Slack app answering usage questions, finding engaged users, and opening draft pull requests after planning, editing files, and running checks; engineers review before changes ship. PostHog AI can connect web analytics with product data to identify content that brings in activating users, suggest target segments, and help with writing. Through the MCP, agents can read live event schemas, add tracking calls, retrieve product metrics, and build dashboards; scheduled summaries can deliver metrics to Slack. PostHog Desktop or MCP can also support building experiments. The article presents shared Slack threads and reusable agent skills as ways to collaborate and pass on team judgment. These are described capabilities and examples from the article, which also cites usage figures for the preceding 30 days.

### Source excerpt

There's a future version of you who isn't boxed into one job function. Someone who has an idea in the morning and the means to ship it the same...

[Next page](<https://devfeed.tech/sources/posthog.md?cursor=WyIyMDI2LTA2LTE1VDAwOjAwOjAwKzAwOjAwIiwgIjg1ZDYzOWYyLTIwMDItNDQwOC05NzA4LWFiMDQyMjRkOWRlNyJd>)