Bryan Helmig 🍻

@ankrgyl congrats! loving braintrust at zapier.

We sent this note to our customers to let them know that Braintrust has raised a new round of funding, and thank them for their support. While the money is exciting, our focus hasn't changed: we're building Braintrust to help our customers ship quality AI products. In 2026, AI is moving to production but teams have never had less conviction about what will fail next. Our customers are building AI products that serve millions and simply need to work. If Braintrust makes their lives easier and their products better, I know we are doing our job. Thank you to @ICONIQCapital for leading our Series B, and to @a16z, @GreylockVC, @basecasevc, and @eladgil for doubling down. Thank you to the Braintrust team for all the incredible work you've done over the past year. And thank you to our customers, who have made this growth possible.

opus 4.6 feels faster but also does things that opus 4.5 wouldn't have, eg. confusing repos/paths, etc.

@claudeai wishlist item for claude code: upon exit, print the `claude -r <uuid>` command to resume the session...

kudos to @Ubiquiti for an increasingly rare product that works just fine locally -- even when the internet is down

@acoyfellow also true

don't all major code agents constantly fill the context window (and prune it, etc..)? ralph is a inverted approach to thinking about context, if i'm grasping it right. my typicla cursor flow: ask > plan > agent. feels reliable.. but i am the harness. i feel like i'm impeding progress. with a ralph loop and way to mentally manage the context, it feels like my job shifts into putting the harnesses (tests, proof oc concepts, reliable production tests) in place and letting the AI decide the details. no more complaing about "oh i'm so tired of hearing claude say 'you're absolutely right'!"... or even being mad at llms. i'm guilty no more of cursing. poor little ralphs is all they are. we've been trying to just overload context windows maybe. i'm sceptical enough to test it but bought in enough to know it's inevitable. x.com/MichaelArnaldi… is a great read.

the primary reason ralph wiggum is effective is because you're forced you to plan upfront

Excited to see what people build with OpenAI's Agent Builder + Zapier MCP. Together you can build Agents that go beyond Chat and do work within your entire stack (Zapier MCP hooks into over 8,000 applications). A few early favorites 👇 * Streamline Onboarding: Use secure People Team credentials to create accounts, assign training in your LMS, and schedule manager check-ins automatically. * Improve Team Engagement: Summarize sentiment from your survey tool, draft follow-up actions, and send those directly to managers with talking points. * Speed Up Customer Support: Triage tickets, enrich them with CRM data, and route alerts to the right team channel. * Streamline Procurement: Validate spend requests, check budgets in your ERP, and push requests to approvers. * Weekly Project Reporting: Pull from project tracking updates, code commits, and design files to generate polished summaries. What will you build with Agent Builder + Zapier MCP?

Every new thing launched at ZapConnect: First up: Unified Copilot. Type what you want, and it builds across Zaps, Tables, Interfaces, and Agents: - Workflows in minutes (not days) - Context flows everywhere (no rework) - AI does the wiring so you focus on strategy (Try it here: zpr.io/7BvrbugqNBze) We also shipped: 1. Human-in-the-loop 2. Over 30 new AI apps: Perplexity, Mistral, Cursor, DeepSeek + more 3. Agent Sharing: Build once, deploy to your whole org 4. Enterprise controls your IT team will actually like Oh and Tables + Interfaces are now included in every @Zapier plan. Missed the event? Replay’s up now.

Zapier MCP is now available inside ChatGPT. That means you can connect to 8,000 apps+ and trigger Zapier workflows, just by writing what you want to happen. @ChatGPTapp + @Zapier MCP will figure out the right tools for the job and run the actions for you. Getting started is easy: 1. Head to the Zapier ChatGPT MCP Server and add the tools you want ChatGPT to access 2. Follow the steps in the “Connect” tab 3. If you're an admin on a ChatGPT Business or Enterprise account, you’ll see extra steps there to enable MCP across your workspace Try it here: mcp.zapier.com/mcp/servers?cl…

check out agents here: zapier.com/agents come help us build agents: #job-openings" target="_blank" rel="nofollow noopener">zapier.com/jobs#job-openi…

zapier agents can work as a team now. killer feature! also, we're hiring! links in thread.

a couple notable points about gpt-5 launch today: * cached token pricing for gpt-5 is 10x cheaper v. ~4x in 4.1/o3/etc. * introduces context-free grammars (!!!) as an alternative to JSON in tool calling * more: preambles to tool calls, "minimal" reasoning, restricting to N tools

IT'S HERE ⚡️ Our @zapier integration has landed, so now you can automate what happens after a meeting.

@AutomationAce_ how far off are sub-zaps from that ideal?

Wish Zapier would release this feature to make it easier to users to do Zap Chaining to send data from a Zap step via a webhook to another Zap that has a Webhook trigger step. Action: Send to Zap Options: - Zap -- This would list the Zaps with these trigger steps: --- Webhook - Catch Hook --- Webhook - Catch Raw Hook - Format -- Form -- JSON -- Raw -- XML The listed Zaps would have this option structure: - Label: Zap Name - Value: Zap ID

Super cool role open @zapier - if you’re interested in working on MCP and AI Agents and you’re a Sr. Full Stack Engineer in North America, you should apply! jobs.ashbyhq.com/zapier/b52d119…

Quarter-billion alert: 250 million AI-powered tasks have been automated on @Zapier We’re officially in the era of AI orchestration.

@AutomationAce_ A fix coming today. Not all APIs support returning lots of results, but we're working on it.

Zapier seems to have really blundered the rollout of the new "If multiple results are found" feature in Zap "find/search" steps. EXAMPLE Broken in the Zap action: Zapier Tables - Find Records Max results returned is 3. Documentation has no indication as to the max # of results that can be returned.

several instances of multi-week engineering estimations falling to claude code within an hour... it's happening!

Copy @zapier's workflows: HR edition 5 templates // save ~20 hours a month Warm intro recruiting (~2-3 hrs/week saved) bit.ly/3Ssxh7G Automated offboarding (~3-5 hrs saved per exit) bit.ly/43II56Y AI-powered hiring docs (~3-5 hrs saved per role) bit.ly/452QYdM New-hire ReadMe generator (~1-2 hrs saved per hire) bit.ly/3SZUDBP 1-on-1 meeting coach/note-taker (~15-30 mins saved per meeting) bit.ly/3HKAwFh

Bringing back the Friday AI leaderboard! ⏫ @zapier agents 🔼 @cursor_ai + sonnet-4 🔼 @chatprd @linear agent 🫳 @chatprd > @v0 🫳 @ChatGPTapp o3 🫳 @DevinAI ⏬ @claude_code ⏬ @ChatGPTapp 4o ❌ @cursor_ai + Sonnet-3.7 ❌ MCPs ❌ @meetgranola (I had v few meetings!)