D

@Shpigford DataForSEO

what's the best SEO MCP/CLI tool? i don't want to dig through some complex UI. i want to hop in claude/hermes and just say "based on my product at replysocial.⁠co, what are 20 starting keywords i should focus on ranking for?"

@JeremyNguyenPhD No I straight swapped it. But likely winning (ie kept team productive) bc GLM handled spreadsheets and specific needs at like 85% good. Team is non technical so hiccups were measured against “really shitty manual data cleanliness work”

@dshan I do have Zai coding subscription but I just have not been able to get GLM to even get me Sonnet level let alone close to Opus. Probably skill issue on my part or maybe I need to try more: are you prompting or treating GLM quite differently from say Claude models?

Direct from Sam Altman: You can use your ChatGPT subscription to power your OpenClaw. What's your current setup? Claw with GPT? Hermes with OpenCode? Asking because my Claw has been mostly out of action since Anthropic stopped allowing it on their subscription.

I think this is accurate And kudos to Sam, masterclass in vague PR

These posts make me miss LinkyDink and day zero of Product Hunt when we all just fired these finds into an email. cc @rrhoover

Man. Opus 4.7 is not good.

@sergiowpf @nbaschez Good question. Not a huge team, but not much of that, focused on the growing client knowledge being used by everyone. I do silo separate agents for our finances and other stuff, but don't put that stuff into markdown knowledge hub.

@dshan @nbaschez That’s really interesting, I’m working on something similar at work. But I was wondering, are you doing anything to make it truly multi-tenant? Make sure that someone’s agent only has access and permissions that are relevant for their role?

I've worked with a few companies on this and spent a ton of time thinking about it The hard part is if you want your employees using Claude, you don't have enough centralized control over their system prompt or skills, so no matter how well-organized your Notion or whatever is, the agent likely won't use it effectively But everyone wants to use Claude and is very happy to be locked in by them, assuming they'll build whatever you need in a few months if it's not already there (probably correctly?) Hard to find the right lane for this as a startup

Right now, it lives in GitHub. My team is non-technical so I’m the primary editor and the agents obviously update edit and keep it fresh. I’ve been pretty adamant about not getting locked into Claude so I use open claw. They love it and we initially focused on markdown files for every client that we have. That’s made open claw really powerful for the team to do productive work against each client relationship and interaction.

@dshan Where does the markdown live?

@_CallMeMacy What’re you using to ingest emails from the team? Agree 100%, close to ditching HubSpot

Lately, I've been advising founders to vibecode their own CRM that reads their emails, calendars, and keeps things up to date in real time. It took me 2 hours to create my own using Claude Code (and I've never coded in my life). Am I crazy for suggesting this or do others agree?

@nickgraynews @MediaKing fwiw i'm moving a bunch of sites to markdown, it's absolutely wild how much more agency you get when your ai agent can just hit the codebase top to bottom. content is just a convo in slack now, UI/UX same thing, all at scale.

I’m open to challenging that assumption! I guess I have always thought that I would want one if someone was making a site for me so that I could easily get in on the desktop to edit some of the content if I needed to But perhaps that is an idea that should be left to the year 2015 And I should full send it to just do it all via an API or with an agent

I’ve spent the past 3 weeks trying to port a handful of WordPress sites to EmDash CMS, a new open source CMS from Cloudflare But it might all be for naught… I believe my love of Cloudflare + excitement for "the shiny new thing" blinded me towards what I really need for my business case After banging my head with build times, load times, etc I had to step back Now I’ve spent the a few hours today really challenging my assumptions and looking at other options like Payload CMS, SonicJS, and some git-based CMS solutions like Alinea and Sveltia I'm still not sure what I'll end up on, but I am excited about what we're building tl;dr I may have wasted TONS of time and thousands of dollars of engineering contractor time on my team for the wrong solution!! FML I've been hammering Opus 4.7 and Gstack’s skills to help me think it through, plus getting second opinions from Gemini and Codex. I'll try getting a second opinion on my PRD from GPT 5.5 too Happy Friday! Any suggestions appreciated

@Shpigford @conductor_build hah! indeed. i'm not smart enough to know which to choose

@dshan @conductor_build why not?

unfortunately i'll never get to install the @conductor_build super extra mega hyperspeed update because YA BOI NEVER STOPS SHIPPING AND THUS NEVER HAS ANY IDLE TIME FOR IT TO INSTALL

@iAmHenryMascot @openclaw Hero capes

If you run into this depend issues Here's the fix: Workaround for OpenClaw 2026.4.20 bundled channel dependency failure: Telegram: pkg="$(npm root -g)/openclaw" npm --prefix "$pkg/dist/extensions/telegram" install --ignore-scripts --package-lock=false --no-save --no-fund --no-audit openclaw doctor --non-interactive openclaw gateway restart WhatsApp: pkg="$(npm root -g)/openclaw" npm --prefix "$pkg/dist/extensions/whatsapp" install --ignore-scripts --package-lock=false --no-save --no-fund --no-audit openclaw doctor --non-interactive openclaw gateway restart Windows PowerShell Telegram: $pkg = Join-Path (npm root -g) openclaw npm --prefix (Join-Path $pkg "dist/extensions/telegram") install --ignore-scripts --package-lock=false --no-save --no-fund --no-audit openclaw doctor --non-interactive openclaw gateway restart

OpenClaw 2026.4.20 🦞 🧠 Kimi K2.6 support + provider-aware /think 💬 BlueBubbles iMessage sends + tapbacks fixed ⏰ Cron state/delivery cleanup 🔐 Gateway pairing + plugin startup hardening Less haunted. More useful. github.com/openclaw/openc…

@ShaanVP Breaking out

dwarkesh 📈

Is my @openclaw lying to me or can we not use agents for @figma new mcp? "mcporter cannot auth with Figma's MCP server. Figma blocks dynamic client reg (returns 403) from unlisted clients. Only whitelisted clients (Claude Code, Cursor, VS Code, Codex, Warp, etc.) can register"

codex is so timid

i've generally cared very little about "privacy" when choosing tools for work stuff - using google, slack, hubspot etc but putting my entire AI agent stack on claude or gpt feels very different to me, and i'm actively shooting for open source, self-hosted

underrated if using @openclaw with a noob team: watch everything (we use slack) and jump in aggressively to re-prompt, diagnose and debrief we've seen fast ramp up on skill in asking for things just by watching me interact w agents

so much of the openclaw vs hermes vs headless and opus vs gpt vs minimax vs glm vs kimi feels like back in the day when everyone was screaming about new tech stacks and @levelsio make like $1m/yr with a single php file just ship, and talk about what you ship