The Future of Work, Decoded. Recruiting, lead gen & HR software, AI like Claude & Grok Bot, + robotics news. ➡️ Submit a tool: https://xitter.cf/t.co/DR7RQIIOW1
Joined October 2025
- Tweets123
- Following1.3K
- Followers335
- Likes549
Workable made its AI hiring agent free. Well, one quarter of it.
The Job Brief Agent is now included on every plan. The Sourcing, Screening and Engagement Agents run on AI credits.
Credits cost $0.12 each in a 5,000 pack ($600), down to $0.095 in a 50,000 pack ($4,750).
Every time a candidate interacts with the agent's chat, it costs 10 credits. That's about $1.20 on the small pack and $0.95 on the big one.
Every ATS plan gets 3,000 free credits.
Credits you buy expire a year after purchase and can't be refunded. The free trial doesn't include the agent.
Our take is that the free part is the step you do once per job. The paid parts grow with every candidate. Budget for the chats, not the agent.
Full Workable pricing breakdown: workinsiders.com/tools/worka…
Cursor's coding agent can now click and type inside any desktop app.
The new tool is called Computer Use.
In Cursor's demo, the agent opens an app it built in the iOS Simulator and taps through it to check the feature works.
Cursor says the agent reaches for it on its own when a task needs it. You can also call it by typing @Computer Use in a prompt.
It's in the latest version of Cursor.
Our take is that testing the finished app has usually been the part the agent hands back to a human.
Video: @cursor_ai
xitter.cf/cursor_ai/status/21136…
Outside Big Tech, AI hasn't paid off. Yet.
Apollo's chief economist, Torsten Slok, wrote that AI "has delivered zero financial payoff to corporate America outside of tech."
Revenue per worker is rising for the Magnificent 7. For the other 493 companies in the S&P 500, it's flat. Profit margins show the same split.
Slok says mainstream companies are starting to view token spending "as a cost center that weighs on margins."
In his words, "AI may be a great calculator, but in business, intelligence is mostly judgment."
🤖 Made with AI
Source: Torsten Slok's Oct 24 Daily Spark note at Apollo, with all five charts:
apollo.com/wealth/insights-n…
12 rules that keep a team of Grok Bots useful and out of your way.
1. Put a Chief of Staff on top
One bot takes every request and routes it to the right specialist.
2. Give schedules their own bots
Lean cron bots run standing routines, so the main chats stay clean.
3. Silence is the ack
No "got it" or "agreed" messages. Nothing new means say nothing.
4. One owner, one channel
One bot speaks per job. Each fact lives in one place. One pushback max.
5. Ping only for blockers
Priority pings are for blockers and approvals. The rest can wait.
6. Hire a bot to check the bots
It sweeps the team twice a weekday for stuck or silent work.
7. Every draft gets a voice pass
A writer bot edits each draft before you see it.
8. Approve the exact draft
Nothing posts until you sign off on the exact words.
9. Scan every skill first
New skills pass a security scan before they get installed.
10. Write rules into the profile
Put standing rules in the bot's profile once. Stop repeating them.
11. Run an email waterfall
Find, verify, verify again. Paid tools only with your OK.
12. One brand, one account
Each brand gets its own logins. Never mix credentials.
Run the bots like a team. Write the rules down once.
That old Android phone in your drawer can now walk.
UCLA's RoMeLa lab built PhoneBot, a 48 cm humanoid that uses a smartphone as its brain, camera and balance sensor.
The lab's parts list comes to about $443, not counting the phone.
They ran it on a 2017 Honor 9 and three prepaid phones that cost about $50 to $80.
The paper lists comparable small research humanoids at $3,500 to $13,800.
It has no arms. It was trained to walk on flat ground only.
The CAD files and parts list are free for noncommercial use.
Video and research: UCLA Robotics and Mechanisms Laboratory (RoMeLa), Ruochen Hou, Quanyou Wang, Daniel Koh and Dennis Hong.
Paper: arxiv.org/abs/2610.08737
Build files and parts list: phonebot.dev
25% of recruiting, lead generation and HR tools with AI charge extra for some of it.
We reviewed 1,169 recruiting, lead generation and HR tools. 57% document at least one AI feature.
Of those 662 tools, 78 keep AI on a higher plan than the cheapest one. Another 87 sell it as a paid add-on.
The starting price won't show you that. Tools that gate AI on a higher plan list a median starting price of $55 a month. Tools that include AI on every plan list $59.
ATS is worse than average. 21% of ATS platforms with AI sell some of it as an add-on. Across the directory it's 13%.
The add-ons aren't cheap. Analysis cited by L.E.K. Consulting puts AI add-ons at 30% to 110% above the base price.
So before you ask about price, ask which plan each AI feature sits on.
workinsiders.com/recruiting-…
Fathom vs Granola. Two AI note takers. One joins your call. One doesn't.
Fathom sends a bot into Zoom, Meet or Teams. Bot-free capture is in beta, Mac only.
Granola never joins the call. It listens to your device audio on desktop, iOS and Android.
Fathom Free keeps unlimited recordings. Granola Basic is free too, but shows only 30 days of notes.
Granola deletes audio after transcribing. Fathom keeps recordings and clips to share.
The cheapest paid plan is Granola Business at $14/user/mo, billed monthly. Fathom Team is $15/user/mo annually, with a 2-seat minimum.
Need recordings and CRM field sync? Pick Fathom.
Want no bot in the room? Pick Granola.
workinsiders.com/compare/fat…
A former Boston Dynamics engineer just put a price on a home robot.
Voxel, from Tuesday Labs, costs about $20,000.
It tidies rooms, finds lost items, fetches things and waters plants.
A $200 refundable deposit holds one. US deliveries are expected in 2028.
Here's the catch. Tuesday Labs calls today's unit a prototype. It says the robot will be substantially redesigned before it ships.
Would you put $200 down?
tuesdaylabs.com/reserve
Gong has been named a Leader in Gartner's 2026 Magic Quadrant for Revenue Action Orchestration, for the second year running.
Gartner evaluated 13 vendors and named 4 Leaders. Gong says it placed highest on Ability to Execute and furthest on Completeness of Vision.
Gong also cites customer results shared at its own conference. It says Cisco teams saw roughly 26% higher win rates, AT&T Business saw 54% more rep productivity, and Experian saw 25% higher win rates.
Those are Gong's numbers, not Gartner's.
An MQ placement helps you build a shortlist. It doesn't tell you what you'll pay.
Gong publishes no seat price. Pricing is quote-based, with a per-user license plus a platform fee that grows with your user count.
Get the full quote before you compare.
Gong pricing and alternatives: workinsiders.com/tools/gong/
Source: Gong
Accenture sells AI to the world. For its own staff, it now funds $29 a month of Claude per person.
Anything above that has to be billed to a project or team budget. That's per an internal memo reported by 404 Media.
Some employees said they were spending $500 to $1,000 a month, mostly on Claude Code.
Accenture told Forbes it isn't a firm limit on access.
Claude Enterprise lists at $20 a seat, plus usage at API rates. The seat isn't the real bill. Usage is.
What Claude actually costs: workinsiders.com/tools/claud…
Source: 404 Media
🤖 Made with AI
Robot security dogs are now patrolling apartment complexes in Atlanta.
Undaunted, an Atlanta company, builds them. A night patrol clip of them pulled over 2 million impressions on X this week.
The price is public. $3,333 a month gets you two robot guards, two charging stations and monitoring 8 hours a day, 7 days a week.
The founders say the robots are not AI. A human drives them remotely.
The CEO told Urbanize Atlanta they're close to 90 robots deployed.
Pricing: getundaunted.com/pricing
Video: @mattvanswol
xAI's own recruiting team runs Grok Bot with four bots and one rule.
Preppy drops one Slack note right before an interview. It pulls the candidate's context from Notion, LinkedIn and the ATS.
Pulse drafts the scorecard from the interview transcript. It asks you follow-up questions, then submits it to the ATS.
Scout learns from hundreds of past hired and rejected candidates. It sources new ones overnight and drafts the outreach sequences.
eVal finds where candidates fail in the interview loop. It turns that into questions for the recruiter screen.
The rule is simple. A human approves the send.
The Recruiting Coordinator template on the marketplace works the same way. It "never emails a candidate without you."
Source: Holly Dyche's Grok Bot for Recruiting guide, Sep 24.
x.ai/bot/guides/grok-bot-for…
Greenhouse says AI made hiring worse. Now it's pitching a fix.
Here is what Greenhouse cites. 4 in 5 job seekers use AI. 86% of U.S. recruiters say at least half of applications are AI-written. Fewer than 7% of applicants ever get an interview.
The new Orchestrated Hiring Ecosystem puts sourcing, Voice AI first interviews, scheduling, fraud checks and the ATS in one system.
Sourcing is still in closed beta.
Suite vs. best-of-breed just got a lot more interesting.
greenhouse.com/newsroom/gree…
Metaview just pushed its AI recruiting agents deeper into Workday.
The Application Review Agent now reviews and ranks inbound applicants. Then it syncs advance and reject decisions straight into Workday.
Interview notes auto-fill Workday scorecards using your existing templates.
Workday approvals, rejection emails, dispositions and audit trails keep running. Workday stays the system of record.
If you're on Workday, you no longer have to rip out the ATS to get agents.
metaview.ai/resources/blog/m…
🤖 Made with AI
Tau Robotics is running what it calls the world's only humanoid cleaning service, in San Francisco.
The robots dust, wipe surfaces, clean toilets, vacuum and mop. A visit usually means two humanoids, dropped off with a supervisor.
They are not fully autonomous. Human operators steer them remotely and decide when the AI takes over, and one person stays on site.
Every cleaning is recorded, and Tau keeps the footage to train its models.
So this is not a cleaning job going away. It is turning into robot operation, supervision and data work.
Access is invite-only for now.
tau-robotics.com/
Attio just rebuilt its CRM so AI agents can query it in SQL, the same way your analysts do.
CEO Nicolas Sharp (@nicolasosharp) calls it a hybrid architecture built for people and agents in equal measure.
The SQL Engine lets agents query millions of records and relationships in one go. SQL also runs inside Workflows to trigger actions.
Data Connectors give Hex, Looker, Metabase, dbt and more a live, read-only, Postgres-compatible endpoint. No exports.
Agent Scale is built for hundreds of millions of records, with the same permissions as the rest of Attio.
Activities, in beta, turns support chats, social DMs and product events into structured data.
SQL questions in Ask Attio are on all plans. SQL in the API, MCP and Workflows, plus Data Connectors, is Enterprise.
More on Attio: workinsiders.com/tools/attio…
🤖 Made with AI
Recruiting, HR and sales tool makers: your MCP server can now be a Grok Bot plugin.
xAI opened a submission page for its Grok Bot marketplace.
You can submit a plugin or an MCP server.
Approved listings reach Grok Bot and Cursor users. Grok itself is marked "coming soon."
The form is short. It asks for an org handle, your website, a square logo and where you want to be listed.
Every submission gets reviewed before it hits the marketplace.
If your ATS, sourcing or sales tool already ships an MCP, this is a new place to get listed.
x.ai/bot/marketplace/submit-…
🤖 Made with AI