Vol. I · No. 32

Back Office Beat

Mon–Fri, 6:30 a.m. MT · Three to five items · Free

Published by J9 Systems
Subscribe

The daily brief · No. 26

Thursday, November 12, 2026

Procore Rate Tables, a DOL comment deadline of Jan. 11, and an OpenAI agent outage

By Carter Josephson · Filed 5:09 AM MST MT

Procore shipped Rate Tables and plans an AI timesheet capture beta for Q4, but it published no prices or plan limits. Email your Procore rep first thing and ask what Rate Tables and the AI credits will cost on your plan.

In this brief

  1. 1Procore ships Rate Tables now, AI timesheet capture beta in Q4 for subcontractorsAsk your vendor
  2. 2DOL proposes five weekly job-search activities for unemployment claimants; comments close Jan. 11Wait
  3. 3OpenAI Workspace Agents down Nov. 11: new runs blocked for 3+ hoursIgnore
  4. 4Google releases Gemini 4 Carbon at $2/$10 per million tokens; no Gmail change listedIgnore
  1. Procore ships Rate Tables now, AI timesheet capture beta in Q4 for subcontractors

    Procore's Groundbreak list has a few things a 20-to-50-person subcontractor can use today, many betas, and no prices or plan limits.

    By Carter Josephson · · Share this item

    What happened

    In a Nov. 9, 2026 post on its Groundbreak announcements, Procore said Rate Tables, SmartBench resource planning, AI agents and the Control Tower AI usage monitor are available now, and that AI-assisted timesheet capture enters closed beta in Q4. Simplified Invoicing, percent-based forecasting and cross-account RFI collaboration are in open beta, while time-and-materials-to-change-order and Production Tracking are still in development, and the post gives no prices or plan requirements.

    What it means for a 10-to-50-person business

    This matters to specialty contractors already paying for Procore, especially those whose GCs also use it. Example: a 30-person mechanical sub with 22 field workers on paper timesheets. If the office manager spends 3 minutes keying each sheet, that is 66 minutes a week, about 57 hours a year, or roughly $2,000 at $35/hr. AI timesheet capture would cut that, but it is only entering closed beta. Rate Tables, live now, lets you set labor and equipment rates once and reuse them across timesheets, T&M tickets and change orders. The bigger money is unbilled extra work: one missed crew day a month (4 workers × 8 hrs × an assumed $95/hr billing rate) is $3,040. The feature that ties T&M tickets to change events is still in development. Control Tower tracks AI 'credit consumption,' so AI agents appear to be metered, at a rate the post does not give. Residential service shops on ServiceTitan, Housecall Pro or Jobber can ignore this.

    What to do
    Ask your vendor

    Before your next Procore renewal, ask your rep in writing whether Rate Tables, Simplified Invoicing, AI timesheet capture and AI agent credits are included in your current plan, what each costs if not, and how to join the timesheet capture closed beta.

    What it costs
    What it costs
    Procore plan or add-on cost for these features
    not disclosed in the post; ask your rep
    AI agent credits (tracked in Control Tower)
    price not disclosed
    Rate Tables setup, assume 40 labor and equipment rates
    4 hrs (≈ $200 at $50/hr)
    Hand-keying paper timesheets today, 22 workers × 3 min/week
    ≈ 57 hrs/yr (≈ $2,000 at $35/hr)
    One unbilled extra crew day a month, 4 workers × 8 hrs × $95/hr (assumed)
    $3,040/mo at risk
    Doing nothing
    $0 now; you keep hand-entering timesheets and looking up rates until the betas ship
    Sources
    procore.com (primary) · Procore Support
  2. DOL proposes five weekly job-search activities for unemployment claimants; comments close Jan. 11

    The proposed rule binds states and claimants, not employers. Comments are open until Jan. 11, 2027; for most small shops the useful step is tidier separation paperwork.

    By Carter Josephson · · Share this item

    What happened

    On Nov. 10, 2026, the Department of Labor published a proposed rule (91 FR 72472, docket ETA-2026-0100) that would require states to make unemployment claimants complete at least five qualifying work-search activities each week and report them before a payment is issued. Comments are due by Jan. 11, 2027.

    What it means for a 10-to-50-person business

    For a 10-to-50-person employer, the direct effect is small. The new duty falls on claimants, not on you, and the rule is only proposed. Where it matters is claims and paperwork. Take an 18-person plumbing shop that lays off 3 helpers each January. Each claim can be charged to the shop's state UI account, which drives its tax rate. Assume a $12,000 taxable wage base and a rate that rises from 2.7% to 3.5% after unchallenged claims. That works out to 18 × $12,000 × 0.8% = $1,728 a year in extra state UI tax. Stricter work-search rules may mean fewer weeks paid on some claims. But a clear separation record and an on-time answer to the state's claim notice still do more for your rate than this rule will. Owners with steady headcount and few separations can ignore it.

    What to do
    Wait

    Take no action on the rule this month; spend an hour making sure every separation has a written reason and that someone answers state UI claim notices by the deadline, and comment at regulations.gov by Jan. 11 only if you have a view on it.

    What it costs
    What it costs
    Direct cost of proposed rule to employers
    $0 (requirement falls on states and claimants)
    Review separation and claim-response process, 1 hr at $50/hr
    ≈ $50
    Optional public comment by Jan. 11, 2 hrs at $50/hr
    ≈ $100
    Example UI rate drift from unchallenged claims, 18 staff × $12,000 wage base × 0.8% (assumed)
    ≈ $1,728/yr
    Doing nothing
    $0 now; the rule is proposed, not final
    Sources
    federalregister.gov (primary) · DOL news release, Nov. 9, 2026 · Comment on regulations.gov (ETA-2026-0100)
  3. OpenAI Workspace Agents down Nov. 11: new runs blocked for 3+ hours

    If your office runs recurring tasks on ChatGPT Workspace Agents, check that they ran today; everyone else can skip this outage.

    By Carter Josephson · · Share this item

    What happened

    OpenAI's status page reported a full outage of the ChatGPT Agent component on Nov. 11, 2026, starting at 22:50 UTC (3:50 p.m. Mountain), with users unable to start new Workspace Agent runs. Three hours after it opened, the incident was still marked "Investigating," and OpenAI had posted no cause, fix or follow-up update.

    What it means for a 10-to-50-person business

    The effect is small for most owners. It matters only if your business runs recurring work through ChatGPT Workspace Agents. Even then it covers only new runs: the status page says users cannot start them and lists no other ChatGPT component as affected. Example: a 16-person accounting firm has a Workspace Agent build the afternoon list of overdue client invoices for its two billing staff. With runs blocked from 3:50 p.m. Mountain onward, one staffer pulls the list by hand from the practice software. Assume 1.5 hours at $45 an hour, or about $68. The bigger risk is a scheduled run that never started and nobody noticed, so client follow-ups slip a day. If you have no agent workflows, the cost to you is zero. When we drafted this, OpenAI had not posted a cause, an end time or any credit. Check the incident page for the final duration.

    What to do
    Ignore

    Unless you run workflows on ChatGPT Workspace Agents, ignore this; if you do, have someone confirm on Nov. 12 that every run scheduled after 3:50 p.m. Mountain on Nov. 11 actually happened, and rerun any that did not.

    What it costs
    What it costs
    Manual fallback, 1 staffer × 1.5 hrs at $45/hr
    ≈ $68
    Checking for missed runs after the fix
    0.5 hr (≈ $23 at $45/hr)
    Service credit from OpenAI
    none announced
    Doing nothing
    $0 for most owners; if you rely on agent runs, a missed run can go unnoticed until a client asks
    Sources
    status.openai.com (primary) · OpenAI Status — all components and incident history
  4. Google releases Gemini 4 Carbon at $2/$10 per million tokens; no Gmail change listed

    Google's new top model is aimed at developers and AI Ultra subscribers; if your team only uses Gemini inside Gmail and Docs, nothing changes this month.

    By Carter Josephson · · Share this item

    What happened

    Google released Gemini 4 Carbon on Nov. 10, 2026, calling it its strongest model for long, multi-step coding, research and agent work, while saying Gemini 4 Argon remains the right choice for most software and knowledge work. Carbon is available through the API in Google AI Studio and Gemini for business to paying developers and enterprises at Argon's introductory price of $2 per million input tokens and $10 per million output tokens, and is rolling out to Google AI Ultra subscribers in Google Antigravity, the Gemini app and AI Mode in Search.

    What it means for a 10-to-50-person business

    For most owners this is a small change. Google's post does not mention Gmail, Docs or Google Workspace plans, so an 18-person accounting firm that uses Gemini only to draft client emails and summarise Docs gets nothing new and pays nothing new. It matters only if you, or a contractor, built an automation that calls the Gemini API directly. Take a plumbing shop with a Zapier or n8n flow that reads 600 supplier invoices a month. Assume 5,000 input and 1,000 output tokens per invoice: that is 3 million input tokens ($6) and 600,000 output tokens ($6), about $12 a month on Argon. Carbon costs the same per token, but Google says it uses more tokens than Argon to reason. The post gives no multiple; if it doubles usage, the same job runs about $24 a month with no clear gain on routine invoice reading. Google also calls the price introductory, so it may rise. Carbon is built for hard engineering and research work. Invoice reading and email triage are not that.

    What to do
    Ignore

    If your team uses Gemini only in Gmail and Docs, do nothing, and if an automation already calls Gemini Argon, leave it on Argon unless it is failing on long, multi-step documents.

    What it costs
    What it costs
    Gemini in Gmail/Docs only
    $0; post lists no change
    API automation on Argon, assume 600 invoices/mo × 6,000 tokens
    ≈ $12/mo
    Same job on Carbon, assume 2× tokens (post says "more tokens", gives no figure)
    ≈ $24/mo
    Testing Carbon in one automation
    1–2 hrs (≈ $50–$100 at $50/hr)
    Google AI Ultra subscription
    price not given in the post
    Doing nothing
    $0; Argon stays available at the same per-token price
    Sources
    blog.google (primary) · Gemini API pricing (Google AI Studio) · Zapier AutomationBench · Gemini 4 Carbon in Google Antigravity

Also happened

Corrections and updates

None. How corrections work.

Get the next one at 6:30 a.m.