Refolk
September 1, 2026·9 min read

Salesforce's Oct 5 Cut: 133 Roles, Zero From Core Agentforce

Salesforce's Oct 5, 2026 WARN filings free 133 roles across SF and Seattle. The reachable pool is MuleSoft plumbing, not Agentforce. Here's the math.

Salesforce layoffs October 2026MuleSoft engineers hiringex-Salesforce sourcingSalesforce WARN BellevueAgentforce talent pool
Salesforce's Oct 5 Cut: 133 Roles, Zero From Core Agentforce

Salesforce filed WARN notices in Washington and California this month disclosing 133 role cuts effective October 5, 2026: 59 across Bellevue and Seattle, 74 in San Francisco. Every sourcer I have talked to this week is asking the same wrong question, which is how to reach the ex-Agentforce engineers. The engineers walking out on Oct 5 are almost entirely from the integration and analytics plumbing beneath Agentforce, not the AI product itself.

What the Oct 5 WARN filings actually say

Salesforce is cutting 133 roles on Monday, October 5, 2026, split 74 in San Francisco and 59 across two Washington worksites (929 108th Avenue NE in Bellevue and 744 North 34th Street in Seattle's Fremont neighborhood). It is the company's third round announced in 2026 and its fourth Bay Area cut in under a year, following 86 SF roles in June.

The filings are unusually senior for a routine trim. Titles named include:

  • Software engineers and technical support engineers
  • Product management directors
  • Incident commanders
  • An SVP of software engineering
  • Tableau roles (Washington filing)
  • 37 technology and product positions, 34 general administration, 3 sales and distribution (SF breakdown)

Neither filing mentions artificial intelligence. That silence matters, because Marc Benioff has spent the last quarter publicly crediting Agentforce for a "30% productivity increase on engineering" and telling investors headcount stays "mostly flat" while sales adds 20% more account executives under CRO Miguel Milano. The WARN paperwork and the earnings-call story are describing the same restructuring in two different vocabularies.

For context on what preceded this: the June 2026 filing cut 86 SF roles across sales, G&A, and tech/product, touching teams connected to Agentforce, MuleSoft, and Marketing Cloud. February 2026 cut roughly 1,000 employees. Customer support alone went from 9,000 to 5,000 headcount over the arc Benioff has described publicly. The pattern is consistent: the plumbing and support layer contracts, the AI product headline stays intact.

Why the reachable pool is MuleSoft, not Agentforce

The reachable Oct 5 pool is overwhelmingly integration engineers, MuleSoft-adjacent staff, and Tableau roles, because Salesforce has consistently spared core Agentforce across all three 2026 rounds. If you are sourcing for an AI product team, this cut does not help you. If you are sourcing for enterprise integration, it is the best window of the year.

The numbers from Refolk's index of professional profiles make the mismatch obvious.

SegmentUS countSource
MuleSoft / Integration Engineer pool2,237Refolk's index, US-only
MuleSoft + Salesforce + integration in headline11Refolk's index, specialist tail
"Agentforce" keyword in headline56Refolk's index; 8 still at Salesforce
Ratio MuleSoft to Agentforce~40xDerived from Refolk's index
Oct 5 SF cuts, tech/product only37 of 74SF Chronicle WARN summary
Combined Oct 5 layoffs133Salesforce WARN, CA + WA
Same-day Seattle senior-tech WARN load~150Salesforce 59 + Zillow 91
40x
MuleSoft/integration pool vs Agentforce pool
Refolk's index shows 2,237 US integration engineers against 56 "Agentforce" profiles, and 8 of those 56 still work at Salesforce.

The Agentforce pool outside Salesforce is effectively a rounding error. If your JD says "5 years shipping agentic AI at scale," you are fishing in a bucket of about 48 people nationally, and half of them are getting recruiter DMs hourly. The MuleSoft cut, by contrast, releases people into a pool that already has 2,237 credible profiles and clear non-CRM landing spots at Apple (4 profiles), Epic (2), Meta (2), and Intel.

The specialist tail is dangerously thin

Inside that 2,237-strong integration pool, the subset that combines "MuleSoft," "Salesforce," and "integration" in a single headline is only 11 US profiles. Three of them still work at Salesforce. The other eight sit at Accenture, Slalom, IntraEdge, and JPMorgan Chase. These are exactly the SI and enterprise buyers who will absorb Oct 5 supply first, because they already have the JD templates and the internal referral graph.

If you want any of those 11 profiles, or the roughly 30 more that will surface once Oct 5 hits and headlines update, the useful move is not a keyword search on LinkedIn. It is describing the person in plain English: "senior MuleSoft integration architect, ex-Salesforce or ex-partner, based in SF or Seattle, has shipped CRM-to-ERP flows in the last 18 months." That is the exact gap Refolk closes, because keyword filters miss anyone whose current headline still says "Principal Engineer" without the tool name.

The Seattle stacking effect on October 6

Two brand-name WARN filings hit the same Monday in Seattle: Salesforce (59) and Zillow (91), roughly 150 senior tech workers landing on the market on a single day. Recruiters who wait for "the market to cool" will miss the 72-hour reply window when these candidates are still triaging inbound.

Here is the mechanism most sourcers underweight. WARN-triggered candidates behave in a predictable arc:

  1. Days 0 to 3: shock, high inbound volume, candidates reply to whichever message names their exact prior team or a specific project
  2. Days 4 to 14: sorting phase, candidates open 2 to 4 loops with warm leads only
  3. Days 15 to 45: decision phase, offers get compared, competing bids from Accenture and Slalom close
  4. Day 45+: the interesting profiles are already in final rounds elsewhere

The Zillow 91 and Salesforce 59 will not compete for the same roles in most cases (Zillow's cut leans product and data, Salesforce leans integration and Tableau), but they compete for recruiter attention on Oct 6. If your outreach template on that Monday says "I saw the news at Salesforce," you are one of 400 identical messages in their inbox. If it references a specific MuleSoft connector, a specific Tableau dashboard project, or a specific integration pattern from a public conference talk, you are one of three.

The WARN paperwork and the earnings-call story are describing the same restructuring in two different vocabularies.

The Tableau signal nobody is searching

The Washington filing explicitly includes Tableau roles, and "ex-Tableau" should be treated as a distinct search vector, not a subset of "ex-Salesforce." Tableau's Seattle office has been quietly shrinking since the acquisition, and headlines on those profiles often still say "Tableau" rather than "Salesforce," which means Boolean searches on "Salesforce" miss them entirely.

Practical implication: two separate search passes are needed on Oct 6.

  • Pass 1: current title contains Salesforce, location SF or Seattle, seniority staff or above, past team includes MuleSoft or Integration Cloud
  • Pass 2: current title contains Tableau, location Seattle metro, past 24 months, any analytics or viz specialization

Pass 2 will return a smaller but far less contested list. The Tableau viz specialists are prized by fintech and healthcare analytics teams, and the same profiles rarely show up in a "Salesforce layoff" search because the candidate's own self-description never updated after the 2019 acquisition.

The reason-gap that undercounts the reachable pool

Neither WARN filing mentions AI, but Benioff's public statements about a 30% engineering productivity gain from Agentforce imply attrition well beyond 133. The reachable pool on Oct 6 is larger than the filings suggest, because voluntary departures track the same underlying signal and do not require a notice.

Consider the gap between two facts:

  • Benioff says Salesforce has about 15,000 engineers, "mostly flat" over the past year, "more productive than ever"
  • Q2 revenue was $11.345 billion, up roughly 11% YoY, with Agentforce ARR crossing $1.5 billion at 240%+ growth

An 11% revenue business with flat engineering headcount and a fast-growing AI SKU has to be moving people around internally. Some of that movement is voluntary, and voluntary movers do not appear on WARN filings. In practice this means the sourceable pool of ex-Salesforce integration and Tableau talent in October and November 2026 is probably closer to 200 to 300 than to the 133 the filings name. Refolk is useful here specifically because voluntary movers update LinkedIn on their own schedule; a natural-language query catches them the day the headline flips, without requiring a keyword you have already guessed.

5,000
Salesforce customer support headcount, down from 9,000
Benioff's own figure. The pattern of "cut plumbing, spare Agentforce" is public and consistent.

What to actually do the week of October 5

Run three parallel searches by Wednesday Oct 7, treat Tableau as its own vector, and message with a specific project reference rather than a generic "saw the news" opener. The candidates who reply first are the ones who feel seen, not the ones who feel counted.

Concrete playbook for the week:

  • Monday Oct 5: file the initial searches, do not send outreach yet, WARN candidates are in shock
  • Tuesday Oct 6: send first messages, prioritize the 11-profile MuleSoft + Salesforce + integration slice, then the broader 2,237-profile integration pool
  • Wednesday Oct 7: run the separate Tableau pass on Seattle metro
  • Thursday Oct 8: revisit voluntary movers, look for headline changes among current Salesforce engineers not in the WARN filing
  • Following week: expect Accenture and Slalom to begin closing on the specialist tail; move to offer or move on

The Salesforce Ben community (salesforceben.com) is worth monitoring for granular role breakdowns beyond the filing's totals. It has consistently broken team-level detail on prior 2026 rounds and is the closest thing to a signal source for ex-Salesforce sourcing intel.

If your team's real job is finding the ex-Agentforce engineer everyone wants, the honest answer is that pool is 48 people and you should stop hunting Salesforce alumni and start hunting the applied-ML engineers at frontier labs who shipped agentic workflows in the last 12 months. If your job is finding integration talent that can move enterprise data at scale, Oct 5 is the best window you will get this year.

FAQ

How many people is Salesforce laying off on October 5, 2026?

133 total, split 74 in San Francisco and 59 across Bellevue and Seattle. The SF portion breaks down to 37 tech and product roles, 34 general administration, and 3 sales and distribution. The Washington portion includes Tableau roles. It is Salesforce's third announced round of 2026 and its fourth Bay Area cut in under a year.

Are ex-Agentforce engineers reachable in this round?

Almost none. Salesforce has consistently spared core Agentforce across all three 2026 rounds, and Refolk's index shows only 56 US profiles with "Agentforce" in the headline (8 of them still at Salesforce). The reachable pool from Oct 5 is overwhelmingly integration engineers, MuleSoft-adjacent staff, and Tableau roles, not AI product talent.

Why do sourcers say to move within 72 hours of the WARN date?

Because WARN-triggered candidates sort inbound sharply in the first three days. On days 0 to 3 they reply to messages that name their exact prior team or project. By day 15 they are in decision mode with Accenture, Slalom, and JPMorgan already competing on the specialist tail. The 11-profile MuleSoft + Salesforce + integration slice in Refolk's index will be off the market fastest.

What is the best search strategy for Oct 6?

Run two separate passes. First, current or recent Salesforce, SF or Seattle, MuleSoft or Integration Cloud on the past-team field. Second, current Tableau, Seattle metro, any viz specialization, because Tableau profiles often never updated to say "Salesforce" and get missed by Boolean. Use natural-language queries to catch voluntary movers whose headlines change after Oct 5 but before you would think to run the search again.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next