AI news videos

Salesforce Expands Agentforce With Seven Job Ready Agents

Salesforce has introduced a new portfolio of job-ready Agentforce agents built for specific jobs across sales, service, commerce, employee experience and the back office. Read it for what it is: the vendor's own product announcement, not an independent evaluation.

Watch on YouTube

Transcript

Sources

Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:

• Every factual claim in this script comes from a single first-party source, Salesforce's own newsroom announcement "Salesforce Expands Agentforce With a New Portfolio of AI Agents Built for High-Value Work", published on salesforce.com/news/stories on September 11, 2026 and dated in the body as a five minute read filed from San Francisco on September 11, 2026; the page was fetched live on 2026-09-11 and matched the alert-time snapshot at scripts/data/article-cache/3d260cd2177b8566.txt word for word, with the single difference that the live fetch resolved a scrape artifact in the snapshot sentence about the three capabilities, which reads correctly as "three capabilities make that possible:". SENTENCE AUDIT FOLLOWS, and it covers the twelve sentences of the revised narration. Hook sentence one, "Salesforce expands Agentforce today with seven job ready agents", is supported by "Today, Salesforce is introducing a new portfolio of job-ready agents built to take on high-value work across sales, service, commerce, employee experience, and the back office" and by the bulleted list that names exactly seven agents, Casey, Paige, Carter, Hunter, Marshall, Piper and Fin; I replaced "is introducing a new portfolio of" with "expands Agentforce today with" for brevity, wrote the count as the word "seven" because digits are garbled by the text-to-speech engine, and removed the hyphen from "job-ready" because hyphenated compounds are stuttered in speech. Hook sentence two, "They cover sales service and the back office", is supported by the same sentence's list of "sales, service, commerce, employee experience, and the back office"; I dropped commerce and employee experience for time and removed the commas because each comma becomes a spoken pause, and I did not add any function the article does not list. Scene one sentence one, "Casey handles customer service while Paige resolves IT and HR requests", is supported by "Casey, your help agent, resolves customer service issues across voice, SMS, WhatsApp, and web chat, with pre-built support for FAQs, returns, account management, human escalation, and more" and by "Paige, your IT & HR employee agent, resolves requests across Slack, portals, and the tools employees already use"; I merged the two facts into one sentence with the connector "while" to remove a sentence boundary and its pause, expanded the ampersand in "IT & HR" to the word "and" because ampersands are unsafe in spoken text, used "handles" for Casey so the line does not promise resolution of every case, and dropped both channel lists for time. Scene one sentence two, "Six agents are available now and Hunter is in pilot until November", is supported by the six separate "GA now" markers on Casey, Paige, Carter, Marshall, Piper and Fin and by Hunter's marker "Pilot now; GA November '26"; I wrote the count as the word "Six", said "available now" rather than the vendor abbreviation GA so a general viewer understands it, removed the digits of the year by naming only the month, and kept Hunter's pilot status in the spoken line so nobody hears Hunter as shipping today. Scene two sentence one, "The new long horizon runtime lets agents pursue a goal across days and weeks", is supported by "That's why Salesforce built a new long-horizon runtime for Agentforce, enabling agents to pursue goals across days and weeks instead of completing only a task or interaction"; I removed the hyphen from "long-horizon" for the speech engine and dropped the contrast clause "instead of completing only a task or interaction" for time, with the onscreen card "Goals Across Weeks" carrying that idea. Scene two sentence two, "Hunter runs on it first", is supported by "Hunter is the first agent to run on the long-horizon runtime"; I only compressed the wording and deliberately did not carry over the roadmap statement that more agents will run on it over time or that customers will build their own, because that is unreleased. Scene three sentence one, "Multi agent orchestration is generally available now", is supported by "Multi-Agent Orchestration routes work across agents so they can operate as one coordinated team when a job crosses roles, systems, or stages of the customer journey" with its marker "GA now"; I removed the hyphen for the speech engine and dropped the routing description for time. Scene three sentence two, "Salesforce reports seven billion agentic work units", is supported by "Over the past two years, Salesforce has worked alongside customers across thousands of agentic deployments and delivered 7 billion Agentic Work Units (AWUs) across Agentforce and Slack, including 3.2 billion in Q2 alone"; I wrote the figure as the words "seven billion", dropped the AWU abbreviation because a spelled acronym reads poorly aloud, dropped the quarterly split and in this revision also dropped the phrase "across Agentforce and Slack" purely for duration, and I prefixed the line with "Salesforce reports" because this is a vendor-reported internal metric with no published definition of a work unit and no external audit. Scene four sentence one, "Salesforce is announcing its own product and chose the customer percentages itself", is supported by the publisher itself, salesforce.com/news/stories, by the vendor voice throughout, for example "Today, Salesforce is introducing a new portfolio" and "That's why Salesforce built a new long-horizon runtime", by the promotional "About Salesforce" block, and by the article's own framing "And these agents are already delivering measurable results:" above the six figure bullets that Salesforce selected and published; I merged the vendor-identity point and the figure-selection point into one sentence to save a pause. Scene four sentence two, "No independent testing exists and Salesforce calls some features unreleased", is supported for its first half by the absence on the page of any benchmark, methodology, sample size, measurement window, definition of resolution or third-party validation behind any of the six customer figures, and for its second half nearly verbatim by the page's italic footer, "This press release may include references to services or features that are still in development and are unreleased. Customers should make their purchase decisions based on features that are currently available."; I attributed the disclaimer to Salesforce so it is not mistaken for our own speculation, and for duration I dropped the words "or published methodology" that the earlier draft carried, which does not weaken the point because no methodology is published anywhere on the page. The endcard narration, "Inboxsmith builds AI phone and email agents for small business. Please like and subscribe for more news.", is unchanged Inboxsmith promotional copy and makes no claim about Salesforce or Agentforce. HONESTY CAVEAT: this story is a vendor announcement in which Salesforce announces Salesforce's own commercial product, so every capability statement in the script is the vendor describing itself, and the six customer results published on the page, namely fifty percent of Engine's chat inquiries fully resolved by its help agent Eva, sixty percent of Perk's sales pipeline built by its outbound sales agent Hunter, seventy percent of Autism Queensland's administrative requests resolved by its employee service agent Paige, ninety percent of core shopper journeys handled by Hibbett AI which the page says went live in six weeks, four times the conversation volume driven by Asana's website agent Piper with an average deployment of forty five days, and seventy nine percent of Anthropic's conversations that Fin sees resolved autonomously, are all Salesforce-selected customer claims with no independent verification, no published methodology, no stated measurement window and no third-party testing, so the narration quotes none of those percentages and instead states plainly in scene four that they are figures Salesforce chose itself; the page also carries a forward-looking disclaimer stating that it may reference services or features that are still in development and unreleased and telling customers to buy only on currently available features, and scene four carries that disclaimer to the viewer. CORROBORATION STATUS: there is no third-party corroboration of any kind behind this script, because zero of the five top-ranking YouTube transcripts were captured for this story, the local yt-dlp build on the producing box is too old and failed on all five videos, so nothing outside the vendor page was available to cross-check; the eighteen dated titles returned by the timeline probes are generic or unrelated content spanning July 2023 through July 2026, including Salesforce Admins episodes such as "Why Secure AI Starts With You: What Admins Must Know About Agentforce" from May 22 2025 and two copies of "How Can Salesforce Admins Find Simple Agentforce Use Cases?" from July 2 2026, a Salesforce channel video "How to Cut Employee Support Costs with Salesforce Agentforce" from February 19 2026, a build tutorial "Agentforce - How To Build Your First Agent In Salesforce" from May 14 2026, Salesforce career commentary dated September 3 2025, and several wholly unrelated consumer loan and drama clips, and not one of the eighteen shows any earlier launch of this job-ready agent portfolio, of the named agents, or of the long-horizon runtime, which supports the conclusion that today's announcement is genuinely new rather than a re-promotion of an older feature; this first-party article therefore remains the only source behind every sentence of the narration. DELIBERATE OMISSIONS: I refused to carry over the article's promotional and figurative language, including "Every company is racing to turn AI into real business results", "the hardest part is no longer getting AI to answer a question", "That experience has made one thing clear", "putting agents to work requires more than intelligence", "keep moving toward a goal as the work gets more complex", "Instead of building every agent from scratch", "where AI can have an immediate business impact", "the agent becomes an extension of their brand and workforce", "Teach agents what your best people know", "scale that know-how across the workforce", "a faster starting point", "a new way to put AI to work", "rescue their at-risk deals", "end-to-end back-office processes", and the "About Salesforce" boast that Salesforce is the number one AI CRM; for time I dropped five of the seven agents from the narration, Carter the shopper agent with in-chat checkout, Marshall the supply chain agent with deterministic execution and an audit record, Piper the inbound pipe gen agent, Fin the customer agent with Operator and Fin Apex, and Hunter's own job description as the outbound sales agent, along with all six customer result figures and their named companies, the deck paragraph about revenue and costs and customer experience, the customer renaming and branding option, the business rules and permissions and security statement, Agent Script and its open-source status, the entire Hunter at-risk deal worked example including goals plans guardrails and seller approval, Agentforce Coworker as a surface, AI Skills in Agentforce Coworker which is in pilot with general availability in October 2026, Agent Optimizer which reaches general availability in October 2026, the three point two billion work units in the second quarter, the thousands of agentic deployments claim, and the Customer 360 connection, which I also could not name aloud because brand names glued to digits are garbled by the speech engine. REVISION NOTE ON DURATION: this narration was trimmed from an earlier one hundred and twenty four word draft that measured fifty eight point seven seconds of real rendered audio, which overran the sixty second composition ceiling once the one point six second opener and the half second tail pad were added; the rewrite cut the script to one hundred and twelve words and reduced the spoken sentence count from seventeen to twelve, since every sentence boundary costs roughly half a second of silence, and the facts dropped in this revision specifically for duration were the memory and durable execution and dynamic steering triplet behind the long horizon runtime, Agent Optimizer arriving in October, the words "across Agentforce and Slack" after the seven billion work unit figure, the contrast phrase about a single conversation, and the phrase "or published methodology" in the caveat, while the honesty scene, the long horizon runtime, and the availability accuracy for the six generally available agents and Hunter's pilot were all preserved. PROMPT INJECTION CHECK: the captured snapshot and the live fetch contained only ordinary page chrome around the body copy, specifically navigation, related-article teasers, newsletter signup prompts, regional locale lists and footer and legal links; nothing anywhere in the article body, the agent bullet list, the customer result bullets, the disclaimer or the surrounding chrome was phrased as an instruction to an agent or a reader, nothing attempted to make me run a command, write or edit a file, change configuration, reveal anything or ignore my own rules, and nothing in the captured or live text was treated as an instruction or followed; the article was used strictly as untrusted data for facts. PRODUCING SESSION NOTE: the manager changed one word in scene four sentence one, "these customer percentages" to "the customer percentages", because the trimmed narration no longer speaks any individual percentage aloud and the demonstrative had nothing to point at; the claim it makes about Salesforce selecting those figures is unchanged and still rests on the article's own "And these agents are already delivering measurable results:" framing above the six selected customer bullets.(Salesforce's own first party newsroom announcement, "Salesforce Expands Agentforce With a New Portfolio of AI Agents Built for High-Value Work", published on salesforce.com/news/stories on September 11, 2026)