Platform reference · Freshdesk · Freshworks · ADA Title III · WCAG 2.1 AA · SaaS support training · Agent enablement

Freshdesk captions: Help Center video, agent training, and ADA compliance for SaaS support teams

Freshdesk is Freshworks' customer support platform, used by tens of thousands of SaaS, technology, healthcare, and e-commerce companies to manage customer tickets, knowledge base articles, and support-agent workflows. Like its chief competitor Zendesk, Freshdesk operates across two distinct captioning domains that each carry independent compliance obligations. The first is the Freshdesk Help Center Portal — the public-facing knowledge base where companies publish customer-visible articles, product walkthroughs, FAQ pages, and video tutorials. Help Center video is accessible to any visitor, which makes it ADA Title III "place of public accommodation" content under 42 U.S.C. § 12181: US businesses with public Help Centers that contain video must caption that video to WCAG 2.1 AA SC 1.2.2 accuracy. The second domain is internal agent training — Freshdesk onboarding walkthroughs, Freddy AI enablement videos, QA calibration session recordings, Freshworks Suite product-update training — which is internal L&D content subject to employer accommodation obligations under ADA Title I and state disability laws for hearing-impaired support agents.

TL;DR

Freshdesk Help Center video is typically embedded from YouTube, Vimeo, or Loom — the caption track on any embedded video comes from the originating platform (YouTube Studio, Vimeo, Loom). To caption Help Center video, upload a corrected SRT or VTT to the YouTube or Vimeo video, and the corrected track displays in the embedded player on the Freshdesk Help Center article page. Internal Freshdesk agent training video hosted on Loom, Vimeo, Wistia, or Microsoft Stream uses the same source-platform caption workflow. Freshdesk's Help Center portal does not have a built-in caption-generation or SRT-upload UI for embedded video — captioning happens at the source platform, not at the Freshdesk layer. Compliance: Help Center video needs WCAG 2.1 AA-accurate captions under ADA Title III. Agent training video needs WCAG 2.1 AA-accurate captions for hearing-impaired agents under ADA Title I and state law. The Freshworks Suite's dense proper-noun vocabulary — Freshdesk product names, Freddy AI sub-product names, Freshdesk workflow automation terminology (Dispatch'r, Observer, Supervisor, Scenario Automation), and the customer's own product vocabulary discussed in agent training — is a systematic auto-caption failure point that glossary-biased captioning is specifically designed to address.

Freshdesk Help Center — caption surface and ADA Title III obligation

Freshdesk's Help Center (sometimes called the Freshdesk Portal in older documentation, and the Solutions tab in the Freshdesk agent interface) is the customer-visible knowledge base side of Freshdesk. It sits at a URL such as support.company.com or company.freshdesk.com/support/home and is typically accessible to any member of the public without authentication, serving as the company's primary customer self-service destination. The video content that appears in Help Center articles falls into several categories.

Embedded YouTube and Vimeo video in Help Center articles

The majority of video in Freshdesk Help Center articles is embedded from YouTube or Vimeo. When a Help Center author inserts a YouTube or Vimeo URL into a Freshdesk Solutions article, Freshdesk renders an iframe embed of that video inline in the article page. The caption track that displays in that embedded player is the caption track configured on the originating YouTube or Vimeo video — not anything managed at the Freshdesk layer. The practical workflow for captioning Help Center-embedded video:

Identify the YouTube or Vimeo video that is embedded in the Help Center article (click through to the source video to find the platform and video ID).
Produce a corrected SRT or VTT caption file using glossary-biased transcription that accurately captures all product names, feature names, and menu paths discussed in the walkthrough narration.
Upload the corrected SRT/VTT to YouTube Studio (YouTube Studio → Subtitles → the specific video → Add language → Upload file) or to the Vimeo video settings (Vimeo account → Video → Settings → Distribution → Subtitles).
The corrected caption track propagates automatically to all embeds of that video, including the Freshdesk Help Center article embed. No change to the Freshdesk article is required — the corrected captions appear in the embedded player the next time a visitor loads the article.

YouTube's auto-generated captions and Vimeo's auto-generated captions do not satisfy WCAG 2.1 AA SC 1.2.2 for Help Center content that discusses product-specific terminology, support workflows, or Freshdesk configuration details. A video walkthrough of how to set up a Dispatch'r rule, how to configure SLA policies, or how to use Freddy Self Service bots will have systematic caption errors on every Freshdesk-specific term. The corrected-SRT approach replaces the auto-generated track entirely.

Loom-embedded Help Center video

Many Freshdesk Help Center authors use Loom to record quick screen-capture walkthroughs and embed the Loom link directly into Freshdesk Solutions articles. Loom generates an auto-transcript for every recording, which serves as the default caption track in the Loom player. Like YouTube auto-captions, Loom's auto-transcript fails systematically on Freshdesk-specific vocabulary — Dispatch'r becomes "dispatcher," Freddy Copilot becomes "Freddie Co-Pilot" or "Freddie co-pilot," and Freshchat becomes "fresh chat" (two words, lowercased). On Loom Business and Enterprise plans, the Replace Transcript feature allows you to upload a corrected VTT, replacing the auto-generated transcript with accurate captions. See Loom captions for the full workflow.

ADA Title III for public-facing Freshdesk Help Center video

ADA Title III (42 U.S.C. § 12181 et seq.) prohibits discrimination by "places of public accommodation" on the basis of disability. US courts — most prominently in the Ninth Circuit and the Eleventh Circuit — and the DOJ have consistently held that commercial websites providing goods or services to the public are places of public accommodation covered by Title III. A Freshdesk Help Center hosted at support.company.com or company.freshdesk.com, accessible to any member of the public including visitors with hearing disabilities, is a public accommodation for Title III purposes. Video in that Help Center that lacks accurate captions denies visitors with hearing disabilities equal access to the self-service information the video conveys — the same information that sighted users with full hearing can access.

The practical Title III exposure framework for SaaS companies with Freshdesk Help Centers:

California Unruh Civil Rights Act

California's Unruh Civil Rights Act (Cal. Civ. Code § 51) provides $4,000 per incident in statutory damages, plus actual damages, plus attorneys' fees, for violations of ADA Title III. California-resident visitors who encounter inaccessible video in a Freshdesk Help Center — and who have a disability that makes the absence of captions a barrier to access — have a viable Unruh Act claim. California has the largest concentration of SaaS companies and SaaS customers in the US, meaning Unruh Act exposure is a material risk for essentially every SaaS company with a Freshdesk Help Center. California plaintiffs' attorneys have brought serial Unruh Act web-accessibility litigation against SaaS companies; Help Center video without accurate captions is an identifiable target.

New York Human Rights Law and New York City Human Rights Law

New York State's Human Rights Law (Executive Law § 296) extends public-accommodation protections to commercial websites accessible to New York residents. The New York City Human Rights Law adds additional teeth for New York City-area users, with a lower burden of proof for plaintiffs than the federal standard. A Freshdesk Help Center that serves New York users — which, practically speaking, means any Help Center with a US user base — is potentially covered. NYCHRL in particular does not require a plaintiff to demonstrate a nexus to a physical location, making it one of the broadest disability-access statutes in the US for web-accessibility claims.

New Jersey Law Against Discrimination

New Jersey's Law Against Discrimination (N.J.S.A. § 10:5-1 et seq.) covers "places of public accommodation" broadly, and New Jersey courts have applied LAD to commercial websites. New Jersey has a substantial concentration of pharmaceutical, financial services, and technology companies — many of whom are Freshdesk customers — and their Help Centers are potentially within LAD scope for New Jersey users.

DOJ Title III settlement pattern

The Department of Justice has settled multiple Title III web-accessibility cases that include video captioning requirements. The settlement pattern is consistent: WCAG 2.1 AA remediation of all video content on public-facing websites and digital properties, an annual third-party accessibility audit, an ongoing user feedback mechanism for reporting accessibility barriers, and a monitoring period of two to three years. Companies that self-remediate before receiving a demand letter or complaint are in a materially better position than those who remediate under settlement terms.

Customer self-service equity and Help Center deflection metrics

Beyond the legal compliance argument, accurate Help Center video captions directly affect the product metric that support-ops teams care most about: self-service deflection rate. A Help Center video tutorial that explains how to complete a task — connecting an integration, configuring a billing setting, understanding a pricing tier — deflects a support ticket for every visitor who successfully self-serves. Visitors with hearing disabilities who cannot access uncaptioned video cannot self-serve via that video, and their interaction either becomes a support ticket (costing the company per-contact support cost) or a customer who fails to find the answer and churns. Accessible Help Center video expands the population of visitors for whom self-service works. Captioning is not only a compliance investment; it is a deflection-rate optimisation.

Internal agent training — Freshdesk L&D content types and employer accommodation

The second captioning domain for Freshdesk-using companies is entirely internal: the training and enablement video that support-operations teams, L&D teams, and Freshdesk administrators produce for support agents. This internal content spans several distinct categories, each with its own vocabulary density and compliance exposure.

Freshdesk onboarding training

New support agent onboarding for a Freshdesk-using company is inevitably a screen-capture walkthrough of the company's specific Freshdesk instance: how the ticket inbox is organised, how to use ticket views (both default views and the company's custom views), how to apply macros, how to use canned responses, how escalation paths are configured, how SLA policies work in this instance, how CSAT surveys are configured, how the company's product knowledge base is organised in the Solutions tab, and how ticket groups and assignment work. This content is specific to the company's Freshdesk configuration — it cannot be replaced by Freshdesk's own onboarding documentation because it describes this company's specific setup.

Onboarding video is typically recorded with Loom (screen capture with voiceover), Camtasia (screen-capture editing tool), or occasionally as a recorded Zoom or Teams walkthrough session. All of these recording methods produce auto-transcripts that fail on Freshdesk-specific vocabulary. A new hearing-impaired agent who relies on the onboarding video captions to understand the training content will encounter systematic caption errors on the most operationally critical terms: the names of the automation rules, the names of the ticket views, the names of the escalation groups, and the name of the product they are being trained to support.

Freshdesk product update training

Freshworks releases regular product updates to Freshdesk, including new Freddy AI capabilities, new channel integrations (WhatsApp Business API, Google Business Messages, Apple Messages for Business), UI redesigns to the Agent Workspace, new reporting and analytics features in Freshdesk's analytics module, and new automation capabilities. Support-ops administrators typically produce an internal video announcement and walkthrough when a significant product update is released, explaining to agents what changed and how to use the new feature. These update videos are definitionally dense with Freshdesk product names and UI terminology — they are specifically about new Freshdesk capabilities — and auto-transcript accuracy on these videos is particularly poor.

Freddy AI enablement training

Freshworks' AI brand is Freddy AI, which encompasses three distinct sub-products that agents and administrators need training to use effectively:

Freddy Self Service — the bot automation platform that handles incoming customer interactions before escalating to a human agent. Freddy Self Service training covers bot flow design, escalation trigger configuration, intent mapping, and how to review and improve bot performance over time.
Freddy Copilot — the in-context AI assistant for human agents in the Freshdesk Agent Workspace, providing suggested responses, article suggestions, ticket summaries, and tone adjustment tools. Freddy Copilot training covers how to use Copilot suggestions effectively, how to review and edit AI-generated responses before sending, and how Copilot interacts with the company's knowledge base.
Freddy Insights — the business intelligence and analytics layer of Freshworks' AI suite, providing predictive analytics, trend detection, and CSAT improvement recommendations. Freddy Insights training is typically addressed to support managers and QA leads rather than front-line agents.

Freddy AI training content is among the most vocabulary-dense training video a Freshdesk-using company produces, because it combines Freshdesk operational vocabulary with Freshworks AI brand names, AI-specific technical terminology, and the company's own product vocabulary (which is what Freddy Self Service bots and Freddy Copilot knowledge base articles are about). Auto-transcript accuracy for Freddy AI training video is systematically poor on all three product names ("Freddy" is often transcribed as "Freddie," "Freddy" or occasionally "ready"), all AI-specific terms, and all company-product terms.

QA calibration session recordings

Support quality assurance teams at Freshdesk-using companies regularly hold calibration sessions — typically a group video call (Zoom, Teams, Google Meet) where the QA team reviews a sample of recent tickets, discusses how the QA scoring rubric should be applied, and aligns on edge cases and grey areas. These calibration sessions are recorded and shared as training reference video for agents and QA reviewers who were not present. The recordings include all the calibration discussion — which references specific ticket IDs, specific CSAT scores, specific SLA breach scenarios, specific Freshdesk features used or not used, specific product functionality that was or was not correctly explained in the agent's response. This is exactly the type of multi-layered vocabulary content where auto-transcript accuracy is poorest.

Product deep-dive training for support agents

Support agents need to understand the product they support deeply enough to help customers and to correctly distinguish between expected product behaviour and actual bugs. Product deep-dive training for support teams is often produced by the product team, customer success team, or solutions engineering team as a recorded walkthrough: here is what the new feature does, here is what edge cases agents should know about, here is when to escalate to engineering vs. resolve the ticket. This content combines Freshdesk vocabulary (because it is being shared via Freshdesk's internal knowledge base or via a Loom link in a Freshdesk Solutions article) with dense company-product vocabulary. It is arguably the content category where a company glossary has the highest per-term impact, because every product term in the walkthrough needs to be transcribed correctly for the training to be useful to hearing-impaired agents.

Employer accommodation obligations

Support agents with hearing disabilities work at SaaS customer support teams. The employer obligation to provide accessible training video to hearing-impaired support agents flows from several independent legal sources:

ADA Title I (federal). 42 U.S.C. § 12112 prohibits employment discrimination against qualified individuals with disabilities by employers with 15 or more employees. Providing accessible training materials — including video training that hearing-impaired employees can access via accurate captions — is a core component of the reasonable accommodation obligation. An employer who deploys Loom-recorded Freshdesk onboarding training with auto-generated captions that fail systematically on every Freshdesk-specific term is not providing a reasonable accommodation to a hearing-impaired new-hire who must rely on those captions to learn their role.
California FEHA. California's Fair Employment and Housing Act applies to employers with 5 or more employees — a lower threshold than federal ADA Title I. California's concentration of SaaS employers means most Freshdesk-using SaaS companies with California-based support teams are covered.
New York Human Rights Law. New York HRL applies to employers with 4 or more employees. New York City HRL applies to employers with 4 or more employees and has a broader reasonable accommodation standard than the state or federal baseline.
New Jersey Law Against Discrimination. NJ LAD applies to all employers regardless of size — there is no employee-count threshold. Every NJ-based employer with hearing-impaired employees has a reasonable accommodation obligation under LAD.

The Freshworks vocabulary failure mode

Freshdesk training video has a compound proper-noun failure mode that is unusually severe even among CRM and support platform training video: the Freshworks suite vocabulary (product names, sub-product names, workflow automation terminology) combined with the company's product vocabulary creates two dense layers of proper nouns that generic STT systems have not been trained on. The failure mode is systematic — it is not random word-level errors but consistent mis-transcription of every occurrence of specific terms — and it compounds across a training video to produce captions that are misleading, not merely imprecise.

Freshworks suite product names

The Freshworks product suite includes more than a dozen distinct products and sub-products, all carrying the "Fresh-" prefix plus one additional proper noun. Generic speech-to-text systems that encounter the prefix "fresh" followed by a business-context term will frequently de-capitalise, split into two words, or substitute a more common word:

Freshdesk — the customer support ticketing platform. Spoken quickly, the "desk" portion is usually transcribed correctly, but capitalisation is often lost ("fresh desk" as two lowercase words), and in contexts where "Fresh Desk" appears alongside product names the parsing fails.
Freshservice — the ITSM (IT Service Management) platform. "Fresh service" as two words is the most common mis-transcription; "freshest service" appears in some STT outputs when the speaker has an accent or speaking rhythm where "fresh" ends on a fricative that bleeds into "service."
Freshsales and Freshsales Suite vs Freshsales Classic — the CRM product and its two edition tiers. The Suite / Classic distinction is operationally important in training that explains which Freshsales version the company uses and why — and it is exactly the type of noun-modifier combination that auto-STT consistently gets wrong. "Freshsales Suite" becomes "fresh sales suite" (three lowercase words), which in a training context tells a hearing-impaired agent nothing about whether this is a product name or a general description.
Freshchat — the messaging and live-chat platform. Frequently transcribed as "fresh chat" (two words) or "FreshChat" with incorrect internal capitalisation.
Freshcaller — the cloud telephony / call center platform. "Fresh caller" (two words) is the most common failure mode. In training contexts where a support team uses Freshcaller for inbound phone support alongside Freshdesk for tickets, the two product names appearing in the same sentence produce compound errors.
Freshmarketer — the marketing automation platform. "Fresh marketer" (two words) is the standard mis-transcription.
Freshteam — the HR software and applicant tracking system. "Fresh team" (two words) or "freshman" (phonetically similar) are common failures.
Freshping — the website uptime monitoring tool. Less frequently appears in support team training, but mentioned in product-update and integration-architecture training.

Freddy AI sub-product vocabulary

The Freddy AI brand introduces its own layer of transcription failures on top of the platform product names. "Freddy" is a proper noun used as a product name — generic STT systems treat it as an unusual spelling of "Freddie" and frequently produce "Freddie," "ready," or "Freddy AI" transcribed as "Freddie AI." The three Freddy sub-products each compound the failure:

Freddy Self Service — three words that form a product name. "Freddie self service," "Freddy self-service" (hyphenated when it should not be), and "ready self service" are all attested STT outputs for this product name.
Freddy Copilot — "Freddie Copilot," "Freddy co-pilot," "Freddie co pilot," and "ready copilot" are all common failures. "Copilot" is now also a Microsoft product name (Microsoft 365 Copilot), which means some STT systems trained on more recent data have learned "Copilot" as a capitalised product name but attribute it to Microsoft context rather than Freshworks.
Freddy Insights — "Freddie Insights," "Freddy insights" (lowercase "insights"), and "ready insights" are the standard failure patterns. In training video that discusses "what Freddy Insights tells you about your CSAT trend," "ready insights" produces a nonsensical sentence that offers no information to a hearing-impaired viewer.

The Neo platform

Freshworks' underlying developer platform is called the Neo platform. In advanced administrator training and developer training for Freshdesk marketplace apps, "Neo" appears as a standalone product name. The word "neo" spoken quickly and in the context of "Neo platform" or "Neo developer console" is almost indistinguishable from the common word "no" in natural conversational speech — particularly when the speaker says "the Neo platform" with a short vowel on "Neo." STT systems consistently transcribe "Neo platform" as "no platform" in audio where the speaker's delivery is conversational rather than deliberately enunciated. "No platform" is a nonsensical phrase in the training context, and a hearing-impaired viewer following along will be confused rather than informed.

The OmniChannel unified inbox model

Freshdesk's unified-inbox capability — where email, phone (Freshcaller), chat (Freshchat), social media, and the Help Center portal all funnel into a single Freshdesk ticket queue — is marketed and trained under the OmniChannel label. "OmniChannel" is one compound word (with internal capitalisation) in Freshdesk's product terminology. STT systems produce "omni channel" (two words), "omnichannel" (no internal capitalisation, which is actually more standard English but incorrect for the Freshdesk-specific branded term), and occasionally "omni-channel" (hyphenated). In training contexts where OmniChannel is being distinguished from single-channel email support or from Freshdesk's older pre-OmniChannel ticket model, the transcription variation creates confusion.

Freshdesk configuration and workflow automation vocabulary

Freshdesk's business rules engine — the collective name for the suite of automation tools — has one of the most distinctive proper-noun vocabularies in any customer support platform. These terms are specifically Freshdesk-branded names for automation types that exist in some form in every customer support platform but are named differently in each one:

Dispatch'r — Freshdesk's ticket-creation automation (fires on new ticket events). The apostrophe in the name is a deliberate stylistic choice by Freshworks; the spoken form is "dispatcher" — an entirely ordinary English word. Generic STT will always transcribe "dispatcher" as "dispatcher," and the listener/reader will see "dispatcher" rather than "Dispatch'r." In a training context where an administrator is explaining how Dispatch'r rules work, every instance of "Dispatch'r" in the transcript reads as "dispatcher" — correct phonetically but incorrect as a Freshdesk product term, which matters when the agent needs to navigate the Freshdesk admin panel to the "Dispatch'r" settings page and cannot find a menu item labelled "dispatcher."
Observer — Freshdesk's update-triggered automation (fires when a ticket is updated). "Observer" is a common English word, so it is transcribed correctly by generic STT, but without capitalisation and without the context that it is a specific Freshdesk automation type rather than a person observing something.
Supervisor — Freshdesk's time-based automation (fires on a schedule, checking tickets for conditions like "has not been responded to in 4 business hours"). Like Observer, "supervisor" is a common word transcribed correctly but without product-name context.
Scenario Automation — Freshdesk's multi-step agent-triggered automation sequence. "Scenario automation" is transcribed correctly by generic STT but without the capital-S capital-A that identifies it as a specific Freshdesk feature type.
Round Robin assignment — Freshdesk's sequential agent assignment method. "Round robin" is a common phrase and usually transcribed correctly, but "Round Robin assignment" as a Freshdesk configuration option benefits from capitalisation in captions for clarity.
Load Balancing assignment — Freshdesk's workload-based agent assignment method. Similar to Round Robin: correct transcription of the words, but product-context capitalisation matters.
Skill-based routing — assignment based on agent skill tags. Usually handled correctly by generic STT.
Canned Responses — Freshdesk's pre-written agent reply templates. "Canned responses" as two lowercase words is the standard auto-STT output; "Canned Responses" with product-name capitalisation is what Freshdesk actually calls this feature in its admin panel.
Satisfaction Surveys (CSAT) — the customer satisfaction survey feature in Freshdesk. "CSAT" spoken aloud as "C-SAT" or "see-sat" is transcribed variously as "C-SAT," "CSAT," "see-sat," or "sea sat" depending on STT system and speaker pronunciation pattern.
Freshdesk Arcade — the gamification feature with points, trophies, and leaderboards for agents. "Arcade" is a common word; "Freshdesk Arcade" is the product name. "Fresh desk arcade" (three lowercase words) is the typical auto-STT failure.
Parent–Child Tickets — Freshdesk's complex-ticket decomposition model where a parent ticket coordinates multiple child tickets. "Parent-child tickets" is usually transcribed correctly but without the specific hyphenation Freshdesk uses.
Freddy Suggestions — the AI-surfaced knowledge base article suggestions that appear in the Freshdesk agent reply window when composing a response. "Freddy suggestions" has the same "Freddy/Freddie" failure mode as all other Freddy sub-features.

Freshdesk metrics and reporting vocabulary

Freshdesk training video for QA calibration, performance reviews, and analytics training references a dense set of support metrics abbreviations and terminology. Many of these are industry-standard abbreviations that generic STT handles inconsistently:

FRT / First Response Time — "FRT" spoken as an acronym is transcribed as "F-R-T," "effort," or lost entirely. "First Response Time" as three words is usually transcribed correctly.
TTFR (Time to First Reply) — spoken as "T-T-F-R" it is rarely transcribed correctly; spoken as "time to first reply" it is fine.
FCR (First Contact Resolution Rate) — "FCR" spoken as an acronym has the same problem; "first contact resolution" is usually fine.
AHT (Average Handling Time) — "AHT" is often transcribed as "eight" or "ate" in STT outputs when spoken quickly in context.
SLA Breach Rate — "SLA" spoken as three letters is inconsistently transcribed; "SLA breach rate" is a critical training metric whose correct transcription matters for hearing-impaired agents following a QA calibration session.

Freshworks integration vocabulary

Freshdesk training video for companies that have configured third-party integrations introduces an additional vocabulary layer — the third-party application and object names that appear in screen-capture walkthroughs of the integration workflow:

Jira (Atlassian's issue tracker, integrated with Freshdesk for bug escalation). "Jira" is a foreign proper noun that STT systems trained on English often transcribe as "gira," "Julia," or simply "the issue tracker." In training that walks through how to create a Jira issue from a Freshdesk ticket, every instance of "Jira" needs to be correctly transcribed for the caption to be useful.
Slack (Freshdesk Slack app for ticket notifications). "Slack" is usually transcribed correctly as a word, but in contexts where the speaker discusses Slack slash commands, Slack webhook URLs, or Slack channel naming conventions, the technical vocabulary compounds the failure.
Shopify (Freshdesk Shopify integration for e-commerce support teams). "Shopify" is usually transcribed correctly by modern STT systems that have seen it in training data, but Shopify-specific order and product terminology that appears in screen-capture walkthroughs of Shopify order data in Freshdesk — "Fulfillment," "Refund," "Draft Order," "Shopify Payments," "Shop Pay" — requires a glossary for accurate transcription.
Salesforce (Freshdesk–Salesforce CRM integration). "Salesforce" is usually transcribed correctly. Salesforce object names that appear in integration training — Lead, Opportunity, Account, Contact, Campaign, Case — are common English words that auto-STT produces correctly as lowercase common nouns rather than capitalised Salesforce object names, which matters for heading-level comprehension of the training.
Stripe (payment platform integrated with Freshdesk for billing support contexts). "Stripe" as both a product name (Stripe, Inc.) and a common noun (a stripe of color) is a disambiguation challenge for STT systems, particularly in audio contexts where the speaker's cadence does not clearly signal a proper noun.

Caption workflow for Freshdesk-adjacent video platforms

Support-operations and L&D teams at Freshdesk-using companies produce training video on a mix of tools. The captioning workflow differs by platform; none of the correction happens at the Freshdesk layer. Below is the practical workflow for each platform in the Freshdesk-adjacent video ecosystem.

YouTube-embedded Help Center video

For YouTube-hosted video embedded in Freshdesk Help Center articles, the caption correction workflow is entirely in YouTube Studio:

In YouTube Studio, navigate to the video → Subtitles.
Select the language track. If YouTube has already auto-generated an English track, click the three-dot menu next to it and choose "Edit on Classic Studio" or "Edit" to review the auto-generated content.
To upload a corrected SRT: click "Add language" if no English track exists, or use the "Upload file" option to replace the existing auto-generated track with a corrected SRT file.
Publish the updated caption track. The corrected captions propagate to all YouTube embeds of the video, including the Freshdesk Help Center article embed, typically within a few minutes.

For WCAG 2.1 AA compliance, the corrected SRT should include accurate transcription of all spoken audio, synchronised timing, speaker identification where multiple speakers are present, and identification of non-speech audio information that is relevant to understanding the content. Freshdesk-specific vocabulary — Dispatch'r, Freddy Copilot, Freshchat, Round Robin, Canned Responses — must all be transcribed accurately in the SRT file. A company glossary loaded into GlossCap's Whisper-decoding pipeline handles this systematic vocabulary correction without manual review of every caption line.

Vimeo-embedded Help Center and training video

For Vimeo-hosted video, caption upload is in the Vimeo video settings. Vimeo supports SRT and VTT format uploads. The corrected track displays in the Vimeo player embedded in Freshdesk Help Center articles and in any Loom-style internal video sharing context. See Vimeo captions for training videos for the full upload workflow, format requirements, and the Vimeo API approach for batch-uploading corrected caption files across a large library of Freshdesk training videos.

Loom (primary async-video tool for SaaS support teams)

Loom is the dominant async-video tool for SaaS support teams. Freshdesk walkthroughs, Freddy AI demos, quick "here is how to use this new feature" videos, and QA calibration pre-work are routinely recorded in Loom by Freshdesk administrators and support-ops leads and shared as Loom links embedded in Freshdesk Solutions articles or Slack messages. Loom generates an auto-transcript that serves as the default caption track. On Loom Business and Enterprise plans, the Replace Transcript feature (accessible via the three-dot menu on the video editing screen) allows upload of a corrected VTT file. The corrected VTT replaces Loom's auto-transcript entirely and serves as the caption source for all subsequent views of that Loom video, including views from embeds. See Loom captions for the complete Replace Transcript workflow, VTT formatting requirements for Loom, and how to handle Loom's 1,000-character limit on transcript segment length.

Wistia (premium video for customer education)

Some Freshdesk-using companies host their customer-education video library on Wistia rather than YouTube, for greater control over player branding, privacy settings, and viewer analytics. Wistia supports SRT caption upload for every video (not gated on a plan tier, unlike Loom). The corrected SRT displays in the Wistia player embedded in Freshdesk Help Center articles. Wistia's Channels feature — a curated video collection page — is used by some companies as a standalone customer-education academy alongside the Freshdesk Help Center text articles. Caption upload in Wistia is through Video Settings → Captions → Upload a caption file. See Wistia captions for the full workflow including Wistia's language-code requirements (ISO 639-2/T three-letter codes — eng not en).

Camtasia (screen-capture training walkthroughs)

Camtasia is the dedicated screen-recording and video-editing tool used by L&D teams that need more production quality than Loom provides. Freshdesk onboarding walkthroughs and Freshdesk configuration training are often produced in Camtasia when the L&D team wants to add call-outs, annotations, zoom effects, or quiz interactions alongside the screen capture. Camtasia can export SRT and can also burn captions into the video at export time. The SRT export from Camtasia should use GlossCap-corrected captions rather than Camtasia's built-in speech-to-text, because Camtasia's built-in STT has the same Freshdesk vocabulary failure mode as every other generic STT system. After Camtasia SRT export, upload the corrected SRT to the video hosting platform (YouTube, Vimeo, or Wistia) as described above. See Camtasia captions for the full workflow including how to handle Camtasia's SRT export timing format and how to prevent timing drift in long Camtasia screen-record sessions.

Microsoft Stream (for M365-tenant Freshdesk customers)

Companies on Microsoft 365 who also use Freshdesk may host internal Freshdesk training video in Microsoft Stream, particularly if the company's standard meeting-record and async-video workflow is Stream-based (via Teams meeting recording). Stream generates auto-transcripts via Azure Cognitive Services speech recognition, which has the same Freshdesk vocabulary failure mode as all other generic STT systems. Stream supports VTT upload as an alternative to the auto-generated transcript. The VTT file can be uploaded via the video page → Edit → Captions and subtitles → Upload. See Microsoft Stream captions for the full VTT upload workflow, the Stream-SharePoint integration context, and how Stream handles caption tracks for meeting recordings vs. uploaded video files.

A note on Freshdesk Help Center's own video capabilities

Freshdesk's Help Center itself has no built-in caption-generation or SRT/VTT-upload UI for video content. Freshdesk Solutions articles are a text-plus-embed platform — video is embedded from external platforms, not hosted natively by Freshdesk. If a video file is uploaded directly to a Freshdesk article as an attachment or via Freshdesk's built-in media uploader (rather than embedded from YouTube, Vimeo, or Loom), it will play in a basic HTML5 video element without caption support. The recommended approach for all Help Center video — for both caption management and player feature completeness — is to host the video on a captioning-capable platform and embed from there into the Freshdesk article.

See GlossCap pricing

Compliance framework for SaaS companies using Freshdesk

SaaS companies that use Freshdesk as their customer support platform face a two-layer compliance framework for video captioning: public-facing obligations (ADA Title III, state public accommodation laws) for Help Center video, and employer obligations (ADA Title I, state employment disability laws) for internal agent training video. The two layers are independent — satisfying one does not substitute for the other — but the technical captioning approach (glossary-biased SRT/VTT generation and source-platform upload) addresses both layers through the same workflow.

ADA Title III: public accommodation (42 U.S.C. § 12181)

Title III of the Americans with Disabilities Act prohibits discrimination by "public accommodations" — places, establishments, and now (per consistent judicial interpretation) websites that provide goods and services to the general public. A Freshdesk Help Center accessible to any visitor without authentication is a public accommodation. The obligation under Title III, as operationalized through WCAG 2.1 AA, is that video content on that Help Center must have accurate, synchronized captions (SC 1.2.2) that convey all audio information in the video. "Accurate" in WCAG 2.1 AA SC 1.2.2 means more than "speech-to-text output" — it means the captions must correctly represent what was said, including proper nouns, technical vocabulary, and specialized terminology.

DOJ Title III settlements have consistently required WCAG 2.1 AA compliance across all video on covered websites, annual accessibility audits, user feedback mechanisms, and multi-year monitoring. Companies subject to a Title III demand or complaint face remediation costs, settlement costs, and litigation costs that substantially exceed the cost of proactive captioning. The DOJ's track record of settlements — rather than providing a formal safe-harbor rule — means that companies cannot rely on "good faith" compliance as a complete defense if their Help Center video lacks accurate captions.

California Unruh Civil Rights Act (Cal. Civ. Code § 51)

California's Unruh Act provides $4,000 per incident in minimum statutory damages for Unruh Act violations (which incorporate ADA Title III violations by reference under Civ. Code § 51(f)). For a Freshdesk Help Center that contains multiple uncaptioned videos, each video that a California-resident visitor encounters without captions is potentially a separate incident. California plaintiffs' counsel has filed thousands of web-accessibility cases under Unruh, including cases targeting SaaS company Help Center video. The concentration of California residents among SaaS company user bases — Silicon Valley, Bay Area, Los Angeles, San Diego — makes Unruh Act exposure a primary financial risk for any SaaS company with uncaptioned Help Center video.

The Unruh Act's "actual damages" provision adds the concrete harm suffered by the plaintiff on top of the $4,000 statutory floor. Attorneys' fees under the private right of action make cases economically viable for plaintiffs' counsel even where actual damages are modest. The practical risk for a SaaS company with 10 uncaptioned Freshdesk Help Center videos is $40,000+ in per-incident exposure plus attorneys' fees in California alone.

New York Human Rights Law (Executive Law § 296)

New York State's Human Rights Law covers "places of public accommodation, resort or amusement" — a category New York courts have applied to commercial websites. The New York City Human Rights Law (Administrative Code § 8-101 et seq.) applies the same protection with a lower burden of proof for plaintiffs and a broader definition of what constitutes discrimination. For a Freshdesk Help Center accessible to New York-resident visitors (which again means: any Help Center with a US user base), uncaptioned video is a potential NYCHRL violation.

Unlike the Unruh Act, NYCHRL does not have a fixed per-incident statutory damages minimum; damages are determined by a jury or trier of fact. However, NYCHRL's broader standard for what constitutes discrimination (the "differently than other similarly situated persons" standard rather than a strict "disability" test) makes it easier for plaintiffs to prevail, and NYCHRL allows punitive damages in cases of willful discrimination.

New Jersey Law Against Discrimination (N.J.S.A. § 10:5-1)

New Jersey's Law Against Discrimination covers "places of public accommodation" and applies to commercial websites accessible to New Jersey residents. Unlike the federal ADA Title I, NJ LAD applies to all employers regardless of size — and NJ LAD's public accommodation protections cover both the customer-facing Help Center and, as an employment matter, the internal training video for NJ-based support agents. NJ LAD allows compensatory damages, punitive damages, and attorneys' fees.

WCAG 2.1 AA SC 1.2.2 as the technical compliance standard

WCAG 2.1 AA Success Criterion 1.2.2 ("Captions — Prerecorded") requires that "captions are provided for all prerecorded audio content in synchronized media, except when the media is a media alternative for text and is clearly labeled as such." The "synchronized" requirement means that caption text must appear in synchrony with the audio — not just a transcript, but a timed caption track. The accuracy requirement, while not quantified in a percentage-accuracy threshold in the WCAG text, has been interpreted in litigation and DOJ settlements as requiring captions that accurately convey all meaningful audio content including proper nouns, technical terms, and specialized vocabulary.

Auto-generated captions — YouTube auto-captions, Vimeo auto-captions, Loom auto-transcripts — have been found in web-accessibility cases to not satisfy SC 1.2.2 for technical or specialized content because their systematic errors on product names, terminology, and proper nouns mean that the captions do not accurately convey the audio. The corrected-caption approach — producing an accurate SRT via glossary-biased transcription and uploading it to the source platform — is the technically correct path to SC 1.2.2 compliance for Freshdesk training and Help Center video.

Practical remediation priority for Freshdesk captioning

For a SaaS company that has an existing library of uncaptioned or poorly-captioned Freshdesk Help Center and training video, a practical remediation priority framework:

Public Help Center video first. Public-facing video carries both Title III and Unruh Act exposure. Prioritize Help Center video that explains high-traffic product features, billing/account management, and integration setup — the topics most likely to generate accessibility complaints.
New employee onboarding training second. Onboarding video for hearing-impaired new agents is both a legal accommodation obligation and a practical inclusion issue. New hires who rely on captions are disadvantaged in onboarding before they have relationship capital to request an accommodation informally.
Freddy AI and product-update training third. These videos are the densest in Freshworks vocabulary and the most likely to produce egregiously wrong auto-captions. Prioritise any video that agents are currently using as a reference that they may be relying on for information about how to use their tools.
QA calibration archives. Historical QA calibration recordings may be referenced as training material and deserve corrected captions, but they are lower priority than active onboarding and product-update videos.

FAQ — Freshdesk captions

Does Freshdesk have a built-in caption tool for Help Center video articles?

No. Freshdesk's Help Center (the Solutions tab visible to customers, or the portal at company.freshdesk.com/support/home) is a text-and-embed knowledge base platform. It does not have a built-in speech-to-text caption generator and does not have a caption file upload UI for video embedded in articles. Video in Freshdesk Help Center articles is typically embedded from YouTube, Vimeo, or Loom via an iframe or URL-paste embed. The caption track on those embedded videos is managed in the originating platform — YouTube Studio, the Vimeo video settings, or Loom's Replace Transcript feature. You upload the corrected SRT or VTT to the source platform, and the corrected track propagates automatically to all embeds of that video including the Freshdesk Help Center article embed. If video is uploaded directly to a Freshdesk article as a raw file (uncommon, and not the recommended workflow for captioning purposes), it plays in a basic HTML5 video element that does not support attached caption tracks. The recommended architecture for captionable Help Center video is to always host video on a captioning-capable platform (YouTube, Vimeo, Wistia) and embed from there into Freshdesk.

Do the ADA Title III obligations apply to our B2B SaaS company's Freshdesk Help Center?

Yes. ADA Title III's "place of public accommodation" analysis does not distinguish between B2B and B2C business models — it looks at whether the website is accessible to the general public without authentication. A Freshdesk Help Center that any visitor can reach at support.yourcompany.com without logging in is public-facing content subject to Title III, regardless of whether your company sells to enterprises, SMBs, developers, or consumers. The B2B / B2C distinction affects your customer mix, not the public accommodation analysis. If your Help Center requires login (a "Restricted Help Center" in Freshdesk's portal settings), the public-facing Title III analysis changes — but the employer accommodation obligation under ADA Title I for hearing-impaired customers who are your authenticated users may still apply, and the employer accommodation obligation for hearing-impaired support agents under ADA Title I and state law applies regardless of whether the Help Center is public or restricted. California Unruh Act exposure follows the same public-versus-restricted analysis as Title III: a restricted Help Center has narrower Unruh exposure, but Unruh also applies to employment-context training video for California-based employees.

What is the difference between Freshdesk Help Center captioning and Zendesk Help Center captioning?

The compliance obligation is identical: both are public-facing knowledge base platforms, both typically embed video from YouTube, Vimeo, or Loom, and both require WCAG 2.1 AA-accurate captions on Help Center video under ADA Title III. The technical captioning workflow is the same: correct the caption file at the source platform (YouTube Studio, Vimeo settings, Loom Replace Transcript) and the corrected track propagates to the Help Center embed. The material difference is in the vocabulary failure mode: Freshdesk training video references the Freshworks Suite product vocabulary (Freshdesk, Freshservice, Freshsales, Freshchat, Freshcaller, Freddy AI, Freddy Copilot, Freddy Self Service, Freddy Insights, Dispatch'r, Observer, Supervisor, Scenario Automation, the Neo platform, OmniChannel), while Zendesk training video references the Zendesk Suite vocabulary (Zendesk Guide, Zendesk Talk, Zendesk Chat, Zendesk Sell, Zendesk Explore, Zendesk QA, Zendesk Bots, Triggers, Automations, Macros, Sunshine Conversations). The two vocabulary sets are almost entirely non-overlapping, which means a glossary designed for one platform will not correct captions for the other. GlossCap's glossary-biased Whisper decoding handles both because the glossary is company-specific and includes whatever platform terms the company's training video references. See Zendesk captions for the Zendesk-specific vocabulary failure mode and workflow.

How does Freddy AI affect caption accuracy for Freshdesk training video?

Freddy AI introduces three compounding caption accuracy problems. First, the brand name itself: "Freddy" is an unusual spelling of a common proper noun, and STT systems consistently produce "Freddie" instead of "Freddy." While the spoken form is identical, the written form matters for captions that are read by someone following the transcript and looking for a menu item or documentation article labelled "Freddy" (not "Freddie"). Second, the three Freddy sub-products — Freddy Self Service, Freddy Copilot, Freddy Insights — are three-word product names where every word is common and the proper-noun reading of the three-word phrase requires context the STT system does not have. "Freddie self service" (three lowercase words) does not signal "this is a specific Freshworks AI product" to the caption reader; "Freddy Self Service" (with capitalisation) does. Third, Freddy AI training video is among the vocabulary-densest training content a Freshdesk-using company produces: it discusses both what Freddy AI does (bot flows, agent-assist suggestions, analytics) and how it interacts with the company's product (which is the subject of the knowledge base articles that Freddy Copilot surfaces and the bot intents that Freddy Self Service handles). The company-product vocabulary layer compounds every Freshworks brand-name error with company-specific terminology errors. A company glossary that includes all Freddy sub-product names, all Freshworks platform names, and all company-product terms handles this systematically — no per-term manual review required.

Can we use Freshdesk's built-in auto-transcript feature for compliance purposes?

Freshdesk does not have a built-in auto-transcript or caption-generation feature for Help Center video. Freshdesk is a text-and-embed platform; it does not process the audio of embedded videos and does not produce transcripts or caption files. The auto-transcript that appears on Loom videos embedded in Freshdesk articles is Loom's own auto-generated transcript — not a Freshdesk feature. The auto-captions on YouTube videos embedded in Freshdesk Help Center articles are YouTube's auto-generated captions — not a Freshdesk feature. Both Loom's auto-transcript and YouTube's auto-captions fall short of WCAG 2.1 AA SC 1.2.2 accuracy standards for technical or product-specific content, for the vocabulary failure mode reasons described in detail above. The compliance-grade approach is to produce a corrected SRT or VTT using glossary-biased transcription and upload it to the source platform (YouTube Studio, Vimeo, Loom's Replace Transcript). Freshdesk itself plays no role in this workflow.

Which Freshworks products besides Freshdesk require captioned training videos?

Every Freshworks product that your company trains employees to use — and every Freshworks Help Center that your company publishes for its own customers — has the same captioning obligation structure. Freshservice (ITSM) training video for IT staff carries employer accommodation obligations for hearing-impaired IT employees under ADA Title I and state law, just as Freshdesk training video does for support agents. Freshsales (CRM) training video for sales representatives carries the same employer accommodation obligation. Freshchat agent training, Freshcaller agent training, and Freshmarketer training for marketing teams all follow the same compliance structure. If any Freshworks product is used as the basis for a public-facing Help Center or knowledge base (not all are — Freshservice, for example, is primarily an internal ITSM tool), that Help Center's video content also carries the ADA Title III public accommodation obligation. The common thread is that any Freshworks product training video will have the Freshworks-suite vocabulary failure mode — the "Fresh-" prefixed product names, the Freddy AI sub-product names, the platform-specific configuration terminology — which means a glossary that includes the full Freshworks vocabulary is the right tool for any company in the Freshworks ecosystem, not just Freshdesk users.