Daily Gig News

GWG Summary: Meta has expanded its Muse personal agent with a small-business setup aimed at handling routine work across connected services. The product is currently available in the United States and Canada. Users can link tools including Canva, Figma, Notion, Slack, Stripe, Shopify, QuickBooks, Dropbox, Zoom, and Meta business accounts, with custom connectors available for unsupported services.

For independent professionals, the useful shift is that Muse can work with business context rather than treating every request as a blank prompt. Meta says it can review sales, campaigns, social performance, books, customer records, email, and calendars to build plans, identify items that need attention, draft campaigns, and flag unusual expenses. It can also surface unanswered messages and prepare first drafts without waiting for a new prompt.

Meta says the agent requires approval before it publishes content, sends messages, or spends money. Most usage is free, while higher limits are sold through subscription plans. The launch does not remove the need to review financial findings, customer communication, or campaign copy, but it gives solo operators a new way to coordinate administrative work across the apps they already use.

Why It Matters - GWG's Take: Freelancers and creators who run their own back office can test whether one agent reduces the time spent moving information between project, payment, accounting, and marketing tools. The approval controls make it practical to trial on drafts and analysis before allowing any outward action. Availability is currently limited to the United States and Canada, and higher usage may require a paid plan.

GWG Summary: Microsoft has expanded the stable Visual Studio 2026 release with a preview of Bring Your Own Model for its new Agent Mode. The option is enabled by default in Community, Professional, and Enterprise editions and can connect to Microsoft Foundry, OpenAI, Anthropic, Ollama, or supported custom endpoints. That gives developers a way to use a preferred or client-approved model in the editor without depending entirely on the standard GitHub Copilot model selection. The update also changes how the preview works: the earlier Bring Your Own Model experience in the old Ask and Agent modes is no longer supported, and previously added Ollama models must be added again after updating. Microsoft warns that model support varies and Visual Studio will flag unavailable agent functions. Beyond model choice, the release lets GitHub Copilot propose fixes for vulnerable NuGet packages from the Error List and adds a Git agent that can summarize pull-request discussions and link developers to relevant files or comments. Visual Studio can also discover and attach to Podman containers for .NET and C++ debugging.

Why It Matters - GWG's Take: Freelance developers can match the coding model to a project, privacy requirement, or existing provider account while staying inside Visual Studio. The built-in package-remediation and pull-request tools may reduce time spent on maintenance and review. Anyone already testing the older model setup should update carefully because the release removes that path and requires Ollama models to be configured again.

GWG Summary: The September product update from Canva brings several practical workflow tools into the design platform. Canva Websites can now use custom navigation and buttons, while Google Analytics support lets users measure traffic and visitor behavior without moving the site to a separate service. A new PayPal app can place payment links directly inside designs, giving creators and small businesses a shorter route from a portfolio, menu, or promotional page to checkout. Canva Sheets also gains tags, filters, and conditional formatting for organizing collaborative work. Whiteboards can be divided into sections with direct links, which should make large planning boards easier to share and navigate. Other additions include automatic brand-kit creation from a website address and custom image frames generated from hand-drawn shapes. Canva Code can now turn spreadsheet data and web research into visual dashboards. The company collected the features in a Design School update last revised September 28. Together, the release expands Canva beyond asset production into lightweight publishing, measurement, payment, data organization, and reporting workflows.

Why It Matters - GWG's Take: Freelancers can now build a simple Canva-hosted landing page, track its performance, and send visitors to a PayPal payment link with fewer external tools. The spreadsheet and dashboard additions may also speed up recurring client reports and content planning. Creators should still test payment links and analytics before relying on them in a live sales workflow.

GWG Summary: Riverside has added full Safari support for its browser-based content workflow, allowing Mac users to record, edit, host, and publish without moving to Chrome or Edge. The company lists the September 27 update as available on its free plan. That broadens the practical setups available to creators, remote interviewers, podcast producers, and freelance editors who collaborate with guests or clients using Apple hardware.

A separate update released the same day lets users regenerate a recording transcript from the editor and choose the language used for the new transcription. The feature is intended for cases where the first transcript was created in the wrong language, avoiding a manual correction pass or a separate transcription service. Riverside also marks transcript regeneration as a free-plan feature.

These are workflow changes rather than new monetization programs, but both address common production friction. Safari support reduces the need to persuade participants to change browsers before a session, while the transcript control gives editors a direct recovery path when automatic language detection fails. Creators should still test recording permissions and audio before client sessions, since browser compatibility does not eliminate device-specific setup problems.

Why It Matters - GWG's Take: Freelance producers can accept a wider range of client and guest browser setups without adding a browser-switching step. Free transcript regeneration may also save editing time when a multilingual recording is initially transcribed incorrectly. The update is most useful to people who deliver podcasts, interviews, webinars, or repurposed video as a service.

GWG Summary: Upwork is testing Upwork Now, an on-demand matching flow available to a limited group of freelancers. Eligible users can turn on a Work Now status from Find Work, enter a pool visible to clients with Hire Now jobs, and browse a live feed filtered around their skills and rates. Listings show a countdown indicating how long the client is actively hiring, and freelancers can apply directly from that feed.

The feature is built around immediate availability rather than the usual proposal cycle. Upwork says participants are expected to begin accepted work the same day. A browser session ends after five minutes of inactivity, while a mobile session can remain active as long as the freelancer responds to invitations or messages within five minutes. Users can restart or disable a session at any time, and they remain free to reject an offer after reviewing the scope, rate, deadline, and client details.

These contracts use the same profile and work history as regular Upwork jobs. Their outcomes also count toward the freelancer Job Success Score, so speed does not remove the need to assess workload and terms before accepting.

Why It Matters - GWG's Take: Selected freelancers gain a new route to jobs from clients who want work to begin immediately. The tradeoff is operational: staying visible requires close monitoring, and every accepted contract can affect the same Job Success Score as regular work. Freelancers who see the toggle should confirm scope and capacity before accepting, rather than treating the short response window as a reason to skip due diligence.

GWG Summary: Fiverr says it will discontinue Personal Assistant, its AI tool for handling first-contact messages from prospective clients, on October 1, 2026. New subscriptions already closed September 1, and the feature will be removed automatically from existing users at shutdown. The tool currently answers inquiries from new buyers, gathers project requirements, handles frequently asked questions, and can transfer complex conversations to the seller; it does not serve repeat clients. Fiverr says existing inbox conversations will remain after shutdown and other Seller Plus benefits will not change. For freelancers who have relied on automation outside working hours, the removal means new buyer messages will reach them directly from first contact. Sellers should review response-time routines, saved replies, notification settings, and lead-qualification questions before October 1 so inquiries are not missed. Users considering Seller Plus should also note that its advertised Personal Assistant benefit is ending even though Standard and Premium subscriptions continue. Fiverr frames the decision as part of its focus on higher-value work and direct, personal communication between freelancers and clients. The notice gives no replacement tool and says no action is required for the removal itself.

Why It Matters - GWG's Take: Sellers using Personal Assistant have only days to replace its lead-screening and off-hours response role or risk slower replies to new prospects. Those evaluating Seller Plus should not count Personal Assistant when deciding whether the remaining benefits justify the monthly or prepaid price. Existing conversations remain, so no export appears necessary, but sellers should test their manual handoff before October 1.

GWG Summary: UserTesting has launched early-access Model Context Protocol servers for both UserTesting and User Interviews, bringing research tasks into AI applications that support the standard. Research teams can recruit participants, create and launch studies, retrieve customer feedback and return findings to tools such as ChatGPT, Claude, Figma Make and Gemini without moving through each platform separately. UserTesting says the two services reach a combined participant network of more than seven million people in 34 countries. The servers can also bring complete session transcripts into an AI workflow and generate findings that remain connected to the underlying research evidence. This release is aimed at organizations and researchers running studies, rather than a new participant feature or a guarantee of additional paid tests. Access is not yet general: both servers are in Early Access, and interested users must request entry. The practical change is that a researcher can move from an AI-assisted concept to participant recruitment and study setup within the same working environment. That could make repeated testing easier to include during faster product-development cycles while preserving access to real participant feedback.

Why It Matters - GWG's Take: Independent UX researchers can use the integration to reduce setup and handoff work when clients already operate in compatible AI tools. It may also help researchers make participant testing part of more iterations instead of reserving it for a separate project phase. Since access is currently limited, freelancers should confirm eligibility and client data rules before building the servers into a promised workflow.

GWG Summary: Fiverr says its 2026 Holiday Marketing Index shows retailers increasing their use of freelance specialists for AI search visibility, user-generated video and ecommerce optimization. The company based the report on a survey of hundreds of top-rated Fiverr freelancers in eight countries, along with marketplace hiring trends, so the figures describe activity reported within its own ecosystem rather than the entire freelance market. Among surveyed search specialists, 67% reported growing retailer demand for generative engine optimization, compared with 56% who saw more demand for conventional search ranking work. Fiverr also found that 69% of UGC freelancers reported increasing holiday demand. Store-management work was another active area: 57% of ecommerce specialists saw higher demand, while 43% reported more requests for Shopify and conversion-rate optimization. The opportunity comes with pricing pressure. Sixty-one percent of contractors said retail clients expected discounts when AI tools were part of the process, even though 89% said clients ultimately evaluated the result rather than the production method. The clearest pattern is a move toward advisory work: majorities across search, ecommerce and UGC said clients increasingly want strategy, not only execution.

Why It Matters - GWG's Take: Freelancers planning fourth-quarter offers have a concrete signal to package GEO with established SEO, or pair UGC production with ecommerce conversion work. The pricing data also supports spelling out the value of research, judgment and strategy instead of accepting an automatic discount for using AI. Because Fiverr produced the survey from its own marketplace, treat the percentages as a directional sales signal rather than a universal rate forecast.

GWG Summary: Microsoft has introduced a redesigned Copilot centered on three capabilities called Home, Code, and Autopilot. Home combines chat with longer delegated tasks and brings Word, Excel, and PowerPoint work into the Copilot interface, where users can create or revise editable files. Code turns natural-language instructions into small apps, dashboards, trackers, automations, and workflows, using technology related to GitHub Copilot and a managed runtime inside Microsoft 365. Autopilot is designed as a persistent agent with its own identity, memory, computer, and workspace that can monitor channels, follow up with people, and continue recurring work without waiting for a new prompt. The availability is staged rather than general. Microsoft says Home and Code will begin rolling out through its Frontier program in the coming weeks, Code is planned for preview for Microsoft 365 Premium and Pro subscribers later in 2026, and Autopilot is expanding to private preview at the end of September. Freelancers should therefore treat these as emerging client capabilities, not tools every Microsoft customer can use today.

Why It Matters - GWG's Take: Freelancers who deliver Office files, internal tools, or workflow automation may face more clients attempting basic production themselves. The same rollout can create work in prompt design, workflow setup, governance, quality review, and customization. Because access is limited and staged, confirm a client environment before promising a Copilot-based workflow.

GWG Summary: Figma has added vertical wrapping to auto layout, giving designers a native way to make content flow into additional columns. When an auto-layout frame uses vertical flow, items that exceed the available height can now move into a new column instead of continuing beyond the bottom edge. Figma says the behavior is aligned with CSS flexbox, bringing the design model closer to the way responsive interfaces can be implemented in code. The update is useful for layouts such as menus, option lists, cards, directories, and other repeated content that must adapt when entries are added or removed. Previously, a designer might have needed to split items across frames manually or build a more complicated structure to demonstrate the same behavior. Freelancers can use the new control to create cleaner responsive specifications and reduce repetitive revisions when a client changes content volume. It is still important to test frames at the dimensions used in the actual project and confirm the intended column order with developers, since wrapping behavior alone does not define every responsive rule or accessibility decision.

Why It Matters - GWG's Take: Freelance product and web designers can represent responsive multi-column behavior with less manual frame management. The closer match to CSS flexbox may also make handoff discussions clearer, but designers should still document breakpoints, reading order, and exceptions for developers.

GWG Summary: UserTesting has added an extension-free option for recorded web Think Out Loud and Interaction Tests as part of its September product update. Researchers can now run those studies without asking participants to install the UserTesting browser extension, removing a technical barrier for people using managed work computers or other devices where browser add-ons are blocked. The same release adds adaptive AI follow-up questions to some unmoderated Think Out Loud studies, updates Live Conversation scheduling and session controls, and introduces card sorting within the newer Think Out Loud format. For participants, the extension change is the most immediate shift. It can make some studies accessible on devices that previously could not meet setup requirements, although each study can still impose its own browser, device, recording, or prototype conditions. AI follow-ups also mean an unmoderated session may react to an answer and ask for more detail, so contributors should expect less predictable scripts and explain their reasoning clearly rather than treating each prompt as isolated. UserTesting labels the page as its September 2026 release but gives no day, so the source date is left blank.

Why It Matters - GWG's Take: Extension-free testing may widen eligibility for contributors who use locked-down or restricted devices. Participants should still check each invitation for device and recording requirements, because the change does not guarantee that every study will work without extra setup. Adaptive follow-ups also make clear, specific verbal reasoning more important during unmoderated tests.

GWG Summary: New Jersey’s labor department says new rules explaining the state’s ABC test for contractor status will become operative October 1. The rules do not create a new classification test or change enforcement policy, acting commissioner Kevin Jarvis told New Jersey Monitor; instead, they codify existing court and administrative decisions. Under the test, a hiring business must establish all three conditions for a worker to remain an independent contractor: the worker is free from its control, the service is outside its normal business or workplace, and the worker operates an independently established business.

The clarification matters because workers are otherwise presumed to be employees under the laws covered by the rules, including wage-and-hour, wage-payment, and unemployment-compensation provisions. Business groups and some freelancers have asked for a delay, arguing that the framework may threaten flexible arrangements. State officials dispute that interpretation and say genuine independent businesses remain protected. As of the September 21 report, the department said the adopted regulations require an October 1 implementation and it was unclear whether the agency could delay them without legislative or executive action.

New Jersey freelancers should review whether their actual client relationships—not just contract labels—fit all three prongs. Clients that closely direct work or hire contractors to perform their core service should also reassess classification before the operative date.

Why It Matters - GWG's Take: Online freelancers serving New Jersey clients may see contracts or work arrangements reviewed as businesses prepare for the October 1 operative date. Maintaining multiple clients, control over methods and hours, separate business expenses, and an independently marketed service may help document a genuine independent business, but classification depends on the full ABC test. Anyone with a close call should use NJDOL guidance or seek qualified legal advice rather than relying on the word contractor in an agreement.

GWG Summary: Anthropic released Claude Opus 5.5 on September 22, positioning it as a faster and less expensive successor to Opus 5 for coding, research, and business workflows. API pricing is $4 per million input tokens and $20 per million output tokens, down 20% from Opus 5, while cache reads drop from $0.50 to $0.20 per million. Anthropic says typical workloads cost about 40% less because the model also uses fewer tokens, and output generation is more than 30% faster.

The model is available across Anthropic’s own products and through AWS, Google Cloud, and Microsoft Azure. Paid Claude subscribers also receive higher five-hour usage limits and a reset they can save for later. Anthropic’s internal and partner evaluations report improvements in long, multi-file coding jobs, data collection, financial analysis, and connected-app workflows. Those performance claims come from the vendor and early testers, so freelancers should validate them against their own recurring tasks before changing a production workflow.

The release also changes some operating constraints. Opus 5.5 cannot run with thinking mode disabled, and newer API accounts receive preserved-thinking protections that limit editing prior reasoning context. Cybersecurity and biology requests can be routed through additional safeguards, which may matter to specialists in those fields.

Why It Matters - GWG's Take: Freelancers who already use Opus 5 can test the newer model for complex, high-value work without automatically accepting the old cost profile. Lower token and cache prices could improve margins on fixed-price coding, research, and automation jobs, while higher subscription limits may reduce interruptions. Compare total task cost, revision time, and output quality on representative client work before switching.

GWG Summary: YouTube announced a package of creator features at its 2026 Made On YouTube event, covering production, audience development, sponsorship work, rights protection, and live video. A conversational editing assistant powered by Gemini Omni is coming to Shorts and the YouTube Create app. YouTube says creators will be able to request edits in plain language, including trimming speech, rearranging frames, synchronizing music, and adding on-screen hooks. YouTube Studio is also getting improved feedback tools, while Creator Partnerships is intended to make channels easier for brands to find and provide resources for responding to sponsorship briefs and preparing pitches. The company is expanding likeness detection to help creators identify unauthorized uses of their face or voice. For live creators, planned additions include automatic dubbing and collaborative showdowns. YouTube will also let creators organize Shorts into seasons and episodes, giving serialized work a clearer viewing path across mobile devices and televisions. The announcement combines features available on different timelines, so creators should check their own Studio and app access rather than assume every tool is immediately available.

Why It Matters - GWG's Take: The editing assistant could reduce production time for short-form client or channel work, while the partnership features may create a more direct route to sponsorship opportunities. Likeness detection adds a practical safeguard for creators whose face or voice is part of their business. Because availability varies by feature, creators should verify rollout status before changing a workflow or promising a capability to clients.

GWG Summary: Anthropic has launched Claude Marketplace, a directory and purchasing hub for software, integrations, and professional services built around Claude. The marketplace is live with more than 2,000 connectors and plugins, alongside third-party agents and products. Enterprise customers can also use part of their committed Anthropic spending to buy eligible software through the marketplace, which could shorten procurement for vendors already selling to those teams. Independent developers and small software firms can apply to list Claude-powered agents or products. Builders can also publish connectors or plugins based on Model Context Protocol and Agent Skills, while consultants and systems integrators can seek placement through the Claude Partner Network. The launch does not guarantee acceptance, visibility, or sales, and Anthropic frames the buyer side mainly around organizational customers rather than individual consumers. Still, it creates a formal distribution route inside the Claude ecosystem for people who package specialized workflows, integrations, or implementation expertise. The marketplace also gives freelancers a single place to compare tools that may fit client delivery work.

Why It Matters - GWG's Take: Freelance developers and small AI consultancies now have a direct application path to put Claude-based products or services in front of business buyers. Builders should review the listing routes and eligibility before investing in packaging work, because inclusion and demand are not automatic. Freelancers buying tools can also use the directory to evaluate integrations for client workflows.

GWG Summary: SpaceXAI released Grok 4.7, a new model aimed at coding and professional knowledge work that can continue working through longer, more complex assignments. The company says the model uses a larger base than Grok 4.6, received more training on tasks that take hours to finish, and is better at checking its own output and handling long context. Those claims come from SpaceXAI and its own benchmark suite, so freelancers should validate the model on representative client work before changing their workflow.

The release is available now in Cursor and Grok Build, as well as through the Grok API, third-party coding tools, model routers, and cloud platforms. API pricing starts at $2 per million input tokens and $6 per million output tokens, matching Grok 4.6 pricing. A faster version doubles output speed and price. SpaceXAI also says Grok 4.7 improves document and presentation creation, although the clearest practical proposition is long-running software and terminal work. Developers can try it in Grok Build for free, which lowers the cost of comparing its results with an existing coding assistant before using it on paid assignments.

Why It Matters - GWG's Take: Freelance developers can now compare another frontier model inside Cursor or Grok Build without rebuilding their toolchain. The unchanged API price may make longer automated coding runs more economical, but vendor benchmarks are not a substitute for testing accuracy, review time, and token use on the kinds of repositories a freelancer actually serves.

GWG Summary: Canva has introduced Canva ProSuite, a connected collection that brings Affinity, Cavalry, Flourish, and Leonardo.Ai into a broader professional creative workflow with Canva. The September 15 launch also delivered substantial updates to Affinity and Cavalry rather than simply applying a new label to the products.

Affinity added more than 60 features across illustration, photo editing, and page layout. Highlights include a Blend Tool, scripting for repetitive production work, a revised masking workflow, expanded RAW photo controls, native InDesign file import, Canva Brand Kit syncing, and a data-merge connection to Canva Sheets. A Claude connector can also carry out batch edits and production tasks from natural-language instructions. Canva says files remain local and are not used to train Affinity AI features.

Cavalry received more than 50 motion-design improvements, including stronger graph editing, Canva Sheets import, and richer SVG handling. Canva also plans to let designers mark selected parts of Cavalry animations, such as text or color, for teammates to edit inside Canva while the rest of the scene stays protected. Affinity and Cavalry remain free at their core, while the Canva AI Studio features in Affinity require a paid Canva plan.

Why It Matters - GWG's Take: Freelance designers can now move more client work across layout, photo, motion, data visualization, and asset generation without stitching together as many separate tools or subscriptions. The free core apps lower the cost of testing this workflow, while native file and brand connections may reduce handoff and revision time. Designers should still distinguish the free production features from paid AI capabilities before changing their software budget.

GWG Summary: Catalant has launched Forward Deployed Experts, or FDX, a consulting model that combines experienced business operators from its vetted independent-expert network with AI engineers on enterprise implementations. Instead of ending at a strategy presentation or pilot, these teams are designed to continue through workflow redesign, deployment, adoption, and change management. Catalant says it shaped the offering after analyzing 6,600 anonymized client conversations across more than 900 Fortune 500 and private-equity-backed companies during the past year. Its internal analysis found that AI had become a central priority while organizations continued to struggle with connecting projects to business strategy and execution. The company reports that 43.7% of the analyzed interactions involved businesses that had already used at least one major consulting firm before seeking Catalant support. FDX starts with business goals and performance measures, then identifies suitable AI applications and reorganizes workflows around them. Each engagement is led by operators with firsthand experience in the relevant business function working alongside technical specialists. Catalant says its expert network is used by more than 30% of Fortune 500 companies and 72% of leading private equity firms by assets under management. The launch formally packages independent expertise into implementation work that extends beyond standalone advice.

Why It Matters - GWG's Take: For independent consultants on Catalant, the platform is framing demand around hands-on operational expertise and the ability to work with AI engineers, not generic AI knowledge. Specialists can respond by emphasizing measurable implementation work, workflow redesign, adoption, and change management in profiles and case studies. The announcement does not describe open enrollment or new pay terms, so it is a positioning signal rather than a promise of assignments.

GWG Summary: GitHub has added a Sentry workspace to the Copilot app that can turn production crash reports into a guided repair workflow. Developers can bring an error report, stack trace, and related context into Copilot, ask it to investigate the cause, validate a proposed fix, and prepare a pull request without manually moving the evidence between tools.

The September 18 weekly release also expands agent work in Visual Studio Code. Support for running agents inside local Dev Containers is rolling out gradually for projects that use Docker and a supported container configuration. Agent Host sessions can now create pull requests from the Agents window, with a chance to review the generated title, description, and draft status before submission. GitHub also introduced optional cleanup for inactive agent sessions after their pull requests merge.

Other changes include three auto-selection tiers that let users emphasize efficiency, balance, or intelligence while drawing from the same available model pool. Business and Enterprise users can request more AI credits after reaching a budget limit, subject to an owner or billing manager decision. The release is available now except where GitHub labels a feature as a gradual rollout or preview.

Why It Matters - GWG's Take: Freelance developers who handle production support can shorten the path from a client crash report to a reviewable fix, especially when the client already uses Sentry. Local Dev Container support also makes agent work more practical on projects with pinned dependencies. Review generated changes and pull-request text before sending them to a client, because the workflow accelerates delivery but does not replace technical accountability.

GWG Summary: Amazon says Mechanical Turk will permanently close on September 30, 2026, ending both worker task submission and requester access to the crowdsourcing marketplace. The company’s closure FAQ tells workers to check that a payment method and transfer frequency are configured in the Worker Portal before the deadline. Approved work will continue to be paid on the existing schedule after the service closes, while requesters can approve or reject pending submissions and issue bonuses through October 30. Work that is still awaiting a decision will auto-approve after the standard 30-day window if the requester takes no action.

Workers will retain access to transaction history until January 28, 2027, and Amazon says it will still distribute required tax documents through its usual process. The same shutdown also removes the Mechanical Turk worker option from SageMaker Ground Truth and Amazon Augmented AI, so labeling and human-review jobs tied to those services need another workforce. The FAQ does not show a publication date, but its September 30 closure date and post-closure payment timetable make the notice current and time-sensitive.

Why It Matters - GWG's Take: MTurk workers should confirm their payment method and transfer schedule now, then save any transaction records they may need before access ends in January. Anyone relying on MTurk income must move to another source of survey, research, or data-labeling work. Pending assignments and bonuses can still produce payments after the platform closes, but workers should track those deadlines rather than assuming the account will remain fully usable.