Get tomorrow's brief in your inbox
Today: OpenAI removed text chat limits for free ChatGPT users and upgraded default models across all tiers. HubSpot launched AEO tools to track brand visibility in AI search engines like ChatGPT and Perplexity. Naïve raised $28.5M to help developers automate company setup and operations through AI agents.
ChatGPT Just Became Unlimited for Free Users
OpenAI removed text chat limits for free and Go tier users starting this week. Previously, free users hit rate limits after too many queries. Now they can have unlimited text conversations, though file uploads and image generation remain capped.
The upgrade includes the new GPT-5.6 Luna model as the default for free users (replacing GPT-5.5) and a "Think" button that triggers higher reasoning for complex questions. Plus and Pro subscribers get an upgraded GPT-5.6 Sol with 68% fewer factual errors than GPT-5.5-Instant, plus a thinking slider to adjust response depth.
For small businesses, this changes the cost equation: ChatGPT's free tier is now viable for text-heavy workflows like customer support draft responses, research summaries, and internal Q&A. You no longer need to budget for a paid subscription just to avoid hitting limits during busy periods. Test it for repetitive writing tasks where accuracy matters less than volume. If you need consistently accurate answers or file processing, the paid tiers still make sense.
ChatGPT crossed 1 billion weekly users. The free tier is now a legitimate business tool, not a trial. Source
HubSpot AEO: Track Your Brand in AI Search Results
HubSpot launched Answer Engine Optimization (AEO) to help businesses track how often their brand appears in AI-generated answers from ChatGPT, Gemini, and Perplexity. It's included free in Marketing Hub Pro and Enterprise plans, or $50/month standalone.
The tool monitors AI visibility, suggests prompts based on your customer segments, and connects recommendations directly to HubSpot's content tools. Internal data shows HubSpot customers saw organic traffic drop 27% year-over-year while AI referral traffic tripled. If your buyers research on ChatGPT before visiting your site, you need to know what those AI engines say about you.
HubSpot AEO is for teams already on HubSpot who want insight-to-action in one platform. Competitor Scrunch offers deeper benchmarking across more engines but requires separate content tools. If you're not on HubSpot, Scrunch is the better monitoring-only option. Source
Naïve: AI Agents That Set Up and Run Businesses
Naïve raised $28.5M (Series A, Nexus Venture Partners) to scale its infrastructure that lets AI agents automate company formation, payments, email, phone numbers, cloud resources, and more through a single API. Developers supply a prompt to tools like Cursor or Claude Code, and the agent handles LLC formation, Stripe setup, QuickBooks connection, databases, and virtual cards.
The platform has 30,000+ developer customers and scaled annual run rate 10x to low double-digit millions in six months. Use cases include AI automation agencies, faceless content channels on TikTok and YouTube, and even an autonomous rental car agency.
Governance features let you set budgets, restrict agent capabilities, and require human approval for sensitive actions. Naïve also sells equity and debt to fund the spending (Alphabet sold $85B equity, announced a $25B bond sale to fund AI infrastructure).
For small businesses: if you're building AI-driven services or exploring autonomous operations, Naïve handles the grunt work of company setup and ongoing infrastructure. It's developer-focused, so you need comfort with APIs. Not for traditional service businesses. Source
Adobe Plugin for ChatGPT: 70+ Creative Tools in One Interface
Adobe launched a ChatGPT plugin combining 70+ tools from Adobe Express, Photoshop, Firefly, Premiere, Acrobat, Lightroom, Illustrator, InDesign, and Adobe Stock. Small businesses can describe what they need in conversational language and let the plugin coordinate tools behind the scenes.
Upload product photos and ask for a consistent visual style. Turn an instructional video into clips for TikTok, Reels, and YouTube Shorts. Upload spreadsheet data and request event badges or infographics. The plugin saves finished assets to Creative Cloud for further editing in Adobe's desktop apps.
This reduces friction for teams without advanced design skills. ChatGPT becomes the starting point for common tasks; Adobe apps handle advanced work. Best for businesses already using Adobe tools who want faster turnaround on routine content. Not a replacement for skilled designers, but a force multiplier for small teams juggling multiple channels. Source
Tax Firm Cuts Research Time by 9 Hours Using ChatGPT Enterprise
HSP GRUPPE, a network of tax advisory firms, reported 98.6% of employees saw higher productivity after deploying ChatGPT Enterprise. The firm logged 500,000+ conversations in six months and estimated 40,000 annual hours of additional capacity.
One partner cut real estate investment analysis from 9 hours to 2 hours. Another uses ChatGPT as a technical sparring partner for complex tax questions and client communications. The firm built custom Agents for client communication drafts and booking classification (final decisions remain with professionals).
The approach: embed AI into daily workflows with strong governance, monthly AI forums for use case sharing, and custom Agents that standardize best practices across the network. Professional review and responsibility always stay with the human expert.
For small businesses: this is a playbook for AI adoption in professional services. Start with one high-value use case (research, draft communications, data analysis), measure time saved, share wins across the team, and build repeatable processes. HSP's 84% weekly active usage shows adoption works when AI solves real problems and fits existing workflows. Source
Malachyte: Spotify's Recommendation Engine for E-Commerce
Three ex-Spotify engineers who built the platform's recommendation system raised $10M (seed, Bessemer Venture Partners and Gradient) to bring behavioral prediction to online stores. Malachyte uses real-time intent tracking to personalize shopping experiences for first-time visitors and repeat customers.
The system predicts what product a shopper wants next based on context (device, time of day, referral source), real-time actions (clicks, hovers, searches), and evolving taste. A search for "heavy-duty boot" followed by clicks on steel-toed boots moves work pants and gloves up the page and pushes dress shoes down, no account required.
It's live on Shopify (native integration) and available via API for larger retailers. The pitch: most e-commerce personalization relies on past purchases or demographic segments, missing real-time intent. Malachyte reads every action to refine recommendations continuously.
For small businesses: if you run a Shopify store, test the native integration. If your average customer browses multiple pages before buying, real-time intent tracking could lift conversions. Skip it if you sell one or two products with straightforward purchase paths. Source
Savers Thrift Store Launches AI Pricing Tool (ThriftIQ)
Savers Value Village deployed ThriftIQ, an AI platform that optimizes pricing for men's and women's apparel across its 375 thrift stores. The tool has priced 25M+ items in 58 pilot stores and will double coverage by year-end, reaching all U.S. and Canadian locations by early 2028.
ThriftIQ uses proprietary data sets (developed over two years) to bring consistency and predictable pricing while keeping average prices 40-70% below traditional retail. The company said it's not dynamic pricing, tags don't change after items are priced, and the goal is to keep prices the same or lower.
CEO Mark Walsh said the tool increases sell-throughs, basket sizes, profitability, and helps new stores ramp faster. It won't eliminate manual labor but makes workers more productive.
For small businesses: AI pricing works when you have enough transaction data to train models and variable inventory (like thrift). If you sell consistent SKUs, traditional pricing strategies still make more sense. Watch whether ThriftIQ delivers on the "same or lower" pricing promise or if it becomes a margin optimization tool that raises prices. Source
AMD Acquires Taalas for Custom AI Inference Chips
AMD bought Taalas, a Toronto startup that makes chips hardwired for specific AI models rather than general-purpose GPUs. Taalas promises thousands of times faster inference for targeted models at lower cost, trading flexibility for speed and efficiency. The startup raised $219M since 2023. AMD did not disclose the purchase price.
Taalas' current chip runs a small version of Meta's Llama 3.1 and uses SRAM memory on-chip for low-latency applications. The company's platform can transform any AI model into custom silicon in two months.
AMD will integrate Taalas chips into its roadmap alongside Instinct GPUs and central processors. This reflects rising demand for integrated systems with specialized accelerators for different workloads, not just GPUs. CEO Lisa Su said there's no one-size-fits-all for AI chips.
For small businesses: this doesn't change anything today, but it signals where the market is headed. If you're building AI products, expect more specialized hardware options that optimize cost and speed for specific models. GPUs remain the flexible choice for experimentation; custom chips make sense for high-volume, stable workloads. Source
Omilia Raises $67M for Customer Support Platform (Not Pure Generative AI)
Athens-based Omilia raised $67M (Series B, Expedition Growth Capital) to scale its customer support automation platform, which has been automating voice calls since 2002. The company grew ARR 10x to $60M since its 2020 Series A.
CEO Dimitris Vassos said most customer queries involve basic information (account balances) and don't need large language models. Omilia uses "any available weapon" to handle different query types efficiently, including self-learning agents across voice, chat, and messages.
Clients include Capital One, Discover, RBC, and Taco Bell (deployed across 1,000+ outlets for voice ordering). The company is in talks with two more quick-service restaurants.
Vassos said Omilia focuses on real ROI and unit economics rather than being "sexy on LinkedIn." The company will use the funding to open a U.S. office, hire a CRO, CMO, and VP of revenue ops, and grow from 500 to 600 employees by year-end.
For small businesses: if you're evaluating AI customer support tools, consider whether you need full LLM power for every query or if rule-based automation handles 80% of volume. Omilia's thesis is that generative AI is overkill for most support tasks. That's worth testing before committing to expensive AI-first platforms. Source
Suno Adds Watermarking Amid Legal Battles
AI music generator Suno will watermark songs created on its platform and limit downloads to prevent misuse on streaming platforms. The company signed a deal with Musixmatch for copyright detection and updated community guidelines to ban deceptive audio and unauthorized voice use.
Suno faces lawsuits from Universal Music Group, Sony Music Group (coordinated by RIAA), a German court ruling that it broke copyright rules, and a class action lawsuit in Massachusetts after a data breach exposed 55 million users. 404 Media reported Suno scraped YouTube, Deezer, and Genius to train models.
For small businesses: Suno is now riskier for commercial music production. If you're using AI-generated music for videos, podcasts, or ads, verify that your tool has proper licensing and won't expose you to copyright claims. Alternatives with clearer licensing include licensed stock music libraries and tools with explicit royalty-free terms. Source
Use ChatGPT's New Think Button for Complex Business Questions
Free and Go tier users get a "Think" button this week that triggers higher reasoning for complex questions. Use it for multi-step analysis, strategic decisions, financial modeling, or anything requiring deeper logic than a quick answer.
Example: "Should I expand to a second location or hire two more employees?" Press Think for a structured breakdown of cash flow, risk, and ROI rather than a surface-level response.
Test AI Observability Tools Before Production
If you're building AI features, add observability early. Tools like Langfuse (open-source), Arize Phoenix (RAG pipelines), Braintrust (evaluations), and LangSmith (execution tracing) help debug, monitor, and improve AI applications before they fail in production.
Start with one tool that fits your stack. Trace every request from prompt to output so you can identify failures, measure response quality, and track token costs. Waiting until production to add observability means you're debugging blind. Source
OpenAI's move to unlimited free text chats commoditizes basic AI access and raises the bar for paid tiers. Businesses that assumed ChatGPT required payment now have a viable free option for text-heavy workflows. Expect competitors to match or expand free tiers.
HubSpot's AEO launch signals a shift: AI search visibility is now a core marketing channel, not an experiment. If your buyers research on ChatGPT or Perplexity, you need monitoring tools to track brand mentions and optimize for AI-generated answers.
Naïve's autonomous business infrastructure and Omilia's pragmatic "not all queries need LLMs" approach highlight the industry split: some companies bet on full AI automation, others focus on targeted efficiency. Watch which approach delivers better ROI over the next 12 months.