Indexed — Wednesday, May 6, 2026
The week's clearest read on AI automation jobs: Freshworks just cut 500 roles while saying more than half its code is now written by AI. That's not a forecast anymore. That's a payroll line.
Today we're unpacking what the SaaS layoff wave means for operators, the stack that actually moves the needle on growth right now, and a prompt to audit your own role before someone else does.
The Signal: AI automation jobs hit the SaaS middle
Freshworks said Tuesday it will cut 11% of its workforce — about 500 roles — as the company restructures around AI in product and engineering. CEO Dennis Woodside told Reuters that "over half of our code is written by AI" and that automation has reduced "rote work that technology can take care of." The restructuring carries a one-time charge of about $8 million. As of December 31, 2025, Freshworks had roughly 4,500 full-time employees. (Reuters)
Reuters frames it as part of a broader pattern: Atlassian disclosed roughly 10% cuts last month, and traditional SaaS makers — Freshworks, Salesforce, ServiceNow — have all seen pressure as buyers weigh whether tools from companies like Anthropic can replace incumbent workflows. Freshworks shares were down more than 8% intraday on the news.
The operator read: AI automation jobs aren't just appearing in "AI engineer" listings. They're showing up as the absence of jobs in customer support, QA, internal tooling, and middle-tier engineering at established SaaS shops. If your roadmap or your résumé still assumes a 2023-shaped org chart, this week is the nudge.
Two more items worth sitting with:
Image models are winning the install war. A new Appfigures report covered by TechCrunch found image-model launches drove 6.5x more downloads than traditional model updates. ChatGPT and Gemini each added tens of millions of downloads off image releases — but Nano Banana drove only an estimated $181,000 in 28-day consumer spend, and Meta's Vibes added 2.6M downloads with no meaningful revenue. Downloads ≠ dollars. (TechCrunch)
Apple is reportedly building a model marketplace into iOS 27. Per Bloomberg via TechCrunch, an "Extensions" feature would let users pick third-party LLMs (Google and Anthropic are testing; ChatGPT's status is unclear) to power Siri, Writing Tools, and Image Playground. Apple hasn't confirmed. (TechCrunch)
If you're an operator, the through-line is clear: the labor market, the install funnel, and the OS layer are all reorganizing around model choice. Pick your spots.
The Stack: Tools that survive the automation cut
If "over half of our code is written by AI" is the new normal, the stack you carry into 2026 should look less like "one chatbot tab" and more like a small operating system. Here's the working set we'd defend:
ChatGPT — default desk for drafting, analysis, and the prompt below.
Claude — long-context reading and careful rewrites for anything legal, exec, or compliance-flavored.
Google Gemini — Workspace-native drafting and live Search grounding.
Perplexity AI — cited research before you ship a claim.
GitHub Copilot and Cursor — IDE-native code generation if you ship anything technical.
Midjourney — the visual lever that, per Appfigures, is now driving more app growth than chatbot updates.
QuillBot — paraphrasing, summarization, and grammar in places where ChatGPT is overkill.
Notice the shape: one default chatbot, one long-context reader, one search-grounded researcher, two code surfaces, one image surface, one writing utility. That's roughly the team you'd hire in 2022 — except today you license it for the price of a midweek lunch.
Prompt of the Day: Audit your role for AI automation
Open ChatGPT (or your default model) and run this:
You are a workforce strategist who has read every recent SaaS layoff memo, including Freshworks cutting 11% of staff with more than half of code now written by AI. I'll describe my role, my outputs, and a typical week. Map my work into three buckets: (1) tasks that are already automatable today with off-the-shelf AI, (2) tasks that require human judgment but could be 5–10x faster with AI assistance, and (3) tasks that are durable for the next 24 months. For each bucket, list the specific tools or workflows I should adopt, the skills I should build, and the metrics that would prove I'm worth keeping. Be blunt. Then propose a 30-day plan to shift my time toward bucket 3.
Paste your role, your last five deliverables, and a representative week. Let it push back. The point isn't to feel reassured; it's to find the bucket-1 work you're still doing on instinct and move it.
Tool of the Day: ChatGPT
ChatGPT is the most widely used AI assistant for conversation, writing, analysis, and coding — now powered by GPT-5, with the o3 reasoning model for multi-step problems.
Why it earns the slot today: when SaaS companies are openly attributing layoffs to AI-written code and automated routine work, the baseline assistant on every operator's machine has to be a Swiss Army knife. ChatGPT covers writing, research synthesis, code in 50+ languages, file analysis, image generation, web browsing, voice, and Custom GPTs in one workspace.
Where it shines for operators:
Turning messy notes into SOPs, briefs, and proposals client-ready in one pass
Analyzing uploaded files, tables, and small datasets without spinning up a notebook
Debugging code, generating tests, and explaining unfamiliar codebases
Building Custom GPTs that bottle a recurring workflow your team runs every week
It's freemium, sits at a 4.7 rating, and a popularity score of 99 in our index — meaning it's the tool your peers have already standardized on. If you're picking one license to defend in a budget review, this is usually it.
Quick Hits
Image > chat for app growth. Per Appfigures via TechCrunch, image-model launches now drive 6.5x the downloads of chatbot upgrades — but Nano Banana converted only ~$181K in 28-day spend. Treat image features as a top-of-funnel lever, not a revenue line.
OpenAI phone rumor — caveat in. The Verge reports analyst Ming-Chi Kuo claims an OpenAI phone is being "fast-tracked" for early-next-year mass production. Single-source, analyst-driven; file under watchlist, not roadmap. (The Verge)
Krutrim pivots to cloud. India's first GenAI unicorn is shifting from model development to cloud services after pausing chip design and cutting 200+ roles, per TechCrunch. Another data point on how brutal the model-builder economics are. (TechCrunch)
Apple's "Extensions" is unconfirmed. Bloomberg-sourced; treat as direction, not date.
<!-- meta-description: AI automation jobs are reshaping SaaS: Freshworks cut 11% as AI writes over half its code. Get the operator stack, prompt, and tool to stay ahead. -->
Midweek closer
Wednesday is the day to stop reading and start auditing. Run the prompt above on your own role before someone in finance runs a similar one on your headcount. The operators who survive this cycle aren't the ones who avoid AI — they're the ones who own which bucket their work lives in.
See you tomorrow.
— The Indexed team
Worth a click:
Forwarded this? Indexed is the daily AI brief for marketers, PMs, consultants, and operators. Reply and tell us what you'd cut from the stack.