Maor Shlomo built an $80M company solo — the playbook is now anyone's
Agentic tools have collapsed the team size needed to ship a real product; across tech, freelancing, and professional services, workers who treat that shift as an invitation are building things that weren't possible two years ago.
The story
Six months after founding Base44, Maor Shlomo — a 31-year-old Israeli developer — accepted an $80 million acquisition offer from Wix. He had no investors, fewer than ten employees, and a product that let non-technical users build software by describing it in plain language. By May 2026, Base44 was tracking $150 million ARR. Shlomo built the feedback-triage system, the QA agents, and the UX audit tooling himself — work that would have required a team of 20 a few years earlier. His story sits at the sharp end of a broader shift: across sectors, workers who treat AI as a force multiplier are finding themselves able to do things that previously required a department. The Challenger report shows AI led all stated reasons for US job cuts in both March and April 2026 — but Upwork simultaneously reports a 109% surge in demand for AI-skilled freelancers. The disruption and the opportunity are the same event, viewed from different angles.
Industries shifting
| Sector | What changed | Scale |
|---|---|---|
| Customer support | Klarna's AI agent reached the equivalent of 853 FT agents, handling two-thirds of inquiries and saving $60M. But CEO Sebastian Siemiatkowski publicly admitted the aggressive cuts hurt service quality — Klarna is now hiring human agents again, settling on a hybrid model where AI handles volume and humans handle escalation and complex cases. | ~853 FTE-equivalent / month (Klarna, company-announced); hybrid model now the stated baseline |
| Tech sector layoffs (US) | Challenger, Gray & Christmas: AI led all stated reasons for US job cuts in both March and April 2026. April alone: 21,490 AI-attributed cuts (26% of 83,387 total). YTD through April: 49,135 cuts explicitly attributed to AI replacement — up from 13% of all cuts in March to 16% in April. | 49,135 AI-attributed US cuts YTD April 2026; tech sector total 85,411 cuts Q1+partial Q2 (Challenger) |
| Software development | Claude Code scores 78.4% on SWE-bench Verified; Replit Agent 3 plans, codes, debugs and deploys full-stack apps from a text prompt. Most production engineers now run two agentic coding tools simultaneously. Entry-level coding tasks are increasingly handled without junior dev involvement, compressing the training pipeline. | 5 agentic coding tools with SWE-bench scores above 54% now in market; 78.4% top score (Claude Code) |
| Marketing and copywriting | 76% of marketers using generative AI report using it for basic content creation and first-draft long-form copy. Agentic marketing platforms like Tofu (B2B) report shipping integrated campaigns up to 8x faster. Gartner projects 40% of enterprise applications will embed AI agents by end of 2026. | 76% GenAI adoption for content creation; 40% enterprise app AI-agent embedding by EOY (Gartner) |
| Freelance knowledge work | Upwork In-Demand Skills 2026: demand for AI-related freelance skills grew 109% YoY. AI video generation up 329%; AI integration services up 178%. Freelancers using AI report deliverables that took 6 hours now take ~2.5 hours at unchanged rates — effective hourly profit nearly triples for early adopters. | 109% YoY demand surge (Upwork 2026); 84% of freelancers now use AI tools, up from 41% in 2023 (Freelancer Kompass 2026) |
| Radiology and clinical imaging | A decade after Geoffrey Hinton declared radiologists obsolete, US radiologist headcount has grown ~10% and median salaries reach up to $571K (Fortune, May 4 2026). Radiologists now use AI for scan prioritisation and report summarisation — the role changed, demand grew, and the specialisation premium held. | ~10% headcount growth and $571K median salary peak despite a decade of 'AI replacement' forecasts (Fortune, May 2026) |
The data, today
| Source | Metric | Value |
|---|---|---|
| Layoffs.fyi | Tech and startup layoffs YTD 2026 (as of mid-May) | 138,988 workers across 327 companies (~993 per day) |
| Challenger, Gray & Christmas (April 2026 report) | US job cuts attributed to AI, April 2026 | 21,490 (26% of all April cuts); 49,135 YTD through April |
| BLS Employment Situation, April 2026 | US total nonfarm payroll change and unemployment rate | +115,000 jobs added; overall unemployment rate unchanged at 4.3% |
| BCG AI Workforce Report (April 2026) | Share of jobs expected to be significantly reshaped by AI within 2-3 years | 50–55% reshaped; 10–15% (16–25M US positions) displaced over 5 years |
| Upwork In-Demand Skills 2026 (published Feb 2026) | YoY growth in demand for AI-related freelance skills on Upwork | 109% YoY overall; AI video generation +329%, AI integration +178% |
| WEF Reskilling Revolution (January 2026) | Global workers reached by reskilling initiative; corporate commitments | 850M+ on track; leading tech companies collectively pledged to support 120M workers by 2030 |
Riders, not victims
Maor Shlomo
Built Base44 — an AI 'vibe coding' platform that lets non-technical users create software through text prompts — without external investors. Used AI agents to handle user feedback triage, QA testing, and UX auditing. Founded, scaled, and sold the company as sole shareholder with fewer than 10 employees.
Outcome: Acquired by Wix for $80M six months after founding; Base44 tracking $150M ARR by May 2026. Shlomo was the sole shareholder.
“The team size required to ship a real product has dropped by an order of magnitude — but only for founders who delegate to agents, not just assistants”
Danny Postma
Built HeadshotPro — an AI professional headshot generator — as a solo operator without a traditional team. Runs a portfolio of ~20 AI-powered products under his studio Postcrafts, using AI throughout product development, customer support, and distribution.
Outcome: $3.6M ARR from HeadshotPro; portfolio generating ~$300K/month MRR at peak — operating essentially alone
“Owning the niche at the intersection of a specific job-to-be-done and an AI capability is more durable than building generic AI features”
Nick Dobos
Built BoredHumans.com — a single domain hosting 100+ free AI tools, from creative generators to interactive games and image tools. Monetised through ads and premium tiers with SEO-driven traffic acquisition and minimal operational overhead.
Outcome: ~$733K/month (~$8.8M ARR) running the platform as a one-person operation
“Volume of useful tools on a single domain compounds SEO value; AI makes the build cost near-zero, so the constraint becomes distribution, not development”
Pattern: mid-career knowledge worker pivoting to AI workflow freelancing (composite profile from Upwork 2026 survey of 4,360 freelancers and Freelancer Kompass 2026 — not a single named individual)
Learned AI workflow tooling — prompt engineering, Claude Code, Zapier AI — over 2–3 months using free resources (Hugging Face, DeepLearning.AI). Left a salaried role to freelance as an 'AI workflow specialist', managing automation stacks for 3–5 agencies simultaneously.
Outcome: Survey respondents in this cohort earn ~40% more per hour than traditional freelancers in the same discipline (Upwork 2026 survey)
“Moving up the stack — from producing the artifact to managing the system that produces it — is the leverage move of this cycle”
Tools workers are picking up
| Tool | What it changes | Cost |
|---|---|---|
| Cursor | AI-native IDE augmenting (and for many replacing) VS Code + Copilot — inline completions, Cmd+K edits, and agent mode for larger refactors. Most production engineers now run it alongside Claude Code for complementary task types. | $20/mo Pro · $200/mo Business |
| Claude Code (Anthropic) | Terminal-native agentic coding tool scoring 78.4% on SWE-bench Verified — highest of any public model as of May 2026. Handles multi-file refactors, test writing, and autonomous debugging that previously required senior dev hours. Runs natively in terminal with MCP integrations. | ~$100/mo (Claude Max plan) or pay-per-use API |
| Replit Agent 3 | Full-stack scaffolder and deployer — describe a project in plain language, the agent plans, codes, debugs, and deploys. Removes the barrier between 'idea' and 'live app.' Particularly popular with non-technical founders validating product ideas without hiring a developer. | $25/mo (Replit Core) |
| NotebookLM (Google) | Instant Q&A, summary, and synthesis across uploaded document corpora — PDFs, transcripts, research reports. Replaces the 'I'll read this later' research backlog. Increasingly used by analysts, lawyers, and journalists to process large document sets in minutes. | Free |
| v0 (Vercel) | AI-driven UI component generator — describe a React component or a full page in text, receive production-quality code with Tailwind and shadcn/ui. Replaces the frontend scaffolding step for full-stack developers and product designers building MVPs without a dedicated designer. | Free tier · $20/mo Pro |
Worth a Saturday — reskill
DeepLearning.AI 'ChatGPT Prompt Engineering for Developers' (Andrew Ng + OpenAI) · Free, ~3 hours, video + interactive Jupyter notebooks; no signup required
For: Anyone using AI in their workflow who wants to move from casual prompting to structured, repeatable prompt design
Hugging Face AI Agents Course · Free, certified, ~3–4 hours/week across 4 main units; community leaderboard for competitive ranking; no deadline
For: Engineers and PMs who want to build and deploy agents using smolagents, LlamaIndex, and LangGraph — not just chat interfaces
Great Learning Academy 'Prompt Engineering for ChatGPT' · Free, 3 hours, fully self-paced, certificate of completion; 206,800+ learners enrolled as of May 2026
For: Non-technical workers who want a verifiable credential to add to a freelance profile or LinkedIn within a weekend
Coursera 'Generative AI for Everyone' (DeepLearning.AI, Andrew Ng) · Free audit / $49 for shareable certificate, ~4 weeks self-paced
For: Managers, operators, and domain experts who need AI fluency to spec and evaluate AI features without engineering depth
IndiaAI FutureSkills / SOAR programme (Indian government, MeitY) · Free online, delivered via Microsoft India, HCLTech, NASSCOM; 211,000+ enrolled as of May 2026
For: Workers in India seeking government-backed AI credentials tied to active industry hiring pipelines
Threads worth pulling
GitHub Copilot 1.5M+ paying seats → senior dev productivity increases → engineering manager span-of-control widens → fewer EMs needed per IC team → middle-management layer contracts while IC headcount holds steady The displacement is running up the org chart, not down. Individual contributors are shipping more; the coordination layer above them is shrinking. Layoff headlines that say 'AI is killing developer jobs' are pointing at the wrong seniority band.
Klarna replaces ~700 staff with AI agent (2023–2024) → CEO admits quality degraded → hires human agents back in 2025–2026 → new baseline is hybrid: AI handles volume, humans handle judgment and escalation The Klarna reversal is the most public stress-test of full AI replacement in customer service — and it failed on quality. The durable model isn't replacement, it's stratification. Knowing how to design the human-AI handoff is a new, billable skill category.
AI tools lower software build cost to near-zero → solopreneur SaaS becomes viable at $150M ARR → VCs reprice team size assumptions → seed rounds get smaller and outcome-per-founder rises If a 31-year-old solo founder can build an $80M acquisition target in six months with no outside capital, the old assumption — that early-stage software requires a team — is being stress-tested at scale. The economics of starting up are being rewritten from the cost side.
AI agents handle entry-level coding tasks → junior developer intake shrinks → fewer roles serve as training grounds → senior dev pipeline eventually thins → mid-2030s skills cliff for companies that optimised too hard on AI efficiency today The short-term efficiency gain of replacing junior tasks with AI may create a medium-term talent problem. Companies building structured AI-assisted apprenticeship tracks now are treating the pipeline as a strategic asset, not just a cost line.
What we’ll be watching
- Colorado AI Act (CAIA) takes effect June 30, 2026 — US employers using AI in hiring or termination decisions must notify affected workers, conduct annual impact assessments, and publish a public statement on AI systems in use. First major US state law to operationalise worker notification requirements for algorithmic employment decisions.
- BLS May 2026 Employment Situation — releases June 5, 2026 at 8:30am ET. Will show whether April's +115,000 payroll gain held and whether the 4.3% overall unemployment rate has moved in either direction.
- Challenger, Gray & Christmas May 2026 job cut report — due first week of June. Tracks whether AI remains the top stated reason for US employer layoff announcements for a third consecutive month.
- AI-Related Job Impacts Clarity Act (Hawley-Warner) — bipartisan US Senate bill requiring employers to file quarterly Department of Labor reports identifying headcount reductions tied to AI automation. Senate hearings ongoing; passage would make AI-attribution layoff data systemic and publicly searchable rather than voluntary and anecdotal.
- IndiaAI FutureSkills mid-year milestone — India's MeitY is targeting 12.5 lakh (1.25M) AI professionals by 2027 (up from ~6.5 lakh today, 15% CAGR). Mid-year progress report expected June 2026 against this baseline.
- OECD AI-WIPS (AI in Work, Innovation, Productivity and Skills) ongoing report series — multi-country employment impact studies using 2025 labour data; next major publication expected Q3 2026 and will be the first multi-country dataset covering a full year of agentic AI deployment.
Sources: layoffs.fyi, challengergray.com, bls.gov, bcg.com, upwork.com (In-Demand Skills 2026 press release), cbsnews.com, fortune.com, cnbc.com, weforum.org, calcalistech.com, timesofisrael.com, twig.so, deeplearning.ai, huggingface.co, natlawreview.com.