Find software for the way you work.
Explore products, comparisons and essays. Twin Scores are legacy editorial estimates.
What Twin Scores are
Twin Scores are legacy editorial estimates on a 0–100 scale. No firsthand product-test log or per-dimension scoring worksheet is attached to these retained scores. Dated documentation checks establish only the specific product facts they cite; they do not validate the scores. The rubric is public on the scorecard. We do not emit AggregateRating for Twin Scores. A scored review uses its own audience score on the native 0–100 scale.
Featured products by human Twin Score
design
Figma
Figma is the browser-native design tool that ended the era of emailed .sketch files. Multiplayer, components, and Dev Mode are the human triumph. REST APIs, plugin APIs, and variables help agents read files, export assets, and lint design systems — but Figma is still a.
Legacy editorial estimates; no firsthand test record attached.
developer-tools
Visual Studio Code
Visual Studio Code is the world's default editor: free, extensible, and fast enough. The human surface is the workbench plus the marketplace. The agent surface is the extension API, language server protocol, and tasks/debug adapters — machines already drive VS Code via those.
Legacy editorial estimates; no firsthand test record attached.
developer-tools
Linear
Linear is the issue tracker designed like a fast native app: keyboard-first, opinionated workflow, gorgeous graphs. Humans who have suffered Jira feel the difference in a day. The GraphQL API, webhooks, and synced GitHub PRs make it agent-friendly. There is no official CLI of.
Legacy editorial estimates; no firsthand test record attached.
developer-tools
GitHub
GitHub is the default public source host and a full DevOps surface: issues, PRs, Actions, Packages, Codespaces, and security scanning. Humans live in the PR timeline. Agents live in the REST and GraphQL APIs, Checks, and workflow YAML. Few products have a more complete twin:.
Legacy editorial estimates; no firsthand test record attached.
developer-tools
Cursor
Cursor is an AI-native code editor forked from VS Code. Cursor documents an IDE and a separate Cloud Agents API. The API reference checked on 2026-10-16 documents POST /v1/agents with user or service-account API keys using Basic or Bearer authentication. No agent run was performed for this audit. Agent score is high because of rules files, SKILL-like.
Legacy editorial estimates; no firsthand test record attached.
design
Canva
Canva is the design tool for everyone who is not a professional designer — and increasingly for those who are. Templates, Brand Kit, and Magic Studio made visual work default worldwide. Partner and Connect APIs exist, but most automation is still export-and-hope. We will not.
Legacy editorial estimates; no firsthand test record attached.
ai-assistants
ChatGPT
ChatGPT is OpenAI's consumer and business assistant, wrapping GPT models in a chat UI with memory, browsing, image generation, and custom GPTs. The public product launched in November 2022 and became the default way millions of people try large language models. The API lives.
Legacy editorial estimates; no firsthand test record attached.
infrastructure
Vercel
Vercel is the company behind Next.js and a hosting platform optimized for frontend and serverless compute. Preview URLs for every PR are the human delight. The REST API, CLI, and deploy hooks are how agents ship. We score both high and still note vendor coupling: the best.
Legacy editorial estimates; no firsthand test record attached.
productivity
Superhuman
Superhuman is a fast, keyboard-first email client on top of Gmail and Microsoft 365. Split inbox, reminders, and snippets are the product. There is no public agent API to drive Superhuman itself — agents should hit Gmail or Graph instead. We include it because it is globally.
Legacy editorial estimates; no firsthand test record attached.
infrastructure
Clerk
Clerk sells beautiful auth components plus a user management backend, especially for Next.js and B2B organizations. Humans get a polished sign-in in an afternoon. Agents get Backend APIs and webhooks. Compared with Auth0 it is more frontend-opinionated and less.
Legacy editorial estimates; no firsthand test record attached.
automation
Zapier
Zapier is the no-code integration fabric: Zaps, Tables, Interfaces, and now agents that run the same connectors. Humans who are not engineers automate Gmail-to-Slack here. Developer Platform and REST let agents create and run Zaps, but the soul of the product is the visual.
Legacy editorial estimates; no firsthand test record attached.
marketing
Shopify
Shopify is the worldwide commerce OS for independents and brands: admin, checkout, POS, and a huge app store. Humans set up a store in a weekend. GraphQL Admin and Storefront APIs, plus webhooks, make it a first-class agent commerce backend. Liquid themes remain a human.
Legacy editorial estimates; no firsthand test record attached.
About the catalog
Editorial · Worldwide · English · Catalog attribution: Edu Tor
Build Something People Want is a worldwide editorial directory of software products scored for humans and agents. We retain legacy editorial estimates, catalog bylines and essays — not crowdsourced star ratings or invented user counts. The twin site, Build Something Agents Want, keeps the same entity slugs for machines.
Every product page answers a buying question in the opening paragraph, then walks H2 to H3 with a capability snapshot, pricing as publicly stated, and links to alternatives. Public catalog pages render from committed seed JSON, so the site is readable before secrets land — including by AI crawlers that do not execute JavaScript or a 3D canvas.
Why two sites
People and agents do not want the same page. People want prose, comparisons, and a magazine that does not look like purple SaaS. Agents want matrices, SKILL.md, CSV, and OpenAPI. Same graph, two contracts.
Categories
- AI assistants — Chat and reasoning products used as daily copilots. Best AI assistants
- Developer tools — Editors, git hosts, issue trackers, and coding agents. Best Developer tools
- Design — Interface, illustration, and generative visual tools. Best Design
- Productivity — Notes, docs, mail, and calendars for teams. Best Productivity
- Infrastructure — Hosting, data, auth, and edge networks. Best Infrastructure
- Automation — Workflow builders and integration fabrics. Best Automation
- Marketing and CRM — CRM, CMS, commerce, and go-to-market suites. Best Marketing and CRM
- Communications — Chat, video, and async recording. Best Communications
- Payments — Billing, invoicing, and money movement. Best Payments
- Data and analytics — Product analytics, warehouses, and observability. Best Data and analytics
Catalog analysis and essays
Start with the journal and the 2026 human-versus-agent software surfaces table. The tables summarize retained estimates; they do not validate product performance. The bylines are catalog attributions; this evidence register contains agent documentation review dated 2026-10-16, with a responsible human editor unassigned. Professional history and firsthand product evaluations are not independently verified in this evidence register. Recorded bylines: Edu Tor, Maya Chen, Jonas Okonkwo.
Latest essays
- Human vs agent software surfaces in 2026: retained Twin Score estimates
- How the Twin Score rubric actually works
- Fake UGC is an SEO tactic that this directory refuses
- The capability matrix agents actually need
Agents should go to Build Something Agents Want developer resources. Public JSON API on this origin: GET /api/v1/products · openapi.json · llms.txt. Errors use RFC 9457 application/problem+json with code, detail, and status.