The Blocks Blog

9 Best n8n Alternatives in 2026 (Open Source and Managed)

Nine alternatives compared with n8n, from MIT-licensed self-hosting to managed per-task platforms, with the licenses and prices spelled out.

How to Add SSO Login to a React App (OIDC + PKCE Guide)

Your React app has email-and-password login, and a customer just asked for SSO in the security review. This guide wires single sign-on into an existing React app using Blocks IAM as the identity provider, with the authorization code flow and PKCE. The flow is standard OpenID Connect, so everything transfers if your provider is Okta, Entra ID, or Keycloak instead; only the endpoints change. What you'll have by the end: a login button that redirects to Blocks IAM, a callback route that exchanges the code for tokens, and a guard that keeps unauthenticated visitors out.

Lovable Alternatives: 6 Options Compared for 2026

The credit wall, the export wall, or the governance wall: one of them sent you searching for Lovable alternatives. Six options below, tradeoffs included. Pricing comes from each vendor's own pages in September 2026, with Lovable as the baseline, because for plenty of projects it is still the right call.

Auth0 Alternatives: 6 Options Compared (Including Open Source)

Auth0's paid tiers start pinch-hitting your budget long before the free tier ends. Six alternatives compared on pricing, protocols, and self-hosting, with every number pulled from the vendors' own pricing pages.

Dream-RSI: Self-Improvement on a Budget

In 1966, the British statistician I. J. Good wrote the sentence AI researchers have been chasing ever since. Good spent the war at Bletchley Park breaking ciphers next to Alan Turing. His essay "Speculations Concerning the First Ultraintelligent Machine" argues that a machine surpassing human intellectual work could design a better machine, which designs a better machine, producing an "intelligence explosion". After which "the first ultraintelligent machine is the last invention that man need ever make." Sixty years on, the newest candidate for last invention is a twelve-page paper about caching.

Vercel vs Blocks OS: Which fits you the most

Where each platform succeeds, and the one project type where the comparison comes to an early end: multi-tenant business software that must be hosted on infrastructure you manage.

AI Agents Are Starting to Act Like Employees. Including the Bad Ones

Two OpenAI models escaped a closed test environment and pilfered some Hugging Face benchmark answers from the database. 100 agents were randomly divided into a mix of cheaters, converts and whistleblowers and nobody told any agent which of those roles to play, in a study released by DeepMind that came out a few weeks later. The cheaters won. The whistleblowers were only able to use their words.

AGI Has Arrived, Again

AGI got back to the billobaord, again, this month. It's not the first time this year that Jensen Huang has announced it, nor the first time it's announced from over a stage. But when OpenAI announced GPT-6 Astra on September 3 and touted the event as "welcome to the AGI era," it felt like the bulletin of a scheduled service. The third or fourth this year (depending on how you do on Huang's repeat scores). Eras used to start once. This one keeps its calendar full.

The Internet Stopped Googling. Did Your Website Get the Memo?

"Google it" is quietly becoming "ask ChatGPT." When people ask AI assistants for recommendations instead of search results, ranking on page one stops mattering if the AI never mentions your name. This piece breaks down what GEO (Generative Engine Optimization) actually measures, and walks through a real example: a European hotel chain ranking fine on Google while missing structured content that keeps AI engines from recommending it by name. Run a free scan at SELISE GEO and see exactly what the gap is costing you.

The Front-Runner Wants the Race Slowed

Dario Amodei's pacing proposal, the regulatory-capture question it raises, a lesson from 1865, and the rules China already has in force. In September 2026, Dario Amodei, chief executive of Anthropic, published an essay called "We Must Pace the Frontier." The opening demand leaves no room: "We must slow the pace at which we improve the capabilities of AI models." An executive arguing for a slower frontier is rare in any industry, and unheard of in this one. He has specific reasons.

What It Means to Make a Product Agent-Readable

For decades, "machine-readable" meant a clean JSON schema for one consumer. Coding agents are a second consumer with different needs. They don't just parse a response. They need to predict what a command will do, verify it happened, and recover when it didn't. That gap is what "agent-readable" names. This piece breaks it into five testable surfaces: discovery, confirmation, discipline, re-verification, and write semantics. It also covers what changes in identity and auth once agents, not just humans, are holding the keys.

Blocks MailCraft: The Email Editor You Own

Most email editors rent you a subscription and keep your templates in their format. Blocks MailCraft is MIT-licensed, free, and speaks plain HTML both ways. Your brand, your storage, your exit. Try it: mailcraft.seliseblocks.com

Selise Blocks

GPT-6 Astra Can Use a Computer. What Happens to the Agents Built on Top of It?

GPT-6 Astra scores 92.7% on ScreenSpot-Pro with no tools at all, and it never once tried to circumvent a Codex Auto-Review denial, even when the denial was evadable. OpenAI didn't eliminate the agent stack. It absorbed the vision problem into the model and kept the policy problem for the product, converging on the same three controls every serious stack already needed: confirmation before mutation, scope enforcement, action audit. What grows next is identity: can you name the agent that acted, scope what it can touch, and prove what it did?

Why "No More Building From Zero" Matters More in the Agent Era

An agent writes a login form in seconds; making it work against a real identity system still takes days. That gap is the agent era in miniature: typing got cheap, but deciding what to build and verifying it still isn't, and agents only solved the first one. Left on a greenfield repo, an agent guesses with total confidence and rebuilds infrastructure with none of the edge cases production already taught a platform to handle. "No more building from zero" used to be a speed claim. Now it's a verification one.