Blog

Aisle is now qbash. AI made judgment programmable and created a new class of application. qbash is the platform where those systems are built, run, and distributed.

The week in AI in one pass: Stripe buys OpenRouter and says the singularity is here, Anthropic's quality complaints turn into churn, a $31,000 lesson in agent autonomy, and Amazon shreds rare books for training data.

A checkout fails after the card is charged. Checkpointing in the Tasks SDK records which steps finished so a retry skips them, and every step runs exactly once.

Integration calls in Aisle Tasks are standard Python API calls, not agent tool invocations. Credentials stay out of the context window and tokens go toward the work.

Build a Task once, expose it as a tool, and call it from Claude Code, Cursor, claude.ai, or ChatGPT. The client passes the inputs, the Task runs on Aisle's managed runtime, and the output comes back in the chat window.

An Airtable base of deals, one Task with four modes, mounted on an MCP server. The sales team updates the pipeline by chatting, without edit access to the table.

Deploy a Task as an MCP server and call it from ChatGPT, Claude, or any client that supports the protocol. Same Task, same behavior, every client.

Mail merge for PowerPoint decks. One template, many rows of data, and a zip of finished decks ready to download.

The most copied lead-gen workflow on n8n scrapes Google Maps and ends in a spreadsheet. Here's the same job rebuilt as an Aisle Task with Google Maps, Apollo.io, HubSpot, and Slack.

We built Tasks so your work stays separate from any one model and the model is a swappable input. On Microsoft's latest earnings call, Satya Nadella described the same architecture.

Anthropic's new flagship is in your model picker today: a 1M token context window, 128K output, and adaptive thinking with full effort control. Plus why your shared Aisle chats can't end up in Google, the Cobalt connector, and a reworked Tasks editor.

Two upgrades to the AI task editor: full create, edit, and remove support for inline prompts, and Postgres connection testing that runs a real read-only query.

How to set up feature and spec memories in Claude Code that survive across sessions and carry across everyone on your team, instead of re-explaining the same architecture every morning.

The best Zapier alternatives in 2026, compared on billing model, hosting, licence and who maintains the automation. Make, n8n, Activepieces, Aisle, Pipedream, Gumloop and Retool.

The AI Task Builder takes a spec and compiles a working Task from it: script, prompts, and wiring, generated end to end.

Building a six-phase competitive intelligence Task with our AI Task Builder made one thing obvious: the model is the smallest part of the work, and the plumbing around it is where the reliability comes from.

A look at how moving from agentic reasoning to a deterministic Python task in Aisle can cut token consumption by up to 50x, with no loss in what the automation actually does.

The best n8n alternatives in 2026, compared on licence, hosting, billing model and what happens after launch. Activepieces, Windmill, Aisle, Make, Zapier, Gumloop and Inngest.

Use an Aisle task to create a deterministic AI automation that summarizes all new commits to a project and delivers them to your team.

Connect an OpenSearch instance to an Aisle project and query your server logs in plain language using AI.

Export controls suspended Claude Fable 5 days after launch. The Commerce Department has lifted them and Anthropic has redeployed the model. The retention terms have not changed, so neither has our position. Plus fourteen connectors, grouped by the function that would own them.

I have been talking about deterministic vs agentic AI for a while, largely with a focus on how it is more reliable. The cost conversation really makes this shine too.

AI assistants and dedicated AI systems do different jobs. One makes a person faster in a session; the other runs a process for the company. Every company should run both, deliberately.

How to build a shared, multi-model prompt library for your team in Aisle - with variables, per-prompt model settings, versioning, sharing, and prompts that run across ChatGPT, Claude, Gemini, Grok, and more.

What shipped this week: a rebuilt Tasks editor with tabs, inline prompts, line-level revisions, and an AI assist; three new connectors and four updates; and two new essays from the blog.

AI does not lift teams evenly. There are two accelerations happening and they are not the same thing. One has a ceiling and a shelf life. The other compounds.

A big iteration on the Tasks editor: tabs for files and prompts, a redesigned sidebar, dirty-state indicators, enhanced revision logs, and an AI assist button.

An X agent built on Aisle that drafts posts and replies, queues them for review, and rewrites its own instructions weekly based on what I approve and reject. It earns autonomy instead of starting with it.

Three ways to build the same automation on Aisle - a two-prompt workflow, a single-prompt workflow, and a native Python Task - and what each one cost in tokens, effort, and maintenance.

Anthropic released Claude Fable 5. It's available in Aisle today, shipped turned off, because Anthropic now retains 30 days of usage on this model regardless of your agreement.

Tasks lets you build AI workflows in Python and run them on a managed runtime. No servers to provision, no infrastructure to wire up.

Tasks are Python scripts that run on Aisle. The runtime supplies the schedule, credentials, retries, prompts, memory, integrations, and audit trail. Live in Aisle today.

Agents are impressive - but once you've run them in production long enough, the trade-offs become clear. Different decisions each run, off days, model personality shifts, and maintenance overhead. Here's how to think about it.

In this video we show using the Playgrounds feature to compare the responses of different models - such as Opus 4.6 and GPT 5.4 - along with different prompt settings.

Anthropic shipped Claude Opus 4.8 today. It's already in Aisle - no setup required.

Workflow canvases fit analysts building reports before LLMs existed. That fit ended when those analysts started using AI assistants. The problem is the canvas shape itself, not its size.

Deterministic vs Agentic, and where your org actually sits on the curve. Two vectors I use to figure out which AI bets are worth making and which ones are giving a dog a gun.

Projects can now export chats and RAG results to PDF or CSV, accept inline file uploads, and surface execution logs. Mixpanel MCP is live, and Memories got a UX refresh.

Most people are treating AI as one thing. It's three different bets with three different risk profiles. The order you take them in matters more than the bets themselves.

Use the Aisle memories feature to create an on the fly, searchable AI RAG database that extends past usual context limits. Upload hundreds of documents including PDF, PPT, DOC and then chat with your data.

The old SaaS moats - accumulated codebases, big teams, hard-won velocity - don't hold up in 2026. The real moat now is how you design your processes, software, and offerings for the new world.

In this tutorial, Mitch takes us through the Aisle Projects feature - how to bundle prompts, workflows, MCP connectors, and knowledge bases into a single agent-run workspace.

Projects puts every Aisle feature into one workspace that an agent orchestrates.

How much time do you spend on prompt structure? Mitchell takes us through a quick demo of the Aisle Prompt Builder, which takes a natural language task description and fleshes it out into the Aisle preferred prompt structure, complete with variables and output structure.

Strong headlines, trickier token math. Compare Opus 4.7 next to what you already run.

It's like Infrastructure Week... But better!

Upload documents, let Aisle index them, and query them from chat, prompts, and workflows in one place. Semantic search, vector descriptions, and real team use cases.

How to build a prompt in Aisle from scratch. Mitch walks through the Prompts feature: setting up a prompt, configuring structured outputs, and deploying it to your team.

All 14 Grok models from xAI, native X Search, Grok 2 Vision, Grok Imagine, and six new image generation options. Let’s get into it.

This week: live news feeds in workflows, a new GPT model, scripts on HTTP endpoints, GitHub and Bugsnag webhook triggers, and two quality-of-life fixes for long-running workflows.