Auto-Pilot: Run Your Entire Content or Dev Workflow in One Click

# Auto-Pilot: Run Your Entire Content or Dev Workflow in One Click

Most AI tools hand you a blank chat box and leave the orchestration to you. You write the research prompt, copy the output, paste it into the writing prompt, check for accuracy yourself, and repeat. It works — but it is slow, inconsistent, and entirely dependent on you remembering every step.

Auto-Pilot is the answer. It is a multi-step automation surface that takes a single brief and runs an entire pipeline to completion: research, writing, fact-checking, code generation, review, tests — whatever your chosen workflow demands. The old content-pipeline and dev-pipeline tools have been consolidated here as dedicated tabs alongside a fully customisable Custom brief mode, so every automation pattern lives in one place.

---

How Auto-Pilot Works

Auto-Pilot is built around three concepts:

Presets are pre-wired multi-step workflows. Each preset chains together specialist AI models — one for research, one for writing, one for fact-checking, and so on — passing outputs downstream automatically. You pick the preset that matches your goal, drop in a brief, and press Launch Auto-Pilot.

Pipelines (the Content pipeline and Dev pipeline tabs) are streamlined, purpose-built versions of the same idea. They expose a tighter UI for the two most common use cases and run slightly faster because the configuration is already baked in.

Supervised Mode is an optional overlay you can toggle on in the Custom brief tab. When enabled, Auto-Pilot pauses after each step and shows you the output before continuing. You can approve it as-is, edit it inline, ask for a re-run with feedback, or skip the step entirely. This is the right choice for sensitive deliverables or any time you want a human checkpoint in the loop.

Credits are consumed as each step runs. The exact cost is shown before you launch — the CreditCostBadge next to the Launch Auto-Pilot button updates live when you switch presets or toggle options. Free and Starter plans can run lighter presets; Pro, Power, and Business plans unlock the full preset library and higher concurrency.

---

Using the Content Pipeline Tab

The Content pipeline tab (URL: /os/auto-pilot?panel=content) is the fastest way to produce a publication-ready article from scratch. It chains three specialist models in sequence:

• Deep Research — Perplexity Sonar Pro searches the web and synthesises findings into a structured brief with sources and key data points.
• Fact Check — Claude Sonnet cross-verifies every claim in the research output, marking each one as Verified, Needs Context, or Unverified.
• Write Article — GPT-5.2 drafts the final article using only the verified claims, producing an SEO-optimised post with headings and a meta description.

The total cost is shown on the Run Pipeline button before you start. Each stage displays its output as it streams in, collapsed into a labelled accordion once the next stage begins — so you can inspect the research or verification at any time without losing your place.

Step-by-step:

• Open Auto-Pilot and click the Content pipeline tab.
• Type your topic into the input field — the more specific the better. "The impact of AI on healthcare in 2026" will produce tighter research than "AI in healthcare."
• Press Enter or click Run Pipeline.
• Watch the stage indicator pulse through Deep Research → Fact Check → Write Article.
• When the pipeline reaches done, your article appears in the Final Article panel. Click the copy icon to grab the full text.

You can expand any stage accordion to read the underlying research or the fact-check verdict. If a claim is marked Unverified in the verification panel, the writer will have omitted it — that is by design.

---

Using the Dev Pipeline Tab

The Dev pipeline tab (URL: /os/auto-pilot?panel=dev) runs the equivalent workflow for software: requirement in, production-ready code plus a full test suite out. Three models handle distinct jobs:

• Write Code — GPT-5.2 Codex writes clean, typed, production-ready code with error handling and comments.
• Code Review — Claude Sonnet audits the generated code for bugs, security issues, performance problems, and style violations, rating the overall quality out of ten.
• Write Tests — DeepSeek R1 writes comprehensive unit tests targeting over 90% coverage, taking the reviewer's findings into account when generating edge-case tests.

Step-by-step:

• Open Auto-Pilot and click the Dev pipeline tab.
• Describe what you need built in the textarea. Be specific about the language, framework, and any constraints — for example: "A REST API endpoint in TypeScript (Express) that handles user authentication with JWT tokens and rate limiting, returning proper HTTP status codes."
• Click Run Dev Pipeline. The button shows the total credit cost upfront.
• The Generated Code panel streams in first. Once the reviewer starts, that panel collapses to a summary.
• When complete, the Generated Tests panel shows the full test suite. Use the copy button in the top-right of the tests panel to grab both the code and tests together in one clipboard operation.

The code review panel is highlighted in amber — worth reading before you ship anything to production. The reviewer will flag specific lines and suggest fixes, and those same issues feed directly into the test suite so edge cases are not missed.

---

Using the Custom Brief Tab with Presets

For everything beyond the two built-in pipelines, the Custom brief tab gives you access to the full preset library. Available presets include:

Step-by-step:

• Open Auto-Pilot — you land on the Custom brief tab by default.
• Click the preset card that matches your goal. The card shows the step chain and estimated credit cost.
• Fill in Your Brief with as much context as you can. For a Blog Post Pipeline you might write: "Write a blog post for SaaS founders about reducing churn in the first 90 days. Conversational tone, data-backed, with actionable takeaways."
• Optionally set Tone and Audience from the dropdowns. These are passed to the writing steps.
• Toggle Supervised Mode on if you want to review each step before it continues. Leave it off for fully automated runs.
• Toggle Hallucination-Free Mode on if accuracy is critical. This adds a verification pass after each verifiable step, costing an additional 10 credits per verifiable step. The badge next to the switch shows the added cost before you commit.
• Check the credit cost badge next to Launch Auto-Pilot, then click it.
• The run view opens with a live progress bar. Each step shows its label and status. If you enabled Supervised Mode, a yellow Review panel will appear after each step.

Supervised Mode in action

When a step completes in Supervised Mode, you will see the step output in a scrollable panel with five action buttons:

• Approve — accept the output and continue to the next step.
• Edit — opens an inline editor so you can modify the output before it is passed downstream.
• Re-run — asks the model to redo the step. Add feedback in the text field below the output to steer the re-run.
• Skip — passes the step with no output (use with care — downstream steps may lack context).
• Copy — copies the step output to your clipboard without affecting the run.

Once the final step completes, the Final Output card appears with a Copy Final Output button. Your run is also saved and listed under Recent Runs on the launch screen so you can come back to any previous output.

---

Tips for Better Results

Write a specific brief. The single biggest lever on output quality is brief quality. Include the target audience, desired tone, key points to cover, word count goals, and anything the pipeline must not do. Vague briefs produce generic output.

Use Tone and Audience selectors. "Executives" as the audience shifts the writing style toward concise, data-led language. "Conversational" tone with "General" audience produces something closer to a personal blog. These settings propagate through every writing step.

Prefer Supervised Mode for first runs. Run a new preset in Supervised Mode once to understand what each step produces. Once you trust the pattern, switch to fully automatic runs.

Hallucination-Free Mode is worth it for anything you will publish. The extra verification credits are a small price compared to the cost of publishing a factual error. It is particularly valuable for the Research Report and Blog Post Pipeline presets.

Check Recent Runs. Every run is stored. If a pipeline fails mid-way — network blip, credit shortage — click the run card to return to its progress view. Completed steps are preserved.

Pair with other tools. Auto-Pilot produces raw deliverables. Use Campaign Studio to generate matching visuals for your content output, or hand off a polished brief to Swarm Builder if you want to build a recurring automation around the same workflow. Workflows is another natural next step for sequencing Auto-Pilot runs with other actions.

---

Which Plans Support Auto-Pilot?

Auto-Pilot is available on all plans including Free, but the preset library and Hallucination-Free Mode scale with plan tier. Starter plans cover the lighter presets (Email Sequence, Pitch Deck). Pro unlocks the full preset library. Power and Business plans add higher concurrency, priority model routing, and increased credit allocations that make running several pipelines per day practical.

Check pricing for the current credit bundles and plan limits.

---

Get Started

The fastest way to see what Auto-Pilot can do is to run it. Open Auto-Pilot, pick the Content pipeline tab, type in a topic you care about, and click Run Pipeline. Three models will do the work. By the time you have made a coffee, you will have a research brief, a fact-check report, and a publication-ready article waiting in the final output panel.

For more complex deliverables, switch to the Custom brief tab, pick a preset, and write the most detailed brief you can. The more context you give, the less editing you will need at the end.