AI tools for workflow automation
Which automation tool can run a branching workflow with an AI step on the plan you will pay for? Compare billing units, free-plan step limits, failure recovery and hosting, then judge each tool on one proposed test.
Which tool fits, in short?
Use Zapier if you connect many apps and want failed runs replayed from Zap history. Make fits branching logic drawn as a diagram, with fallback routes and per-module error handlers. n8n suits teams that want self-hosting, since a full run counts as one execution. Zapier Free allows only two-step Zaps. A branch plus an AI step needs a paid plan.
This is a source-led shortlist, not a performance ranking. Open each profile for the provider source, check dates and plan conditions.
About this shortlist
What this page compares
- 5 published profiles
- Plans and access shown where documented
- Official source dates stay attached to each profile
- No invented score or hands-on result
How each option fits workflow automation
Each note comes from the tool's source-checked profile. The condition line is what to confirm before you rely on it for this task.
Zapier
Connects a trigger in one app to actions in others across a large app catalog, with AI steps that summarize, classify or draft text mid-workflow; Zap history logs every run and lets you replay unsuccessful ones.
Check first: Free is limited to 100 tasks a month and two-step Zaps, so a branch plus an AI step needs Professional (USD 19.99 per month billed yearly), and each AI step uses more than one task.
Make
Routers and filters draw each branch as a diagram with a fallback route for unmatched records, and module-level error handlers (Skip, Retry, Resume, Commit, Rollback) decide what happens when one step fails.
Check first: Free covers up to 1,000 monthly credits, 2 active scenarios, a 15-minute scheduling floor and 7 days of logs; paid plan names and prices vary between public views, so confirm the quote at checkout.
n8n
Counts one execution per complete workflow run however many nodes it has, and adds Code nodes, HTTP requests, error workflows and retry of failed runs, on n8n Cloud or your own servers.
Check first: The only ongoing free route is self-hosted Community, where servers and upkeep are yours; the Cloud trial lasts 14 days, and Starter (EUR 20 per month billed annually) caps concurrent runs at 5.
Bardeen
Fits workflows whose first stage is collecting rows from web pages: it scrapes, enriches and AI-qualifies each row, then exports to Google Sheets, Airtable, Notion or CSV for the next tool.
Check first: Credits are charged per output row (enrichment costs 3 per row), Basic is scraper-only, and runs while the browser is closed are described only for paid services.
Copy.ai
Workflows chain research, scraping and writing actions into a repeatable process started from a form or the API, with Brand Voice and Infobase supplying the same context every run.
Check first: There is no free plan; building and test-running a workflow uses no credits, but monthly workflow credits start with Growth (USD 1,000 per month billed annually).
Considered but not listed
- ChatGPT: ChatGPT can run scheduled tasks and, on eligible plans, supported app-triggered tasks. This shortlist focuses on dedicated automation platforms with explicit routing, run history and workflow billing; app support and account access still need their own check.
- Jasper: Grid runs prompts and agents across spreadsheet rows for bulk content, which is a content pipeline rather than app-to-app automation, and it is listed for Business only.
- Claude: Claude can serve as an assistant or connected-tool client within an automation setup. This page compares the dedicated workflow platform that owns routing, execution history and billing rather than the chat interface.
Compare tools for workflow automation
Use the profiles below to inspect the specific workflow, pricing and limits before making a choice.
Zapier
AutomationZapier is a no-code automation service that connects web apps so an event in one triggers actions in others, billed by successful tasks.
Make
AutomationMake is a cloud automation builder that links app modules in visual scenarios and bills each module run in credits.
n8n
Automationn8n is a workflow automation platform you can run on managed Cloud or self-host for free, billed per whole workflow run rather than per step.
Bardeen
AutomationBardeen is a Chrome extension that scrapes websites, enriches contacts and qualifies leads with AI, billed in credits per output row.
Copy.ai
Writing & contentCopy.ai is a sales and marketing content platform that pairs a multi-model AI chat with brand-trained content agents and credit-metered workflows.
What the shortlist costs to start
Cheapest paid plan billed month to month, in USD, as listed on each profile.
Bardeen$10/mo
Copy.ai$29/mo
Zapier$29.99/mo
What to compare for this task
These questions keep the shortlist tied to the work instead of turning it into a generic feature list.
Step budget
Does the plan you can afford allow the number of steps, branches and AI actions this workflow needs?
Billing unit
Is usage counted per action (task), per module run (credit) or per full run (execution), and what does your monthly volume come to?
Failure recovery
When one step fails, can you find it in run history and retry it without rerunning the steps that already succeeded?
Hosting
Must the workflow run on infrastructure you control, or is a hosted service acceptable?
Compare the shortlist
Values come from the linked profile fields. Blank or unverified details stay visible rather than being filled by assumption.
| Documented detail | Zapier | Make | n8n | Bardeen | Copy.ai |
|---|---|---|---|---|---|
| Access | Free access available | Free access available | Free access available | Free access available | No free tier |
| Paid plans from | USD 19.99 per month, billed yearly (USD 29.99 billed monthly; 750 tasks) | Free + paid; quote varies | EUR 20 per month, billed annually | $10 per month, billed monthly (100 credits) | USD 24 per month, billed annually as USD 288 per year (USD 29 billed monthly) |
| Best starting point | Connecting web apps without code | Visual automation for app workflows | Self-hostable workflow and AI-agent automation | Building and enriching lead lists from websites | Brand-consistent sales and marketing content for teams |
| Source status | Source-checked: 24 Sep 2026Pricing checked: 24 Oct 2026 | Source-checked: 24 Sep 2026Pricing check incomplete | Source-checked: 24 Sep 2026Pricing checked: 24 Sep 2026 · Update due | Source-checked: 24 Sep 2026Pricing checked: 24 Sep 2026 · Update due | Source-checked: 24 Sep 2026Pricing checked: 17 Oct 2026 |
| Profile | Zapier profile | Make profile | n8n profile | Bardeen profile | Copy.ai profile |
Head-to-head comparisons for this task
Both tools in each comparison are on this shortlist.
One fixed task to compare them
Proposed evaluation, not performed: run the same input through two or three shortlisted tools and judge each against the same acceptance line. We have not run it; results depend on your plan and account.
Sample input
Ten order records, IDs O-01 to O-10, each with name, email, product and amount. O-03 and O-07 have no email and O-09 has an amount of 0; the other seven are complete. Complete orders go to an AI step that writes a one-sentence confirmation note; the three flagged orders go to a review list.
Steps
- Download the exercise files linked on this page. Keep the supplied acceptance rules separate from your observations; record results only after running the exercise on your actual account.
- Check that the selected plan supports the trigger, branch, AI step and destination. Zapier needs paid multi-step access for this flow; model-provider usage may be separate. Bardeen is evaluated only for collection and qualification handoff, not assumed to own every routing or retry step.
- Build the flow: trigger (new row, webhook, form or import, whichever your plan allows), branch on missing email or zero amount, AI step, write the note back. Save confirmation notes only to a test destination; do not send emails or customer messages.
- Feed in the ten records, one run per record.
- Break the destination for O-05 (for example, rename its target column), let that run fail, restore the column and retry the run from the run history.
- Read the run log and record how many tasks, credits or executions the ten records used.
- Create a run ledger with order ID, route, AI called yes/no, destination record ID, attempt number, final state and billed units. For O-05, preserve the first failed attempt and the retry as separate entries tied to the same order ID.
- Before a retry can write a note, check the destination for the order ID and update or skip an existing record. If the platform cannot implement that guard on the chosen plan, record the limitation instead of claiming duplicate protection.
- Passes if
- Seven notes exist, one per complete order; O-03, O-07 and O-09 sit on the review list and never reached the AI step; O-05 succeeds on retry without duplicating other rows; the usage figure is read from the tool's own run log. The ledger attributes O-05 retry cost separately, and any unavailable stage is reported instead of being treated as a failed AI answer.
- Fails if
- Any order gets two notes, a flagged order reaches the AI step, the retry reprocesses successful records, or the flow works only after an upgrade you had not planned.
Take sample records into a test workflow
Start with the ten-order exercise above. The pack includes expected routes, a blank retry log and a separate five-event form exercise for duplicate, missing-field and formula-like input checks.
Original editorial exercise files. Expected outcomes are acceptance rules; observation fields are blank. No hands-on product results are included.
No account required to download. Your completed files stay with you.
Read the exercise stepsGuides for this task
Browse by category
The categories of the shortlisted tools, with every profile in each.
Questions about workflow automation with AI
Does a ten-step workflow cost ten times a two-step one?
It depends on the billing unit. Zapier uses a task for each successful action step, AI steps use more than one, and triggers, filters, paths and delays use none. Make normally charges one credit per module run, and empty scheduled checks count too. n8n counts one execution per complete run, however many steps it contains.
What happens to my workflows when the n8n Cloud trial ends?
If you do not upgrade, the 14-day trial expires and its workspace is removed. Workflows can still be downloaded for 90 days after the trial ends, so export anything you want to keep, then either choose a paid Cloud plan or import the files into a self-hosted Community installation that you run yourself.
Is the AI model billed separately from the automation plan?
Sometimes. In Make, an eligible custom AI-provider connection means the provider bills its tokens in addition to the credits Make charges, while Make-managed AI is billed differently. Zapier prices AI steps by model tier inside its task count. Copy.ai workflows draw on monthly workflow credits, which start with the Growth plan.
How quickly does a workflow react to new data?
On Zapier, polling checks for new data every 15 minutes on Free, every 2 minutes on Professional and every minute on Team, and webhooks are listed with Professional. Make scenarios can start as soon as a webhook receives a request, while scheduled scenarios on Free run at most every 15 minutes. Choose by how late a hand-off may arrive.
What are AI automation workflows?
An AI automation workflow is a repeatable chain that starts with a trigger, such as a new row, a webhook or a form, and then runs actions in other apps. At least one step is an AI step that summarizes, classifies or drafts text in the middle of the run. The exercise on this page uses a trigger, a branch, an AI-written note and a write-back.
Sources for this task
These checks cover the task conditions below. Each linked profile keeps its own feature and pricing evidence.
- Make webhook processing orderhelp.make.com · Checked 2026-09-27
Default parallel processing and Process data in order; not a plan price or free-access check.
- n8n error handlingdocs.n8n.io · Checked 2026-09-27
Error workflows and intentional failure; not a completed workflow test.
- Google Sheets input modesdevelopers.google.com · Checked 2026-09-27
RAW preserves values without parsing in the separate form-input exercise; connector access must be checked.
How to use this shortlist
Start with the task fit, then open the profile whose documented workflow matches your constraints. Check the source date, plan terms and limits on the provider site before you subscribe or publish an output.



