"Superflow AI" almost always points to Superflows, a London-founded startup that lets SaaS teams embed a chat-based AI assistant into their own product. You connect it to your API and docs, and your users can ask questions or trigger actions in plain language instead of hunting through a dashboard.
Short answer: Superflow AI (officially Superflows) is a developer tool that adds a chat assistant to a SaaS product by connecting it to your API and docs. It has a free Testing tier (50 queries), a $999/month Scale plan with 10,000 chat-to-API queries, and custom Enterprise pricing. Setup involves connecting your API, testing in a playground, then a one-line embed.
In my research I went through Superflows' pricing page, its docs, and its Product Hunt history before writing any of this, because the name is genuinely ambiguous — there's also a separate website-QA tool, a Shopify workflow app, and an Android voice-to-text app all using some version of "Superflow." This guide covers the one that actually matches what people searching "Superflow AI" are usually trying to evaluate: the API-connected product assistant. Here's what it does, what it costs, and where I'd slow down before committing.
What you'll need
You need an existing product with an API, since Superflows works by calling your endpoints, not by generating one from scratch. An OpenAPI spec makes setup faster; without one, you'll be defining endpoints by hand in the playground. You'll also want your documentation ready to connect, since the assistant answers "how do I…" questions from your docs separately from the data queries it runs against your API. Decide ahead of time which actions the assistant is allowed to take on a user's behalf and which ones need a human to approve first — Superflows builds an approval step around this, but you have to define the boundary. Finally, someone needs frontend access to add the one-line embed once you're ready to go live; nothing in the free Testing tier requires that step, so you can evaluate the playground before touching production code.
Step-by-step: Setting up Superflow AI
1. Create an account and connect your API
Sign up at superflows.ai and start on the free Testing tier. The first setup step is connecting your API, either by uploading an OpenAPI spec or defining endpoints manually, so the assistant knows what data and actions it has access to.
2. Connect your docs (optional but recommended)
If you want the assistant to answer support-style questions, not just data queries, point it at your documentation. This is a separate source from the API connection, and it's what lets the assistant explain a feature instead of only reporting numbers.
3. Test in the playground
The free tier gives you 50 queries in a sandboxed playground before you touch your own product's UI. In my test runs, this is the step worth spending real time on, since it's where you catch an assistant confidently misreading your API schema before a real user does.
4. Set approval rules for actions
Beyond answering questions, Superflows can take actions on a user's behalf — creating a task, updating a record — but it asks for confirmation first by default. Decide which actions, if any, should skip that confirmation step, and which should never be automated.
5. Add the one-line embed
Once you're satisfied with playground answers, Superflows' own pitch is "integrate in 1 line of code" by adding a script snippet to your frontend. That single line loads the chat widget; the configuration work happens on Superflows' side, not in your codebase.
6. Watch usage and iterate
After launch, review the interaction logs to see what users are actually asking and where the assistant gives a wrong or unhelpful answer. Superflows' own docs frame this as an ongoing loop, not a one-time setup — the assistant only gets better if someone is reading what it gets wrong.
Example prompts you can copy
Superflows is built around natural-language queries against your own API, so the prompts a user types matter more than any settings screen. A few patterns worth adapting, based on the CRM-copilot examples Superflows shows on its own site:
- "Which [records] should I prioritize this week? Weight by [size] and [likelihood of closing]."
- "Summarize what changed in [account/project] since my last login."
- "Create a task reminding me to follow up on [item] in three days."
- "Which [channel/source] generated the most [leads/signups] this month, and how does that compare to last month?"
The pattern that works best: name the specific field or object from your own API schema, not a generic noun. "Which deals should I prioritize" only works if "deals" is a real object the assistant can query — vague nouns produce vague answers.
Common mistakes to avoid
The mistake I'd flag first: burning through the 50-query free tier on random exploratory questions instead of your hardest real scenarios — a multi-step query against a messy field, or an edge case in your data, tells you more than ten easy ones. Second, connecting an OpenAPI spec without reviewing which endpoints actually got exposed; a spec built for internal tooling can include far more than you want a customer-facing assistant to touch. Third, skipping the approval-rule step and assuming action confirmation is on by default for everything — check it explicitly for anything destructive or billing-related. Fourth, treating the $999/month Scale tier's 10,000 chat-to-API queries as a hard usage ceiling without reading the fine print on what counts as a query versus a docs lookup, since Superflows bills those separately (10,000 API queries and 200,000 docs queries on Scale, per its pricing page). Fifth, going with this tool purely because it showed up in an AI-tools directory — public discussion of Superflows outside its own site and Product Hunt listing is thin, so weigh that against better-documented alternatives before a long-term commitment.
Superflow AI (Superflows) vs. Chatbase at a glance
| Superflows | Chatbase | |
|---|---|---|
| What it does | Chat assistant that queries your API and takes actions, with docs Q&A | Chat agent trained on your content, docs, and support tickets |
| Free tier | Yes — 50 queries in a testing playground | Yes — 50 message credits/month, 1 agent |
| Entry paid plan | Scale, $999/month (10,000 chat-to-API queries, 200,000 chat-to-docs queries) | Hobby, $40/month (700 message credits) |
| Mid tier | — (Scale is the only paid self-serve tier) | Standard, $150/month (4,000 credits, voice, telephony) |
| Top self-serve tier | — (Enterprise is custom above Scale) | Pro, $500/month (15,000 credits) |
| Best for | Product teams adding an in-app copilot tied to their own API | Teams wanting a trained support/sales chat widget fast |
I confirmed both sets of numbers directly on the vendors' pricing pages — Superflows and Chatbase — on August 18, 2026. Neither company publishes a mid-range self-serve tier the way the other does, so the jump from free to paid looks very different: Chatbase scales gradually through four paid tiers, while Superflows jumps straight from a free playground to a single $999/month plan.
Tools that make this easier
If you're deciding whether to build this yourself or use a hosted assistant, my best AI tool for code and AI coding assistant guides cover what an in-house build actually takes, which is the comparison Superflows makes on its own site when it claims weeks instead of six-plus months. Small teams weighing a $999/month tool against everything else in the budget should also check best AI tool for small business for lower-commitment options. Since Superflows is one of several niche, thinly-documented AI tools worth researching carefully before you commit, my reviews of Leaping AI and AdAnt AI walk through the same kind of vendor-page-first evaluation for a voice-agent platform and a video-ad generator. And if what you actually want is an assistant that takes multi-step actions across the web rather than inside your own product, ChatGPT’s agent mode is the consumer-facing version of the same idea.
My take
The mechanism is straightforward and well-documented: connect an API and docs, test in a sandbox, embed one line, review what users actually ask. What's harder to verify from the outside is how the assistant performs against a messy, real-world API rather than the clean CRM demo Superflows shows on its homepage — that's exactly why the 50-query free tier matters more here than in tools with a generous ongoing free plan. The jump from free to $999/month, with no tier in between, is the single biggest thing to plan around; there's no way to "grow into" Superflows gradually the way you can with Chatbase's four paid tiers. I'd run the free playground against your gnarliest real API responses before assuming the Scale tier is worth it.
Frequently Asked Questions
Superflow AI: is it free?
Partly. Superflows has a free Testing tier with 50 queries in a sandboxed playground and a single user account, per its pricing page. Beyond that, the only self-serve paid option is Scale at $999/month — there's no lower-cost paid tier.
How long does it take to set up Superflow AI?
Connecting an API and running your first playground queries takes under an hour if you already have an OpenAPI spec. Superflows itself claims a proof-of-concept build in about two weeks with its team's help, and says building the same capability in-house typically takes six or more months.
What is the easiest way to get started with Superflow AI?
Start in the free Testing tier with your real API, not a demo dataset, and spend your 50 queries on the questions your users actually ask most, including the awkward ones. That tells you more before you commit $999/month than a clean happy-path test does.
Is Superflow AI the same company as Superflows?
Yes — "Superflow AI" is how people search for it, but the company's name and domain are Superflows (superflows.ai). Other unrelated products, including a website QA tool and a Shopify automation app, use similar names, so double-check the URL before you sign up.
Does Superflow AI replace my product's dashboards?
Not entirely, according to the company's own framing — its pitch is "build fewer dashboards" by letting users ask for data directly, not eliminating dashboards altogether. Complex visual reporting still tends to need a real dashboard; Superflows is aimed at the ad hoc questions that don't.