Comparison
Refik and Siena, read on the same day.
Both sides sell an assistant that talks to shoppers, so the difference worth printing is not what it is called — it is the shape of the bill. One side charges a fixed platform fee every month and then adds a price for each ticket it automates on top of that. The other charges one price a month with a block of conversations already inside it. The rows below are the evidence for that sentence, each one dated, each one lifted from a page you can open yourself.
Method
Two sides, two dates, and an empty cell that means something.
The right-hand side of every row was read out of Siena’s own pages on 10 October 2026 — the pricing page, the home page and the integrations page, nothing else, no analyst report and no review site. The quotes are theirs, word for word, including the dash that sits in the middle of two of their sentences and one line that arrived cut off. The left-hand side was read out of our own source code on 4 October 2026, and the plan figures come from the pricing page in this same directory, which fails its build if a price in the code changes without the page changing with it.
An empty cell is a third state and it is on purpose. It does not mean “they do not have it”. It means we did not read it on that day, so printing a dash there would turn a gap in our reading into a claim about their product. Where a cell is empty, the reason is written underneath the table, row by row, and one of those empty cells is ours.
Three things are missing from this page by design. There is no verdict row and no score, because a price for an automated ticket and a price for a storefront conversation are not the same unit, and dividing one by the other would produce a number that looks like a measurement and is not one. There is nothing here about how well either assistant answers a shopper, because that was not measured on either side. And one line from their home page, the one that lists the platforms they connect to, was read and is deliberately left out: it names other vendors this page does not measure, and a single phrase is not enough to carry a sentence about a third product.
Our own side has a limit too, and it is the same one on every row. The 4 October reading walked the source code and checked four links in a chain: a surface is written, something calls it, the path is actually reached, and something reads the result. That is a statement about what the code produces. It is not a statement that the surface was watched working in a live shop on a real catalogue, and those are two different claims.
The rows
What each side charges for, and where the floor sits.
Read the first two rows together. They are the reason the rest of the table looks the way it does: one bill opens at a fixed figure every month and then grows with the work it does, and the other opens much lower with a block of work already paid for. A shop automating thousands of tickets a month and a shop whose owner answers everything between orders land on opposite sides of that line.
Siena read from www.siena.cx/pricing, www.siena.cx and www.siena.cx/integrations; Refik read from this repository and the pricing page.
| What the row measures | Refik | Siena |
|---|---|---|
| What the bill is made of | ||
| What you pay before any volume at all | Starter, $39 a month or $390 a year, for the whole shop, with 300 conversations already inside it. | “Platform fee Access Siena’s core AI engine and unlimited sandbox for $750/month—reliable, high-quality automation.” |
| What the volume itself costs | Nothing, up to the band. Above the band and on monthly billing only, $0.15, $0.12 or $0.10 for a conversation, and the charge stops at one band’s worth. On yearly billing there is no charge above the band at all. | “Automation Pack Automate tickets at just $0.90 each—scale effortlessly while keeping costs in check.” |
| Upper tiers, list price in USD | $99 and $249 a month; $990 and $2,490 a year, carrying bands of 1,000 and 3,000 conversations. | |
| A separate line on the bill for getting started | None. Setup is five steps the merchant walks alone, there is no onboarding service to buy, and the setup page makes no claim about how long it takes. | “Support & Implementation Expert onboarding plus dedicated Slack su…” |
| A sandbox to try changes in | Named inside the platform fee line above, as an “unlimited sandbox”. | |
| Free trial | 14 days on every plan, with a conversation allowance scaled to the plan. | |
| Where it runs, and what it reads | ||
| Platforms it connects to | Shopify, and nothing else. The catalogue, the policies and the order lookups all come from one store’s Admin API, and there is no second platform anywhere in the code. | “E-commerce Shopify Handle Shopify orders from inquiry to resolution automatically.” |
| Channels a shopper can reach it on | The web storefront, and only that. WhatsApp, Instagram, Messenger, email and SMS are not built, and that is written on our integrations page rather than left to be discovered after the first invoice. | |
| What feeds the answers | The store’s own catalogue, the shop policies pulled from Shopify and offered to the merchant as a draft rather than applied, and question-and-answer pairs typed by hand, capped at 300 records. There is no URL crawler, no document upload, and no ingestion of shop pages or blog posts. | |
| How each side describes itself | A shopping assistant on the storefront: it recommends, it looks up an order, and it hands a hard question to a person instead of settling it. | “Experience AI agents that combine empathy with automation for commerce. Provide personalized shopper support.” |
- Upper tiers, their cell is empty. Their pricing page describes three line items and does not draw a tier table, so package names and the volume included in each were not measured. That is the gap the row is reporting, and it is the reason there is no like-for-like band figure on their side.
- Their onboarding line arrived cut off. The sentence was captured truncated at “Slack su”, so the rest of it was not read, and neither was what that line costs. We print it in the state it was captured rather than completing it from memory.
- Free trial, their cell is empty. None of the three pages read that day stated a trial length. That is a gap in our reading, not a statement about their product.
- Channels, their cell is empty. A channel list was not among the pages read. Ours is written out because a missing channel is the kind of thing a shop finds out after it has paid.
- What feeds the answers, their cell is empty. Nothing on their side of the fact base this page is written from speaks to where the knowledge comes from, so there is nothing to quote and nothing to infer.
- The sandbox, our cell is empty. The 4 October reading did not cover whether the product has a place to try changes away from a live storefront, so there is nothing honest to put in the cell. It is the one empty cell on our side and it is empty for the same reason as theirs.
- No search or traffic figures for either side. They were not measured for their side, and we are not going to publish an estimate dressed up as a reading.
- Neither side’s answer quality is here. Not theirs, not ours. Measuring it would need the same shoppers asking the same questions of both, on a real catalogue, and that has not been done.
The difference that matters
A bill with a floor and a bill with a ceiling.
A platform fee with usage charged on top is a bill with a floor. It is the same figure before the first ticket of the month as after it, and then it grows with the work the engine does. There is a real argument for that shape: if the automation is the product, the fee is paying for an engine you keep whether or not this month was busy, and the per-ticket line makes a busy month cost more than a quiet one in a way you can predict. A band price is the other shape. The volume is already bought up to a line, crossing the line costs something only on monthly billing, and nothing at all on yearly. The first shape asks what the engine is worth to you. The second asks how many shoppers you expect to talk to.
The unit is not the same either, and that is the whole reason there is no verdict row on this page. One side counts a ticket it automates. The other counts a conversation that starts on a storefront. A shopper who asks three questions on a product page before buying is one conversation here; whether that same shopper is one ticket, three tickets, or none at all on the other side was not read, so there is no conversion between the two units and any arithmetic across them would be invented. The figures in the first two rows are comparable only in the narrow sense that both are printed in dollars.
What the two assistants are pointed at is also different, and their own sentences say so more clearly than ours could. Theirs is written about handling an order from the first question to the end of it, on its own. Ours is written about the questions that arrive before the order exists, and it is built to stop rather than finish: it opens a handover, it tells the shopper a person will pick it up, and it is instructed never to tell anyone a refund has been approved. That is a narrower promise on purpose, and it comes with a narrower bill.
The handover has an honest state and this is it. The server can open a ticket in one help desk, that help desk is not named on this page because this page measures one product and that is not it, and the settings screen for choosing a target is not built. So a shop cannot wire the handover up from our panel today. Our integrations page lists what is connected and what is not, in the same words.
Who each one fits
Where we would tell you to buy the other one.
Their shape fits you if automating support volume is the thing you are actually buying — if there is already a queue large enough that a fixed monthly engine fee looks small next to what that queue costs you in people, and a price per automated ticket is a number you can put in a spreadsheet against a salary. It also fits you if you sell in more places than one storefront. Our platform list is one platform long and our channel list is one channel long, and no amount of wanting it to be otherwise changes the code.
We fit you if the questions that cost you money arrive before the order exists: which of these two is right for me, is this in stock in my size, where is the order I placed on Friday, can I send this back. Those are storefront questions, they arrive on a product page, and answering them needs the catalogue and the policies of one shop rather than a ticket queue. If your monthly conversation count is in the hundreds rather than the thousands, a platform fee of $750 before the first question of the month is answered is the part of the other bill you would feel first.
We are the wrong tool for you if your answers live somewhere we cannot read. We read a Shopify catalogue, the shop policies, and question-and-answer pairs somebody types in, with a cap on how many. We do not crawl a page, open a document, or learn from a help centre you already wrote. If that is where your knowledge is, the work of getting it into us is yours and it has a ceiling, and you should know that before you pay rather than after.
Neither of us fits you if what you actually need is a person. Both products put a human at the end of some path; ours gets there more often, by design, because it is told to hand over rather than resolve. How quickly a person actually arrives is not ours to promise and was not measured. If there is nobody on the other end to pick that handover up, no assistant on either side of this table fixes that, and neither of these bills is the thing standing between you and a better week.