Ask what Sierra costs and the honest answer is that Sierra will not tell you, not on a page anyone can read without a sales call. There is no pricing page, no tier list, no self-serve signup. What is public is the philosophy: outcome-based pricing, where Sierra is paid when its agent resolves something and not when it fails to. That is a genuinely interesting idea. It is also a model built for a company that already has a support volume worth metering and a budget line that survives a six-figure enterprise contract, which describes a small fraction of the businesses that need help with customer conversations.
This is a fact-checked comparison. Sierra's own site was read the same week this was written, because a platform with no public rate card and a philosophy this widely discussed deserves the numbers stated carefully rather than repeated as folklore. Where a figure below is a third-party estimate rather than something Sierra confirms, it is labeled that way.
What Sierra actually is
Sierra is a conversational AI platform built for enterprise customer service and support, with three named pieces: Ghostwriter, which turns a company's SOPs, conversation records or plain-English descriptions into a production agent; Insights, which analyzes conversations, tracks performance and runs A/B tests on agent behavior; and Horizon, which handles longer-horizon planning for outcomes that take more than one turn, proactive engagement and multi-step resolutions. The company's own language is "beyond customer service to drive revenue growth," and its named customers, Rocket Mortgage, Uber, Vanguard, SiriusXM, are the kind of company that already runs a support organization at a scale worth optimizing with a dedicated platform.
The commercial model is outcome-based pricing: Sierra is paid per resolution, not per seat or per message, so an unresolved conversation or one that escalates to a human is reported to cost nothing. Sierra has never published the rate. Third-party trackers and industry reporting place per-resolution pricing somewhere between $1 and $2.50 depending on complexity and volume, with a commonly cited midpoint near $1.50, and describe annual platform contracts starting around $150,000, with first-year cost, contract plus implementation, running $200,000 to $350,000 or more for a base enterprise deployment. None of those numbers come from a Sierra pricing page, because Sierra does not have one; they come from analysts and competitors estimating a private enterprise contract, and this piece treats them as estimates rather than fact for that reason.
What Sanaf is instead
Sanaf, the AI employee, is not a platform for building a support agent. It is a hire that can hold a support job alongside thirty-four others: chasing invoices, keeping a CRM current, drafting the weekly update, booking appointments. You write the support job description the way you would for a person joining the support team, connect the help desk and the inbox it needs to work in, and Sanaf answers the routine questions, resolves what it has been trained to resolve, and escalates anything that needs a person, in the same Slack or Microsoft Teams channel where the rest of the team already works.
The difference is not "outcome pricing versus something else." It is what has to be true before either one starts paying for itself. Sierra's outcome-based model only makes sense against a support volume large enough that per-resolution economics beat a fixed cost, and a budget process built to sign a six-figure annual contract before the first agent goes live. Sanaf starts with $120 of free credit and no card, and the same credit that answers a support conversation also chases an invoice or drafts a report, so a business with modest support volume is not paying platform overhead sized for Uber's ticket queue.
Side by side
| Dimension | Sanaf | Sierra |
|---|---|---|
| What you get | One AI employee with a written job description for every job you hand it, including support | A platform for building enterprise conversational agents, built specifically for support and service |
| Where it works | Slack, Microsoft Teams, and its own chat in the console | Your own channels, configured as part of an implementation |
| What it needs first | A Slack or Microsoft Teams workspace and the apps you already use | A support operation with enough volume to justify a metered, per-resolution contract |
| Who sets it up | You, in plain words, in an afternoon | Sierra's implementation team, alongside your own, over a sales-and-deployment cycle |
| Jobs it can hold | Support, plus 34 others: inbox, invoicing, scheduling, reporting and more | Support and service conversations, its stated focus |
| Pricing model | Free to start with $120 of credit and no card, then credits from $20 as you need them, and they never expire | Outcome-based, per resolution; no published rate; reported to start near $150,000 a year in contracts |
| Getting started | Self-serve, no sales call | Enterprise sales, no self-serve path visible on Sierra's own site |
| Best for | A business that wants one hire covering support and everything around it | A large support organization with the volume and budget to run a metered enterprise contract |
The real difference: one job, or a roster
Sierra's strength is real: a platform purpose-built for support, with tooling to turn existing SOPs and transcripts into an agent, analytics to see what is and is not resolving, and planning for outcomes that take several turns. For an enterprise support organization already handling large ticket volumes, that specialization and the outcome-aligned pricing behind it can be a genuinely good trade, since paying only for what resolves removes a real risk from a metered rollout.
The limit is that Sierra answers exactly one question: how do we handle support conversations. Most businesses do not have a support-only problem. They have an inbox nobody owns, invoices going unchased, a calendar that eats a morning, and a support queue that is real but is one job among several, not the whole operation. Buying a support-specialized enterprise platform for that business means solving one of five problems at enterprise cost and leaving the other four exactly where they were.
Sanaf is built for that wider list. The support job runs with the same care a specialized platform would bring, escalating anything genuinely unhappy or outside policy straight to a person, but it is one of thirty-five jobs on the same hire, not the entire product. A business that needs support handled today and invoicing handled next month does not need a second platform and a second sales cycle; it needs the same hire to pick up the next job description.
Approvals and who is actually accountable
Sierra's Insights component gives an enterprise buyer the analytics layer a metered contract demands: conversation-level tracking, A/B testing, performance monitoring, the instrumentation that lets a company justify paying per outcome. That is appropriate for a platform priced on results, since a company paying per resolution needs to see exactly what it is being charged for.
Sanaf's accountability model runs at the level of the job rather than the platform: every kind of action the support job can take, replying to a customer, escalating an issue, updating a record, carries its own rule of do it, ask first, or never, with conditions attached. Every job starts in shadow mode, drafting for a person to review before acting unattended, and earns more autonomy only once a reviewer is comfortable handing more of it over. That is a lighter mechanism than an enterprise analytics suite, built for a team that wants to see what happened in the thread where it happened, not a separate dashboard to check.
The honest pricing comparison
Sierra's outcome-based model is real and its logic is sound for the buyer it targets: pay for resolutions, not for seats or messages, and let volume determine the bill. The catch is that the number itself is not public. Third-party estimates put per-resolution pricing at roughly $1 to $2.50, with reported annual contracts starting near $150,000 and first-year totals, including implementation, commonly cited at $200,000 to $350,000 or more. A business has to enter a sales process to learn what any of that means for its own volume, and the contract floor alone rules out most companies before the per-resolution math even starts.
Sanaf's price is public and the same for every business: $120 of free credit with no card to start, then packs from $20 that never expire, priced for the work actually done rather than a seat or a resolution count. A short reply in Slack barely touches it; a worked task that reads a few apps and drafts something back uses more. There is no six-figure floor, no separate contract for support work, and the same credit balance that resolves a support conversation also chases an invoice, because it is one hire with one credit meter, not one platform per job.
Where Sierra wins, honestly
Choose Sierra if you run an enterprise support organization with the ticket volume, the budget process and the internal team to run a metered, outcome-based contract, and if support is a large enough share of your cost structure that a platform purpose-built for it, with its own analytics and long-horizon planning, is worth a six-figure annual commitment on its own. Choose it if resolving conversations at scale is the single highest-leverage problem in the business this year, ahead of everything else on the list.
Where Sanaf wins, honestly
Choose Sanaf if support is one real job among several, not the whole problem, and you would rather one hire cover the inbox, the invoices, the schedule and the support queue than buy a separate platform for each. Choose it if there is no support-specific budget to sign a metered enterprise contract against, and you want to see what a support conversation actually costs before committing anything beyond the free credit. Choose it if the team asking for help this week is the same team that will ask for invoice follow-up next month, and one job description away is exactly how far that request should have to travel.
Can you use both?
For a company already running Sierra at enterprise scale, yes, and it is a sensible split rather than a hedge. Sierra is well placed to keep handling the high-volume support queue it was built and tuned for. Sanaf is built for the parts of the business that sit outside that queue entirely: the invoice chased through a bookkeeping tool, the weekly update pulled together from three apps, the appointment nobody had time to confirm. A company with Sierra already live can reasonably keep it on support and add Sanaf around it. A company that has never signed a six-figure support contract, which is most small and mid-sized businesses, is choosing between the two, and the honest answer for that business points toward the one that starts free.
The bottom line
Sierra is a serious platform for a narrow, high-stakes job: enterprise customer support at a volume that makes outcome-based pricing pay for itself, sold through a sales process because that is how a six-figure contract gets signed. Sanaf answers a more common situation: a business that needs support handled well, alongside the four or five other jobs sitting in the same inbox and the same spreadsheet, without a separate platform or a separate sales cycle for each one. Outcome-based pricing is not the question that decides this comparison. Whether support is the whole problem or one job on a longer list is.
Explore every job Sanaf can hold, see what starting free actually costs, or get in touch to talk through where the line falls in your own business.