Guzli sells one of the three products below and wrote this page. It is not an independent analyst review. The job is a cost forecast, not a feature bake-off. You are choosing a meter: Chatbase message credits, Intercom Fin outcomes plus seats, or Guzli published plans (chat credits; voice ¢/min on Standard and Pro).
Method: desk research on public vendor pages, 19 September 2026 — not a logged-in tenant, not a purchase order. Re-open each vendor’s current published prices before you commit spend. This is planning math for support and finance leads, not tax, legal, or investment advice.
Read the same day: Chatbase pricing, Chatbase credit FAQ, Fin pricing, Fin outcomes, and Guzli plans. Published prices move.
Product shape lives on Guzli vs Intercom vs Chatbase and Chatbase vs Guzli. Phone: Does Intercom Fin answer the phone?.
The three meters
Chatbase: credits per agent reply
Public tiers on 19 September 2026: Free 50 credits, Hobby 700, Standard 4,000, Pro 15,000. Dollar amounts on that page are JavaScript-rendered in this environment, so this page does not freeze Hobby/Standard/Pro list prices. Confirm them on Chatbase’s current pricing page. Older third-party writeups quoted Hobby about $40 monthly / $32 annual-equivalent, Standard $150 / $120, and Pro $500 / $400 — treat those as dated, not as today’s invoice.
The FAQ is the usage rule that matters:
- Credits renew on the 1st of the month, not your anniversary.
- One agent reply is 1–6 credits depending on the model (default models 1; several current models 2–6).
- When the pack is empty, visitors see: “This AI Agent is currently unavailable.”
- Auto-recharge exists so unused extra credits do not expire. Confirm the current pack price on the pricing page (commonly cited as $40 per 1,000).
Actions (Shopify, Stripe, Calendly, lead capture) still consume reply credits. They can shorten a thread. They do not change the meter.
Intercom Fin: $0.99 per outcome, plus seats
fin.ai/pricing (19 September 2026): $0.99 per outcome, 50 outcomes/month minimum. On Intercom, that stacks on helpdesk seats that start at $29 per full seat / month on that page — confirm Essential / Advanced / Expert on Intercom’s current seat cards.
| Outcome | Public price | Counts when |
|---|---|---|
| Resolution (confirmed or assumed) | $0.99 | No further help after Fin’s last answer |
| Procedure handoff | $0.99 | A Procedure you configured ends in a handoff |
| Disqualification | $0.99 | Sales Fin marks the prospect out |
| Qualification | $9.99 | Sales Fin matches and routes (meeting / trial) |
| Self-serve qualification | $0.99 | Routing marked Self-serve |
| Customer asks for a human, or a Procedure fails | $0 | Not an outcome |
One conversation, one charge, even if Fin ran several Procedures. A customer who leaves after an answer is an assumed resolution and is billed. A clarifying question with no reply is abandoned, not billed.
Fin Voice is custom / sales-gated. Do not put ¢/min into this forecast. See Does Intercom Fin answer the phone?.
Salesforce completed its acquisition of Fin (formerly Intercom) on 10 September 2026. Published Fin outcome prices on fin.ai/pricing and Intercom’s outcome article were still the live meter we used for this forecast. Do not replace that meter with Salesforce seat math until your order form says so.
Guzli: published plan + billed overage
See current plans and pricing. Hobby is $49/mo chat-only (1,000 credits, 1 member). Voice starts on Standard: $149 + 16¢/min talk (5¢/min transfer, 3 concurrent). Pro is $499 + 14¢/min (4¢ transfer, 8 concurrent). Standard includes 12,000 chat credits; Pro 40,000. Team seats 1 / 3 / 10.
When chat credits run out, the widget stays up and overage is billed at the listed add-on rate on that page. We will not freeze an overage ¢ here if checkout shows a different pack.
Guzli is not an Intercom or Shopify App Store app. Handoff can open a ticket in Salesforce, Zendesk, or Guzli’s help desk.
Four inputs, then the formula
- Monthly conversations (or tickets the bot will see)
- Share that becomes a billable Fin outcome (not the same as “AI touched it”)
- Average bot replies per conversation, and which Chatbase model (credit multiplier)
- Whether you also pay Intercom seats, Chatbase auto-recharge, or Guzli voice minutes
Copy:
Chatbase ≈ plan_price + extra_thousands × recharge_price
(credits_needed = conversations × replies × model_credits)
Fin ≈ max(50, outcomes) × $0.99 + seats × seat_price
(plus $9.99 per full qualification if you turn Sales Fin on)
Guzli ≈ plan_price + billed chat overage + (minutes × ¢/min if Standard/Pro)Actions change (2) and (3). They do not invent a fourth vendor.
Planning examples (not quotes)
Shared assumptions: 2,000 / 6,000 / 20,000 conversations. 40% become a Fin resolution or Procedure handoff (800 / 2,400 / 8,000 outcomes). 4 bot replies when a connected action finishes the job; 7 when the bot only talks. Chatbase on a 1-credit model first; a 3-credit model triples the Chatbase column. No Fin qualifications. No Guzli voice minutes in the chat-only rows. Five Intercom Advanced-class seats are not in the table — add them on your side from the live seat card.
| Volume | Fin outcomes @ $0.99 | Chatbase credits @ 4 replies | Chatbase credits @ 7 replies | Guzli chat plan that covers 4-reply volume |
|---|---|---|---|---|
| 2,000 | $792 (above the 50-outcome × $0.99 floor) | 8,000 → above Standard 4,000 | 14,000 → fits Pro 15,000 | Standard 12,000 credits ($149) |
| 6,000 | $2,376 | 24,000 → above Pro 15,000 without recharge | 42,000 | Pro 40,000 ($499) covers 4-reply; 7-reply needs overage |
| 20,000 | $19,800 | 80,000 | 140,000 | Enterprise / custom; do not use Hobby |
Read the table as shape, not a bid:
- Fin’s bill rises with successful outcomes. At 20,000 conversations and a 40% outcome rate you are modeling almost $20k/month in Fin before seats. That can still be cheap versus human handle time. It is not “$0.99 somewhere.”
- Chatbase’s bill rises with replies × model weight. Standard’s 4,000 credits are 1,000 conversations at 4 one-credit replies — or ~333 conversations on a 3-credit model. Empty credits take the agent offline unless auto-recharge is on.
- Guzli’s published Standard/Pro credit packs are larger than Chatbase’s Standard/Pro packs (12,000 vs 4,000; 40,000 vs 15,000). That is pack size, not a cheaper-dollar claim — Chatbase list prices were not frozen here. Voice is a second Guzli meter (¢/min), not a Fin-style outcome.
If billing and WISMO (“where is my order”) are most of the queue, a working Shopify or Stripe action (Chatbase documents both; Guzli ships the same named tools) usually cuts replies and Fin outcomes. Model that as a second pass: drop replies from 7 → 4, and outcomes from 40% → 25%, and re-run. Do not treat “we added actions” as a 0% ticket rate.
What to ask so the invoice matches the deck
- What is one unit — reply credit, outcome, seat, or minute?
- If the customer asks for a human after an answer, is that billed? (Fin: often yes if they already got an answer and did not ask; Chatbase: credits already spent; Guzli: credits already spent, widget stays up.)
- Do actions add units or only shorten threads?
- Can I set a hard cap? Fin documents reminders and limits. Chatbase documents auto-recharge. Guzli bills overage.
- Can I see usage by intent (billing, order status, returns)?
- Is voice on a published ¢/min, an outcome, or an order form?
Buyer checklist: AI chatbot buyer’s guide. ROI inputs: AI chatbot ROI and the ROI calculator.
When each meter is the honest fit
Fin + Intercom if the suite (inbox, series, tours, Procedures) is already the operating system and you will pay seats anyway. Stay if assumed resolutions match how you define “done.” Leave the phone out of the model until Fin Voice is on your order form.
Chatbase if you want a custom-GPT builder, you can keep threads short, and you accept a visitor-facing stop when credits die — or you will pay auto-recharge. Voice and SIP (Session Initiation Protocol — the telephony trunk that puts the agent on a phone number) sit on Standard/Pro; Chatbase does not publish a ¢/min on the page we opened.
Guzli if you want website live chat (and inbound voice on Standard+) on a published plan, billed overage instead of “unavailable,” and a ticket back in Salesforce, Zendesk, or Guzli’s desk. Hobby cannot take the store number.
FAQ
What is one billable unit on each product?
Chatbase bills message credits per agent reply (1–6 by model). Fin bills outcomes ($0.99, once per conversation) on top of Intercom seats. Guzli bills a published plan plus chat overage and, on Standard/Pro, voice minutes.
If the customer asks for a human after an answer, is that billed?
On Fin: often yes if they already got an answer and did not ask — that is an assumed resolution. If they ask for a human, or a Procedure fails, Fin documents $0. Chatbase: credits already spent. Guzli: credits already spent; the widget stays up.
What does a visitor see when Chatbase credits run out?
The Chatbase FAQ says the agent shows “This AI Agent is currently unavailable.” Credits renew on the 1st. Auto-recharge exists so unused extra credits do not expire. Guzli’s public pricing is the other contract: the chatbot stays up and overage is billed.
Did Salesforce buying Fin change this forecast?
Salesforce closed the Fin acquisition on 10 September 2026. This page still models Fin’s published $0.99 outcome meter plus Intercom seats. Do not swap in a different Salesforce price until that price is on your contract.