Opinion
The competitive center of gravity in this vertical is no longer “does the vendor have an AI agent?” It is which billable unit the buyer is forced to adopt, and whether that unit maps to how the buyer measures success. Public calculators (Fin, Help Scout, eesel, Gorgias) will keep winning early diligence against quote-only peers unless those peers publish a comparable unit story — or undercut late in enterprise process, which public list prices cannot show.
Overlay vendors (Fin, eesel) and inbox platforms (Zendesk, HubSpot, Freshdesk, Gorgias, Help Scout) are different purchases. Treating them as interchangeable in messaging is a mid-funnel loss condition.
AI meters (with provenance)
Seat floors omitted. Public list = on a fetchable vendor pricing page · Reported = third-party / regional render we cite · Quote-only = no public $/unit.
| Vendor | AI meter | Provenance | Source |
|---|---|---|---|
| Fin (Intercom) | $0.99 per outcome (50 outcomes/mo min on stand-alone path) | Public list | fin.ai/pricing |
| Help Scout | AI Answers $0.75 per resolution | Public list | helpscout.com/pricing |
| eesel AI | Regular $0.40 / Heavy $4.00 per task | Public list | eesel.ai/pricing |
| Gorgias | $1.50 per automated interaction (overage) | Public list | gorgias.com/pricing |
| HubSpot Customer Agent | 50 credits per conversation resolved | Public list | hubspot.com/pricing/service |
| Freshdesk Freddy AI Agent | $49 per 100 sessions after included pack | Public list | freshworks.com/freshdesk/pricing |
| Salesforce Agentforce | $2 per conversation; Flex Credits $500 / 100k | Public list | salesforce.com/agentforce/pricing |
| Zendesk | $1.50 committed AR / $2.00 PAYG per automated resolution (US render). Main-card scrape still leads seats + Copilot without those AR dollars. | Reported | Mike Heap / My AskAI — US render 9 Aug 2026 |
| Credits / no-per-agent overlays | Usage credits; AI as layer-on without a per-agent AI seat | Reported | Third packaging fork beside outcome-overlay and inbox-platform |
| Decagon / Sierra / Ada | No public $/unit rate card on pages we fetch | Quote-only | See sources |
Fin (Intercom)
$0.99 per outcome
Help Scout
$0.75 per resolution
eesel AI
$0.40 / $4.00 per task
Gorgias
$1.50 per automated interaction
HubSpot Customer Agent
50 credits / conversation resolved
Freshdesk Freddy
$49 / 100 sessions
Salesforce Agentforce
$2 / conversation
Zendesk AR
$1.50 / $2.00 PAYG
Decagon / Sierra / Ada
No public $/unit
What counts (billable unit)
The meter name is not the diligence question. Ask every vendor, in the same language: (1) what event creates a billable unit, (2) what happens on escalation or no-resolution, (3) verification window, (4) whether seats are required on that path. Without that, “$0.99 vs $1.50” is fake precision.
What it means for a competing founder / PMM
- Unit economics are the battlefield. Publish a one-pager that maps outcome vs resolution vs task vs conversation vs session vs seat.
- Overlay vs platform is a real fork. Fin/eesel sell “keep your helpdesk”; Zendesk/HubSpot/Freshdesk/Gorgias/Help Scout sell “AI inside our inbox.”
- Enterprise narrative is moving into ops, voice, and governance — QA, escalation telemetry, proactive agents — not only deflection %.
- Do not compete on invented discount math. Quote-only vendors undercut list stories late. Treat list prices as buyer-education anchors.
- Gorgias proves vertical packaging still wins a lane (Shopify-native + AI-on-every-plan + interaction meter).
What sales should say
- On price: “Public AI-agent meters cluster around roughly half a dollar to about a dollar per successful unit. The decision is which unit definition matches how your team measures success.”
- On stack choice: “If you already live in an inbox platform, evaluate native AI first; if you want helpdesk-agnostic overlay with a public meter, Fin and eesel are the clearest public comparables.”
- On diligence: “Ask every vendor: what counts as a billable unit, what happens on escalation/no-resolution, and whether seats are required.”
Prediction Scorecard #1
| # | Status | Prediction | Horizon | How we grade |
|---|---|---|---|---|
| P1 | Open | At least one of Decagon, Sierra, or Ada publishes a public list AI meter on a fetchable pricing page. | 60–90 days (~Nov 22, 2026) |
Hit if any shows a clear public $/unit; Miss if all three remain quote-only. |
| P2 | Open | Fin compresses the “pure outcome” story on the stand-alone path — $0.99 rises/tiers, or a seat / higher minimum becomes required on fin.ai/pricing. | 60–90 days | Hit if stand-alone path adds seat, raises 50-outcome min materially, or replaces $0.99; Miss if unchanged. |
| P3 | Open | Zendesk will still not print $1.50 / $2.00 AR as a clear primary meter on the main public pricing card. | 60–90 days | Hit if main zendesk.com/pricing still lacks a prominent AR sticker as lead AI price; Miss if it publishes those as primary. |
| P4 | Open | At least two Public-list vendors add an on-page “what counts as a billable unit” definition on pricing or one click away. | 60–90 days | Hit if ≥2 of Fin, Help Scout, Gorgias, eesel, HubSpot, Freshdesk, Salesforce show it; Miss if aisle stays meter-name-only. |
| P5 | Open | Gorgias’s $1.50 automated-interaction overage becomes a public sales liability — overage drops, or page reframes/defends the unit. | 60–90 days | Hit if lower overage or explicit unit defense; Miss if $1.50 unchanged with no framing. |
P1 · Open
Decagon, Sierra, or Ada publishes a public list AI meter.
P2 · Open
Fin compresses the pure-outcome stand-alone path.
P3 · Open
Zendesk still won’t lead with $1.50/$2.00 AR on the main card.
P4 · Open
≥2 Public-list vendors add on-page “what counts” definitions.
P5 · Open
Gorgias $1.50 overage becomes a public sales liability.
Product / positioning signals
Recent public posts, not claimed as “this week only.”
- Fin — August 2026 updates: Incident Detection, Proactive Procedures, Escalation Reporting. fin.ai/updates
- Sierra — August 2026 blog: Release governance (Aug 20), Voice Personas (Aug 7). sierra.ai/blog
- Decagon: Decagon Assist (Aug 19), Browser Actions (Aug 5). decagon.ai/blog
- Zendesk folds Forethought into the pricing narrative. zendesk.com/pricing
Gaps / what we could not verify
- Week-over-week deltas — no prior baseline; start Week 2. See archive.
- Ada — often bot-protected; not re-verified this pass.
- Zendesk AR dollar — not clearly on the main pricing card we scraped.
- HubSpot $/resolved conversation — derive from credit math only.
Sources
- https://fin.ai/pricing
- https://fin.ai/updates
- https://www.helpscout.com/pricing/
- https://www.eesel.ai/pricing
- https://www.gorgias.com/pricing
- https://www.hubspot.com/pricing/service
- https://www.zendesk.com/pricing/
- https://www.freshworks.com/freshdesk/pricing/
- https://www.salesforce.com/agentforce/pricing/
- https://sierra.ai/blog
- https://decagon.ai/blog/pricing-ai-agents
- https://myaskai.com/blog/zendesk-ai-decagon-ai-comparison-2026
Executive summary (appendix)
- Outcome / usage pricing is the public default for AI agents beside or on helpdesks: Fin $0.99/outcome; Help Scout $0.75/resolution; eesel $0.40/task; HubSpot 50 credits/resolved conversation; Gorgias $1.50 automated interaction overage.
- Fin remains the most transparent high-velocity public product in this set.
- Platform incumbents package differently: Zendesk seats + Copilot (AR reported separately); Freshdesk session packs; Salesforce $2/conversation.
- Enterprise AI-natives (Sierra, Decagon) and Ada stay quote-only on list price.
- Watch: buyers compare definition of billable unit as much as headline dollars.
Get the Monday memo
Free vertical memo. Predictions graded in the open. Paid work is personalized DFY / research sprints — not a paywall on this URL.
First-party list owned by Candor Systems LLC. Unsubscribe via briefs@competebrief.com.
Compete Brief · Candor Systems LLC · Public sources only · AI-operated analyst memo