Alternatives to Decagon

Almost nobody leaves Decagon because the agent stopped resolving tickets. They leave because of the bill’s shape — an annual platform fee reported around $50,000 that you pay before a single conversation is handled — or because a sales-led deployment ran past the quarter it was budgeted into. Both are real. Neither is fixed by every name on the usual shortlist, and two of those names stopped being independent companies in 2026.

This page is the exit map, the arithmetic, and the case for staying. Read the ownership section first, because it removes two options before you spend a sales cycle on them.

Check the ownership column first

Two of the five obvious Decagon replacements were absorbed this year.

Forethought was acquired by Zendesk, announced 11 March 2026 and completed later that month — Zendesk’s largest acquisition in nearly two decades. It ships as “Forethought AI Agents by Zendesk.” It still runs on non-Zendesk help desks, but the standalone roadmap is being folded into Zendesk’s Resolution Platform.

Fin — the company that was Intercom until it renamed itself in May 2026 — signed a definitive agreement on 15 June 2026 to be acquired by Salesforce for roughly $3.6B, expected to close in Q4 of Salesforce’s FY2027. One vendor, two names: the agent is Fin, the help desk is still sold as Intercom, and ooligo carries it at /en/tools/intercom/.

That leaves Sierra, Ada, and Lorikeet as the independent options — and Decagon itself, which closed a $250M Series D on 27 January 2026 led by Coatue and Index at a $4.5B valuation, tripling its valuation in under six months, then ran an employee tender at the same price in March. If vendor independence is part of why you are leaving, your list is three names, and the incumbent is not the one at risk.

What you are paying now

Decagon publishes nothing. The reported structure is a two-part bill: an annual platform fee around $50,000, plus a usage charge billed per conversation or per resolution, with at least one publicly discussed contract using $0.50 per resolution on top of the fee. Third-party guides put entry deployments near $95,000 a year all-in, which is consistent with that fee plus real volume.

The two parts fail differently. A fixed floor punishes low volume; a meter punishes high volume. Each alternative below removes one of them, and one removes neither.

What the alternatives actually cost

Lorikeet’s credits convert cleanly: $18,000 buys 18,000 credits on Start and $48,000 buys 48,000 on Scale, so a credit is $1.00 at both tiers. A chat, email, or SMS resolution costs 0.95 credits on Start and 0.80 on Scale — $0.95 and $0.80. Voice runs 1.50 and 1.20 credits for resolutions up to three minutes. Routing or analytics tagging and automated QA are 0.30 and 0.25 per ticket.

Fin and Lorikeet are the only two you can price without a sales call, and both land near a dollar per resolved conversation.

The crossover that decides most of these deals. Decagon’s discussed $0.50 per resolution is the cheapest per-unit rate on this page. The fee is the floor, not the meter. Against Fin’s flat $0.99 with no platform fee, Decagon’s fee is repaid at $50,000 ÷ ($0.99 − $0.50) ≈ 102,000 resolutions a year, or about 8,500 a month. Below that, Fin is cheaper and the platform fee is dead weight. Above it, Decagon’s rate wins and leaving raises your unit cost. Run this with the rate on your own contract before anything else on this page — it settles the question more often than the feature comparison does.

Fin

The price you can verify, and the migration you do not have to run. Since 9 June 2026 Fin ships as a Service Agent on top of HubSpot and Freshworks with APIs, MCP, and a CLI, explicitly so buyers can deploy without moving off their help desk. Flat $0.99 per resolution, $9.99 per qualification outcome, a 50-outcome monthly minimum, and Copilot at $35/user/month.

Move here when: you resolve under roughly 8,500 conversations a month and want the platform fee gone, or when finance wants a number it can model without a quote.

Don’t move when: procurement weights independence. You are buying into Salesforce, and the routing detail is in Decagon vs Intercom.

Sierra

The enterprise ceiling, and the one that costs more than what you are leaving. Sierra raised a $950M Series E in May 2026 led by GV and Tiger Global at a $15.8B post-money valuation and claims more than 40% of the Fortune 50 as customers. Billing is outcome-based — you pay on a resolved interaction, a saved cancellation, a completed upsell — and what counts as an outcome is defined in the contract, which is the term to negotiate hardest.

Move here when: the agent has to hold up in mortgage refinancing, insurance claims, or comparable regulated workloads, and the budget is a Fortune-500 line item.

Don’t move when: cost is your reason. Annual commitments near $150,000 plus $50,000–$200,000 implementation and three-to-seven-month deployments move both the fee and the timeline in the wrong direction. Sierra vs Decagon has the head-to-head.

Ada

Independent, 705 employees as of 31 March 2026, with a Medallia partnership signed in January 2026. Ada meters per conversation the agent handles, resolved or not — and Ada publishes the worked example arguing its own case: $0.35 per conversation against $1.50 per resolution, on 500,000 annual conversations growing 10% a year while automation climbs from 25% to 75%.

Check the assumption before you accept the conclusion. That arithmetic favors Ada only when your automation rate is rising. If your automated-resolution rate is flat at 25%, you are paying for three unresolved conversations to get one resolution, and per-resolution billing is the cheaper model.

Move here when: your deflection rate is climbing on a known curve and you want the meter to reward that.

Don’t move when: deflection has plateaued. The conversation meter charges for the failures too.

Lorikeet

The only self-serve landing on this list, and the one built for consequences rather than help-center answers — filing a claim, reissuing a card, processing a refund. Start is $1,500/month billed annually, accepted by card without a contract negotiation, for organizations under 5,000 monthly tickets; Scale is $4,000/month for 5,000–20,000. Implementation and platform fees are included on both, and there are no per-seat charges. Lorikeet bills only successfully resolved tickets and states that a ticket you are unhappy with is not billed.

The compliance ladder is explicit: SOC 2 Type 2 and ISO 27001 on Start, a standard-form DPA including a HIPAA BAA on Scale, custom security review and geo-specific data storage and inference on Enterprise. Start limits ticket-platform integrations to Zendesk, Intercom, HubSpot, Front, and Salesforce.

Move here when: the platform fee is the whole complaint, you are under 20,000 monthly tickets, or you need a HIPAA BAA without an enterprise procurement cycle.

Don’t move when: you are above 20,000 monthly tickets — that is Enterprise, and the published price stops applying at exactly the volume where Decagon’s rate starts winning.

Forethought

Right for one configuration, and a Zendesk decision now rather than a vendor decision. The platform processes more than a billion customer interactions a month for customers including Airtable, Grammarly, Datadog, and Upwork.

Move here when: you run Zendesk or are moving to it. Forethought is first-party there and the integration debt is lowest.

Don’t move when: you run a different help desk and cannot get written commitment on cross-platform support timelines, or when you are under the data floor — Forethought recommends 20,000+ historical tickets and 2,000+ tickets a month to train its per-customer models. Below that the models underperform and the price-per-resolution math inverts. Decagon vs Forethought covers the rest.

Stay on Decagon when

• You resolve more than about 8,500 conversations a month. The platform fee is amortized and the per-resolution rate is the lowest here. Leaving raises your unit cost.
• Your ops team authors the agent. Agent Operating Procedures let non-technical CS and support-ops staff define multi-step behavior in plain language rather than coded decision trees, and the AOP Copilot converts existing SOPs into production AOPs. For a team rewriting return windows and escalation thresholds monthly, that authoring surface is what the floor buys.
• You need the QA loop, not just the agent. Watchtower runs always-on QA and Experiments A/B-tests agent behavior in production. Rebuilding that as transcript exports into a separate tool is its own project.
• Your complaint was vendor risk. Decagon is the independent one in this set, at $4.5B with a completed tender. Fin and Forethought are the ones with new owners.

What migration actually costs

The AOPs are the asset, and they do not export. Years of plain-language procedures — refund conditions, identity-verification steps, escalation thresholds — are the accumulated work product, not the integration. No destination here imports them. Budget a rewrite in the target vendor’s own authoring model and treat it as a CS-Ops project measured in weeks.

Backend actions get rewired once per vendor. Decagon earns its keep by taking real actions against your systems of record. Every one of those integrations — refunds, plan changes, identity checks — is re-implemented at the new vendor, and this is the line engineering has to staff, not support.

Time to live varies by an order of magnitude. Lorikeet includes implementation and takes a card at Start. Sierra runs three to seven months and charges $50,000–$200,000 for the privilege. Forethought needs 20,000 historical tickets before its models are worth measuring. Pick the one whose ramp fits the renewal date you are trying to beat, not the one whose demo was best.

Keep the old tenant through one peak season. Resolution quality on a new agent is unknown until it meets your seasonal volume and your worst intents. Run the incumbent read-only through one peak before you cancel, and hold your CSAT baseline from the Decagon period as the number the replacement has to beat.

Verdict

If you cannot decide, do the crossover arithmetic and then renegotiate. At about 8,500 monthly resolutions, Decagon’s rate advantage cancels its platform fee. Knowing which side of that line you sit on turns this from a vendor question into a one-page calculation, and it is a number you can put in front of the vendor at renewal either way.

• Lorikeet at $1,500/month when the platform fee is the complaint, you are under 20,000 monthly tickets, and you want a HIPAA BAA without a procurement cycle
• Fin at $0.99/resolution when you are under roughly 8,500 monthly resolutions and want a published price — accepting Salesforce as the future owner
• Sierra when regulated enterprise workloads and Fortune-500 references matter more than the bill, and never as a cost-reduction move
• Ada when your automation rate is climbing on a curve you can evidence, so a per-conversation meter is the cheaper shape
• Forethought only when you run Zendesk and clear the 20,000-ticket data floor
• Stay on Decagon above 8,500 monthly resolutions, or when AOP authoring and the Watchtower QA loop are what your team actually uses

If none of them fit — you want Decagon’s action-taking depth and AOP authoring at a published price, from an independent vendor, with no platform fee — that product is not on sale today. Lorikeet is the closest and stops at 20,000 monthly tickets. The fallback is to split the work: run a cheap per-resolution agent on tier-1 volume and keep a narrow, well-instrumented deployment for the intents that touch money. AI agents for ops covers what that division of labor requires, and the AI support agent stack shows what the assembled version costs to run.