Table of Contents
- Intercom Fin AI Review: Quick Verdict
- What Is Fin & How It Works
- What Counts as a Resolution
- Fin Pricing Explained (Per-Resolution)
- How Fin's Per-Resolution Model Works
- The Per-Resolution Math at Real Volume
- The Real Total Cost of Ownership
- Hidden Costs to Watch
- Strengths
- High Resolution Quality
- Works With Any Helpdesk
- Mature Escalation and Handoff
- Strong Brand Trust and Track Record
- Limitations & Hidden Costs
- Per-Resolution Pricing Penalizes Success
- Costs Stack Across Seats, AI, and Add-Ons
- Resolution Counting Can Be Generous
- Premium Analytics Are Gated
- Why SiteGPT Is a Better Alternative for Predictable Budgets
- Best Fin AI Alternatives
- 1. SiteGPT
- 2. Tidio
- 3. Crisp
- 4. Chatbase
- 5. Freshchat
- 6. Zendesk AI
- 7. MyAskAI
- How Fin Compares to SiteGPT
- Who Should and Shouldn't Use Fin
- Fin Is a Good Fit If You:
- Fin Is Not a Good Fit If You:
- What Real Customers Are Saying
- Frequently Asked Questions
- What is Intercom Fin AI?
- How much does Intercom Fin AI cost in 2026?
- What counts as a resolution with Fin?
- Why is per-resolution pricing a problem?
- What are the best Intercom Fin AI alternatives in 2026?
- Is Fin worth it for a small business?
- Does Fin work without Intercom's helpdesk?
- How can I make AI support costs predictable?
- Final Verdict: Is Intercom Fin AI Worth It in 2026?
Created by
Do not index
Created time
Jun 25, 2026 08:15 AM
Intercom Fin AI is one of the most capable customer service agents on the market, with resolution rates that genuinely impress in production. But this Intercom Fin AI review keeps coming back to one issue that decides whether Fin is right for you: the per-resolution price. At $0.99 per resolution on top of seat costs, Fin gets more expensive precisely as it gets better, and that pricing model is the single biggest reason teams start looking for alternatives.
The AI customer service market is growing fast, and so are buyer expectations. According to Zendesk's 2026 AI customer service report, 74% of consumers now expect 24/7 service availability driven by AI. That demand is exactly what tools like Fin are built for, and it is also why the meter on a per-resolution bot can run hotter than anyone budgeted for.
This review covers what Fin is, how it actually works, what the per-resolution pricing means at real volumes, where Fin shines, the limits and hidden costs to plan for, and the best flat-rate alternatives, including SiteGPT, which trains on your existing help-center content (Intercom included) and bills a predictable monthly rate instead of charging per outcome.
Intercom Fin AI Review: Quick Verdict
Aspect | Rating | Notes |
Resolution Quality | 4.5/5 | Strong, accurate answers from connected content |
Setup & Onboarding | 4/5 | Fast if you already run Intercom; more involved otherwise |
Pricing Predictability | 2/5 | Per-resolution model makes budgets hard to forecast |
Pricing Value at Scale | 2.5/5 | Cost climbs as resolution rate climbs |
Analytics & Reporting | 3.5/5 | Good, but best features sit behind paid add-ons |
Overall | 3.3/5 | Excellent agent, pricing model is the catch |
Best for: Large support teams already standardized on Intercom that can absorb per-seat plus per-resolution costs and want a best-in-class agent without leaving the platform.
Not ideal for: Startups, SMBs, and high-volume teams that need predictable monthly costs. For these use cases, SiteGPT offers flat-rate pricing that does not punish you for resolving more tickets, while still training on your Intercom help center.
Bottom line: Fin is one of the better AI agents available in 2026. Whether it is worth it depends almost entirely on your conversation volume and how comfortable you are with a bill that scales with success.
What Is Fin & How It Works
Fin is the AI agent built by Intercom. It reads your support content, answers customer questions in natural language, and resolves conversations without a human agent stepping in. In 2026, Intercom repositioned Fin as a standalone agent: it now works with any helpdesk (Salesforce, HubSpot, Zendesk, and others), not just Intercom's own Inbox. You can run Fin on top of your existing stack or pair it with Intercom's full helpdesk.
Fin works using a Retrieval-Augmented Generation (RAG) approach. Rather than answering from a general model, it pulls from sources you connect, then generates a response grounded in that content:
- Connect your content - Fin ingests help center articles, public URLs, PDFs, snippets, and past conversations.
- Set guidance and guardrails - You define tone, escalation rules, and which topics Fin should or should not handle.
- Deploy across channels - Fin runs on web chat, email, and messaging surfaces tied to your helpdesk.
- Resolve and escalate - When Fin cannot help, it hands off to a human agent with conversation context preserved.
The result is a capable agent that answers most routine questions accurately. Intercom backs this with a "Million Dollar Guarantee" on resolution outcomes, a signal of how confident the company is in Fin's resolution rate. The quality is real. The question this review keeps returning to is what that quality costs.
Fin also leans on conversation history. By learning from past tickets and the way your human agents have answered, it can match tone and handle edge cases that a help-center article alone would not cover. That history-aware behavior is part of why resolution quality is strong, but it also means Fin needs a reasonable volume of prior conversations to perform at its best, which can slow time-to-value for newer teams without much support history.
What Counts as a Resolution
This detail matters more than any feature, because it is what you pay for. With Fin, a billable outcome is counted when "no further help is requested after Fin's last answer." In practice, that means if a customer asks a question, Fin answers, and the customer does not reply again, Fin treats the conversation as resolved and charges you for it.
That definition has drawn criticism. Customers have pointed out that Fin can count a conversation as resolved even when the customer was not actually helped, simply because they stopped replying. There are also higher-priced outcomes: lead qualification routing is billed at $9.99 per outcome rather than the standard $0.99. Understanding what triggers a charge is essential before you turn Fin loose on real traffic.
Fin Pricing Explained (Per-Resolution)
Fin's headline price is simple to state and hard to forecast: $0.99 per resolution, with a 50-outcome monthly minimum when used outside Intercom's own helpdesk. There is a 14-day free trial with unlimited outcomes and no credit card required, plus an Early Stage Program offering eligible startups a year of Fin free and steep Intercom discounts.
How Fin's Per-Resolution Model Works
The core mechanic is straightforward. Every time Fin resolves a conversation, you pay $0.99. There is no flat monthly cap on a base plan, so your AI bill is a direct function of how many conversations Fin successfully closes.
Cost Component | Price (2026) | Notes |
Fin AI Agent | $0.99 per resolution | 50 outcomes/month minimum off-Intercom |
Lead qualification outcome | $9.99 per outcome | Higher-value routing billed separately |
Intercom helpdesk seat | From $29/seat/month | Required if you use Intercom's Inbox |
Copilot (agent assist) | $35/agent/month | In-inbox AI assistant for human agents |
Pro analysis add-on | $99/month | Analysis of 1,000 conversations (CX Score, AI Topics) |
When you add Fin to Intercom's full helpdesk, you stack the $0.99 per resolution on top of per-seat pricing that runs from roughly $29 to $132 per seat per month depending on tier. The AI fee and the seat fee are separate lines, and both grow as you scale.
The Per-Resolution Math at Real Volume
Here is where the model gets uncomfortable. Imagine Fin handles 5,000 conversations a month and resolves 60% of them. That is 3,000 resolutions at $0.99, or $2,970 per month in AI fees alone, before a single seat is counted.
Now consider what happens as Fin improves. The per-resolution pricing trap documents the core problem clearly: when a bot's resolution rate climbs from 30% to 70%, the bill more than doubles even though your traffic stayed flat. You are charged for success, so every improvement in the model, every better-tuned guardrail, every new help article that lets Fin close more tickets, pushes your invoice up.
Monthly Conversations | Resolution Rate | Resolutions | Fin AI Cost/Month |
2,000 | 50% | 1,000 | ~$990 |
5,000 | 60% | 3,000 | ~$2,970 |
10,000 | 65% | 6,500 | ~$6,435 |
20,000 | 70% | 14,000 | ~$13,860 |
These figures exclude seats, Copilot, and the Pro add-on. As Listicler's chatbot cost analysis notes, the sticker price on a pricing page rarely matches the real cost at scale, and per-outcome models are the clearest example of that gap.
The Real Total Cost of Ownership
To budget honestly for Fin inside Intercom, you have to add three moving parts: AI resolutions, seats, and the analytics layer. Consider a mid-sized support team running 5,000 conversations a month with five agents and a 60% resolution rate:
Base AI fees: 3,000 resolutions x $0.99 = ~$2,970/month
Plus seats: 5 agents x ~$99/seat (Advanced tier) = ~$495/month
Plus analytics: Pro add-on = $99/month; Copilot for 5 agents = $175/month
Realistic monthly total: roughly $3,700-$3,900/month for one mid-sized team, and that total climbs as either traffic or resolution rate grows. The AI line alone (~$2,970) dwarfs the seat budget (~$495), which is exactly the surprise that catches teams who plan around seats first. By contrast, a flat-rate platform fixes the largest, most volatile line item entirely.
The deeper issue is forecasting. With per-resolution billing, a successful quarter, more traffic, a better-tuned bot, a richer help center, directly raises your costs. Finance teams that want a number they can plan against find that uncomfortable, and it is the most common trigger for evaluating alternatives.
Hidden Costs to Watch
Beyond the per-resolution meter, several costs catch teams off guard:
- Seat costs stack on top. Using Intercom's helpdesk means seats are billed separately from AI resolutions.
- The best analytics are add-ons. CX scoring, AI topics, and custom scorecards live in the $99/month Pro tier; agent-assist Copilot is another $35/agent/month.
- Higher-priced outcomes. Lead qualification routing costs $9.99 each, ten times the standard rate.
- Budget unpredictability. Multiple analyses report final bills landing 2-5x higher than the pricing page implied once seats, AI fees, and add-ons combine.
Strengths
Fin earns its reputation. Before weighing the limitations, it is worth being clear about what Fin does genuinely well.
High Resolution Quality
Fin's answers are accurate and well-grounded when your content is solid. It handles nuanced, multi-part questions better than most no-code chatbot builders, and the RAG approach keeps hallucinations low. For teams that have invested in a thorough help center, Fin converts that content into real ticket deflection.
Works With Any Helpdesk
The 2026 shift to a standalone agent is a meaningful upgrade. You no longer need to migrate your whole support operation to Intercom to use Fin. It connects to Salesforce, HubSpot, Zendesk, and other helpdesks, so you can layer it onto an existing stack.
Mature Escalation and Handoff
When Fin cannot resolve something, the handoff to a human is clean. Conversation context carries over, so customers do not have to repeat themselves, and agents pick up where Fin left off. This is an area where many cheaper tools fall short.
Strong Brand Trust and Track Record
Intercom holds a 4.5/5 rating on G2 across more than 3,000 reviews, and 4.5/5 on Capterra across 1,000+ reviews. The Million Dollar Guarantee on outcomes reflects real confidence in the product. For risk-averse enterprise buyers, that track record carries weight.
Limitations & Hidden Costs
The catch with Fin is rarely the technology. It is the economics and a few structural gaps that matter more at scale.
Per-Resolution Pricing Penalizes Success
This is the headline limitation. A model that charges per resolution means the better your AI performs, the more you pay. Founders and operators repeatedly report that they can no longer predict their support bills, which makes budgeting and forecasting difficult. A viral week, a product launch, or a seasonal spike can turn a manageable invoice into a shock.
The better Fin works, the more you pay. If Fin resolves 60% of 5,000 monthly conversations, that is $2,970 in AI fees alone, before seats.
The reality is that per-resolution pricing aligns the vendor's revenue with your traffic, not with your budget. That is great for the vendor and stressful for the buyer.
Costs Stack Across Seats, AI, and Add-Ons
Fin is rarely a single line item. Between per-seat helpdesk costs, per-resolution AI fees, Copilot, and the Pro analysis add-on, the total cost of ownership is far harder to model than a flat subscription. Teams budgeting for seats alone are often surprised when AI fees alone exceed the seat budget.
Resolution Counting Can Be Generous
Because a "resolution" is triggered when a customer simply stops replying, some billed outcomes may not reflect a genuinely helped customer. When a human takes over from Fin mid-conversation, the original resolution can still be counted in some cases. Over thousands of conversations, that ambiguity adds up.
Premium Analytics Are Gated
The reporting that helps you justify and optimize an AI deployment, CX scoring, AI topic clustering, and custom scorecards, sits behind the $99/month Pro tier. The base experience gives you the agent, but the insight layer costs extra.
Why SiteGPT Is a Better Alternative for Predictable Budgets
For teams that want Fin-class resolution without a meter on every success, SiteGPT addresses the core problems above directly. SiteGPT trains on your existing content, including your Intercom help center, websites, sitemaps, PDFs, YouTube, GitHub, Notion, and cloud storage, and charges a flat monthly rate rather than per resolution. Your bill does not rise when your chatbot gets better at its job. You also get native human escalation, auto-sync to keep answers current, and affordable white-labeling, all on predictable pricing that starts at $39/month.
Best Fin AI Alternatives
If Fin's per-resolution model does not fit your budget, several flat-rate and lower-cost alternatives deliver strong AI customer service without the meter. Here are the best options for 2026, starting with the top recommendation.
1. SiteGPT
Best for: Teams that want flat-rate, content-trained AI support without per-resolution surprises.
SiteGPT is purpose-built for the exact buyer who finds Fin too unpredictable. It ingests your help-center and website content (including a live connection to Intercom help-center content) and turns it into an accurate AI agent, billed at a flat monthly rate.
Key Features:
- Trains on websites, sitemaps, files, YouTube, GitHub, Notion, Google Drive, Dropbox, Box, OneDrive, Confluence, Zendesk, and Intercom content
- Auto-sync keeps answers current (monthly on Growth, weekly on Scale, daily on Enterprise)
- Native "Escalate to Human" handoff with team notifications
- White-labeling for $39/month and support for 95+ languages
Pricing: Starting at $39/month (4,000 messages, 1,000 pages). Growth $79/month, Scale $259/month.
Pros:
- Flat-rate pricing that does not penalize higher resolution rates
- Broad content ingestion, including live Intercom help-center sync
Cons:
- Message-based tiers rather than per-outcome billing may suit some enterprise procurement less
- Not a full ticketing helpdesk on its own
SiteGPT is the strongest fit for startups, SMBs, and high-volume teams that want best-in-class resolution on a budget they can forecast to the dollar.
2. Tidio
Best for: Small businesses wanting AI plus live chat with a generous entry point.
Tidio pairs live chat with its Lyro AI agent and is popular with smaller teams.
Key Features:
- Lyro AI for automated conversations
- Live chat, ticketing, and visitor tracking
- Easy widget setup
Pricing: Free plan available; Starter $24.17/month, Growth $49.17/month (annual). Lyro AI conversations billed separately.
Pros:
- Affordable entry tiers
- Strong reviews: 4.7/5 on G2 across 1,400+ reviews
Cons:
- Lyro AI conversations are metered above the included allowance
- Less depth for complex, multi-source knowledge bases
3. Crisp
Best for: Teams wanting per-workspace pricing instead of per-seat or per-resolution.
Crisp bills per workspace, so adding agents does not increase the bill.
Key Features:
- Shared inbox, chatbot, and AI features
- Per-workspace pricing model
- Multichannel messaging
Pricing: Free plan; Mini $45/workspace/month, Essentials $95/workspace/month, Plus $295/workspace/month.
Pros:
- Predictable per-workspace cost
- Solid ratings: 4.5/5 on G2
Cons:
- AI usage is capped on lower tiers (50 uses/month on Essentials)
- Unlimited AI requires the top Plus plan
4. Chatbase
Best for: Quick-to-deploy custom chatbots trained on your data.
Chatbase is a fast no-code builder for AI chatbots.
Key Features:
- Train on PDFs, URLs, and Q&A pairs
- Multiple AI model options
- WhatsApp and Slack integrations
Pricing: Hobby $40/month, Standard $150/month, Pro $500/month. Credit-based.
Pros:
- Very fast setup
- Good ratings: 4.7/5 on G2
Cons:
- Credit-based pricing adds its own unpredictability
- White-labeling costs $199/month
5. Freshchat
Best for: Teams in the Freshworks ecosystem wanting AI plus live chat.
Freshchat from Freshworks offers Freddy AI on top of live chat.
Key Features:
- Freddy AI bot sessions
- Omnichannel messaging
- CRM and helpdesk integration
Pricing: Free for up to 10 agents; Growth $19/agent/month, Pro $49/agent/month, Enterprise $79/agent/month (annual). Freddy AI is an add-on.
Pros:
- Free tier for small teams
Cons:
- Freddy AI features cost extra above base plans
- Bot sessions limited on lower tiers
6. Zendesk AI
Best for: Teams that want AI included in suite pricing, not billed per resolution.
Zendesk AI bundles AI into its suite plans rather than charging per outcome.
Key Features:
- AI agents and agent assist within the Zendesk suite
- Mature ticketing and reporting
- Large integration ecosystem
Pricing: Suite Team $19/agent/month, Growth $55, Professional $115, Enterprise $169 (annual). AI included in suite plans.
Pros:
- AI bundled rather than per-resolution
- 4.3/5 on G2 across 5,800+ reviews
Cons:
- Per-agent pricing climbs quickly for larger teams
- More platform than smaller teams may need
7. MyAskAI
Best for: Adding an AI deflection layer on top of Intercom, Zendesk, or Freshdesk.
MyAskAI sits in front of your existing helpdesk to deflect tickets.
Key Features:
- AI ticket deflection for major helpdesks
- Per-ticket pricing with clear overage rates
- Fast setup with a 30-day free trial
Pricing: Pro $199/month (1,000 tickets), Scale $499/month (2,000 tickets), Enterprise from $999/month.
Pros:
- Designed to layer onto Intercom and others
- Transparent overage rates ($0.10-$0.12/ticket)
Cons:
- Still volume-based above the included tickets
- Branding removal and API are paid add-ons on Pro
MyAskAI is a sensible middle ground if you want to keep your current helpdesk but deflect more tickets before they reach an agent. It does not solve the predictability problem as cleanly as a fully flat-rate tool, but its tiered ticket allowances and published overage rates make costs easier to model than a pure per-resolution meter.
How Fin Compares to SiteGPT
For the buyers most affected by per-resolution pricing, the clearest contrast is Fin versus SiteGPT.
Feature | Intercom Fin | |
Pricing Model | $0.99 per resolution + seats | Flat monthly rate |
Cost as resolutions rise | Increases with success | Stays the same |
Content Sources | Help center, URLs, PDFs, snippets | Websites, sitemaps, files, YouTube, GitHub, Notion, Google Drive, Dropbox, Box, OneDrive, Confluence, Zendesk, Intercom |
Auto-Sync | Supported | Automatic monthly/weekly/daily refresh |
Human Escalation | Native | Native with team notifications |
White-Label | Enterprise | $39/month |
Starting Price | $0.99/resolution (50/mo min) | $39/month (4,000 messages, 1,000 pages) |
Best For | Large Intercom-standardized teams | Teams wanting predictable, flat-rate AI support |
The deciding factor is predictability. Fin gives you a best-in-class agent whose cost rises with its own success. SiteGPT gives you flat-rate pricing and trains on the same kind of content, including your live Intercom help center, so you can forecast your support automation budget without watching a meter.
Who Should and Shouldn't Use Fin
Fin Is a Good Fit If You:
- Already run Intercom as your primary helpdesk
- Have predictable, moderate conversation volume
- Can absorb per-seat plus per-resolution costs
- Want a best-in-class agent and value the Million Dollar Guarantee
- Need mature, context-preserving human handoff
Fin Is Not a Good Fit If You:
- Are a startup or SMB watching every dollar
- Have high or spiky conversation volume
- Need to forecast support costs precisely each month
- Want your AI bill to stay flat as resolution rates improve
- Prefer flat-rate or per-workspace pricing over per-outcome billing
For these needs, SiteGPT addresses the limitations with flat-rate pricing, broad content ingestion (including live Intercom help-center sync), and native escalation, so resolving more tickets never raises your bill.
What Real Customers Are Saying
Intercom and Fin earn strong ratings overall, with pricing predictability the most consistent complaint.
Platform | Rating | Review Count |
G2 | 4.5/5 | 3,000+ reviews |
Capterra | 4.5/5 | 1,000+ reviews |
The praise centers on resolution quality, polish, and the maturity of the platform. The criticism is almost entirely about cost: an analysis of 200+ Capterra reviews found that founders and operators struggle to predict their support bills under the per-resolution model, with some teams budgeting a few hundred dollars for seats and landing past $1,300 because of AI fees alone.
Teams that switch to flat-rate tools often cite predictability as the main reason. As one SiteGPT customer put it:
"An easy solution to provide round the clock support for your customers - without having it feel like 'just another chatbot'... The conversations speak for themselves, the positive feedback from the customers go a long way, and the sales both (made & recovered) speak for themselves." Brent Burrows II, Co-Founder, Starfish Web Ventures
"We no longer talk about 'the bot' as just a tool. Over these months, we've expanded our team with five new virtual colleagues who work 24/7, never get tired, and are surprisingly competent and pleasant." Ecommerce Customer (5-store multilingual deployment)
Frequently Asked Questions
What is Intercom Fin AI?
Fin is Intercom's AI agent for customer service. It reads your support content using a Retrieval-Augmented Generation approach, answers customer questions in natural language, and resolves conversations without a human agent. As of 2026, Fin works as a standalone agent with any helpdesk (Salesforce, HubSpot, Zendesk, and more), not just Intercom.
How much does Intercom Fin AI cost in 2026?
Fin costs $0.99 per resolution, with a 50-outcome monthly minimum when used outside Intercom's helpdesk. Higher-value lead qualification outcomes are billed at $9.99 each. If you use Intercom's helpdesk, seats start at $29/seat/month on top of the AI fee. Add-ons include Copilot at $35/agent/month and a Pro analysis tier at $99/month. There is a 14-day free trial.
What counts as a resolution with Fin?
A billable outcome is counted when no further help is requested after Fin's last answer, in other words, when the customer stops replying after Fin responds. This has drawn criticism because a conversation can be counted as resolved even if the customer was not genuinely helped. Lead qualification routing is a separate, higher-priced outcome at $9.99.
Why is per-resolution pricing a problem?
Per-resolution pricing means your bill rises as your AI gets better. When a resolution rate climbs from 30% to 70%, the cost can more than double on the same traffic. This makes budgets hard to forecast and effectively penalizes you for success, which is why many teams look for flat-rate alternatives like SiteGPT.
What are the best Intercom Fin AI alternatives in 2026?
SiteGPT is the strongest flat-rate alternative, training on your existing content (including your Intercom help center) and billing a predictable monthly rate. Other options include Tidio, Crisp (per-workspace pricing), Chatbase, Freshchat, Zendesk AI (AI bundled into suite plans), and MyAskAI for a deflection layer on top of an existing helpdesk.
Is Fin worth it for a small business?
For most small businesses, the per-resolution model makes Fin hard to justify, because costs are difficult to predict and rise with volume. Smaller teams are usually better served by a flat-rate tool like SiteGPT starting at $39/month, which delivers strong resolution quality without a meter on every success.
Does Fin work without Intercom's helpdesk?
Yes. As of 2026, Fin is a standalone agent that connects to other helpdesks including Salesforce, HubSpot, and Zendesk. You can run Fin on top of your existing stack and pay $0.99 per resolution with a 50-outcome monthly minimum, without adopting Intercom's full Inbox.
How can I make AI support costs predictable?
Choose a tool with flat-rate or per-workspace pricing rather than per-resolution billing. SiteGPT charges a fixed monthly rate (from $39/month) regardless of how many conversations it resolves, while Crisp bills per workspace. Both keep your bill stable even as resolution rates and traffic grow.
Final Verdict: Is Intercom Fin AI Worth It in 2026?
Fin is a genuinely strong AI agent. The resolution quality is high, the handoff is mature, the standalone-agent shift makes it more flexible than ever, and Intercom's track record gives enterprise buyers confidence. If you already run Intercom and your volume is predictable enough to model, Fin is a defensible choice.
The catch is the pricing model, and it is a big one. Per-resolution billing means Fin gets more expensive exactly as it gets better, stacking on top of seat costs and gated analytics add-ons. For startups, SMBs, and high-volume teams, that unpredictability is hard to live with.
The Bottom Line: Fin is worth it for large, Intercom-standardized teams that can absorb per-seat plus per-resolution costs. For everyone watching the budget, the better value is a flat-rate tool that does not charge more when your AI succeeds.
For most businesses, SiteGPT offers a more predictable path to the same outcome:
- Flat-rate pricing from $39/month that never rises with resolution rate
- Training on your existing content, including a live Intercom help-center connection
- Native human escalation and auto-sync to keep answers current
- Affordable white-labeling at $39/month and support for 95+ languages
Ready to see the difference?
---
Last updated: June 2026. Pricing and features verified from official sources as of June 2026.
Written by
