What is Penfriend.AI?

Simply put, Penfriend AI creates the most realistic human-like first draft copy within minutes using AI.

They position themselves as saving you hundreds of hours on prompting (which it does), but what it does behind the scenes is where the time savings happen.

During the article process, you also get the added benefit of a Title Generator and an SEO-backed outline.

Who made Penfriend and How it Started

Penfriend was co-founded by these three folks below:

• Inge von Aulock – COO
• Tim Hanson – CCO
• John Harrison – Head copywriter

I could try to explain their history, but I wouldn’t do it justice because John’s copy is already perfect. Here’s an excerpt from the launch campaign they created.

Why John, why did you do this?
Because I needed Penfriend around 5+ years ago when “the blog” was one of my 1,000 tasks.

I was part of a tiny marketing team in a startup with huge ambitions. The uncrackable puzzle was between quality and quantity. We didn’t have the time or money to invest in content [rather, we chose not to…] and so traffic dwindled and the blog became a box on a marketing checklist I rarely checked.

By the way, Tim was our contracted content strategist back then. We started out-ranking banks for mortgage related keywords. So, Tim’s strategy works. It’s just we couldn’t crack the quality/quantity puzzle.

Inge needed Penfriend about 7 months ago too
Inge had contracted Tim for content strategy on her AI tools directory, TopApps.

She spent a lot of money getting content made for the blog. It was insanely expensive. She contracted several freelance writers and 3 content agencies! She even hit the limit on what those agencies could produce each month.

The content ranked, and fast. But it was in no way a sustainable way to build a traffic generating blog.

So Inge proposed an idea.

AI-ify Tim’s brain
Turn that man into a robot please.

Make an AI blog article writing tool that replicates Tim’s content strategy.

That was the plan.

Tim’s been an SEO content strategist for over a decade now. And he’s a systems based guy, which meant when ChatGPT came along last year, he feverishly started building prompts to replicate his processes. Processes that have delivered £24.6M in the last 3 years. So far, he’s spent over 600 hours refining prompts that build high quality blog articles.

And so, for the last month, we’ve been working with a dev agency turning Tim’s prompts into a tool anyone can use.

[…]

– Official Penfriend waitlist campaign email by John

And of course there are the developers and I’m sure other key people involved to make this product happen.

Needless to say, this isn’t just another wrapper app around Chat GPT put together overnight.

How does Penfriend AI work?

Now that you know what it is and who’s behind this product, let’s get into how it works.

Get the Free Credit

First, it’s worth noting that everyone who signs up gets three free credits, but before you claim the credit, read below on what types of articles NOT to use with this writer so you get the most out of your test. (hint: using Penfriend to write this review would not have worked out – more on this later).

Penfriend Onboarding

You’ll land on the dashboard after you go through the usual sign-up flow (no credit card required). Just the usual email and password.

There’s no need to sign up for a paid plan yet.

The dashboard is very minimal by design:

My Penfriend – this is your home for the dashboard and where you’ll spend most of your time going through the drafting flow on the right.

History – once you finish your articles, they will be archived here so you can retrieve them easily.

Profile – Edit your name – that’s it for now.

Billing – where you subscribe when ready. Details on pricing are below.

In the top right-hand corner, you’ll see your total credits.

As you can see, the next logical step is to add your keyword and press that nice orange Generate button. This is one of the things I enjoy about this tool: just how straightforward it is.

How do credits work

Your credits are in the top right corner of your My Penfriend home screen. You should start with 1 credit when you sign up.

Once you press “Generate Outline” the writer will start creating your article outline and the credit will be consumed. This step is immediately after the Title Generation step.

Creating an article with Penfriend

These are the steps to starting an article:

• Add a keyword to generate titles
• Select a Title to generate Outlines
• Confirm the Outline to generate the Draft

Generating Titles

To start the draft process, you need to have decided on the keyword to generate the article with. Assuming you’ve done the keyword research already, type in the keyword and press Generate.

Notice that a credit is not consumed during Title generation. This saves you time from going back and forth to another tool like Chat GPT.

If you don’t like the Titles, just click the “X” and Generate again until you find a compelling title.

Behind the scenes, Penfriend is going out to the SERPs, analyzing the results of your keyword and suggesting ideal titles for your topic while giving it a score.

According to their website it pairs down the 19 titles into the top 5 based on score – though I’ve been getting 6 titles sometimes.

Generating Outlines (uses 1 credit)

When you are ready to generate an outline, select the Title from the proposed list and click Generate Outline. Penfriend will then create a compelling outline based on its proprietary prompts.

What sets their outlines apart from other AI writers is that each section of your article and outline is crafted with a dedicated prompt. This is a much better method than trying to one-shot the article like many people tend to do.

Here’s an outline generated for the keyword “penfriend ai review” as a test. What is interesting about this outline is that despite the lack of SERP data (this tool just came out Dec 1 2023.), it still nailed it. You’ll see that I used several headers from the outline in this review.

The outline had these helpful features.

• Each header was clearly marked and well written, H2, H3 etc.
• It included an introduction and concluding paragraph bullets
• It included the ideal call to action based on its research
• The logical order of the headers makes sense
• It adds related keywords which is always helpful to have in the back of your mind when writing.

Is the AI Generated Output Good?

Now onto the main event. Because the keyword “Penfriend ai review” doesn’t have enough data, I will show you the draft for “Jasper ai review,” which is a very similar topic, and how I would have expected this article to turn out had it had enough data.

Here’s a snippet of the output below, but here is the full article draft output from Penfriend, if you want to read the entire draft.

What I liked about the output

• Overall, the draft quality is excellent, the most human like content I have seen from AI. Sentences like these have never come from Chat GPT when I prompt it: “Over time, I’ve taken Jasper AI under my wing. Not for the sake of experimenting, but because it has proven its worth”
• The output was generated in minutes. No more waiting 5 business days for a writer to reply.
• There are really helpful tips that nudge you into adding more to the content to enhance it. They are usually labeled starting with “MANUAL CHECK”
• Word count is very high (in a good way) – this example outputs over 3500 words which is what I would expect from a topic like this.
• Uses elements of EEAT – the output showcases many elements required for EEAT with sentences like “Personally, I found the pricing of Jasper AI justified, given the quality and versatility of the service provided.”

What I didn’t like about the output

• Introductions are a bit over the top. Let’s face it, intros are hard to write. Everyone is going to have a different style, and right now, I just prefer to craft my own.

• Three bullets are added to most H2s. I’m not sure if this is an SEO thing I just don’t understand or this was just a note for the editor, but right after the H2s, there are almost always three bullet points before going into the H3s. Typically, I write a lead in a sentence and move right to the H3. Some documentation around the intent of this feature would be great.

• The conclusion paragraph had a headline. Again, this might be obvious for someone else, but I couldn’t figure out why there was a headline in the conclusion. Was this meant for using that headline for the conclusion? I would have preferred the H2 to just use the headline instead of having it beneath and maybe “Conclusion” in brackets after the H2. I’m just being nit-picky at this point. But again, documentation might be warranted here.

• Sometimes “manual check” tips get buried in the copy. Just a heads up that sometimes these tips get lost in the copy, but we should all be proof-reading our articles for this type of stuff anyway.

Articles you should NOT write with Penfriend

As with all AI writers (and tools in general), you need to know when to use them and when not to. As a rule of thumb, you don’t want to use this writer on any topic that has little to no SERP data. Here’s an example.

I really wanted to use Penfriend to write this article you’re reading now, but I suspected it would be a bad idea and wanted to confirm.

I started with Google Trends and as you would expect, with a new tool, very little interest.

Another quick SERP check before I added this keyword into Penfriend shows that Google has no idea really what this is about. Only the branded and one other website related to the product show up. Everything else is about another similarly named product.

At this point, I only used the Penfriend Outline feature but not the actual writer itself. I didn’t want to take the chance of writing generic content on a topic that wasn’t heavily researched to date.

Since the tool looked at the SERPs, it was likely to get confused with the other Penfriend product.

Penfriend Pricing – Is it Worth The Investment?

For me, Penfriend is absolutely worth the investment, but here is why.

Assuming one of the reasons you want Penfriend is either to save time, money or both, all we need to do is account for the current expenses.

First, here are the subscription tiers for Penfriend. They have monthly and annual plans to fit any budget. I’ve also broken it down by a per article cost for simplicity.

Depending on the project, I pay between $100–$300 per article. But, before I can even get the article, I have to do my keyword research, generate an SEO-optimized outline, and put it into SurferSEO before it is handed off to a writer.

Round trip, before Penfriend, the time from when I generate the outline to get an article back is 3-5 business days – minimum.

That would typically be considered fast by most standards. The total cost of my previous setup is several hundred dollars plus at least 1 hour of my time to create a good SEO outline.

With Penfriend, I just need the keyword, and I have my draft back within 10 minutes. And the total cost? On my Advanced tier plan, a generous $14.11 per article.

So the comparison really needs to be between whether the quality you’re receiving from your freelance writers today is better or worse, given the savings of time and per article expense for your particular content pipeline.

You need to also factor in editing time. Again, it was simple for me: reduce the number of articles I outsourced and have Penfriend do more heavy lifting for instant savings.

Penfriend vs Other AI Writers (Alternatives)

All of the AI writers are so different now that it’s not even an apples-to-apples comparison.

Here is how I use the writers based on my needs for the article.

• Penfriend.ai – Best for when you need a really human like draft and a great first draft quickly. Penfriend only handles blog posts right now.
• Frase.ai – Best for when you need an SEO keyword optimizing tool, I use this in conjunction with Penfriend AI – read my review here.
• SurferSEO – similar to Frase, when I need an SEO keyword optimizing tool, I will also use this with Penfriend AI
• ZimmWriter – Best for when you need to have full control of the headers you are writing about or doing bulk ai writing on a budget (will add a review soon)
• Jasper.ai – Best for when you want to quickly generate ai content for the entire marketing pipeline and not just blog posts (social media etc.)

Tips When Using Penfriend

• The output is in British English – You’ll need to do a quick find and replace if you are writing for the American market.
• You don’t need a markdown converter – When you copy from the app, it has all the right headers, no need to use a markdown to HTML converter like other apps.
  • This has also changed recently. You’ll now require a markdown converter when you copy the text from Penfriend’s output.
• Be sure to remove MANUAL CHECK lines – The manual check tips are to help you make your article more unique. Be sure to remove these as you review them. Although they mostly start with the term “MANUAL CHECK” sometimes they use other phrases. I like to highlight them as I go so I don’t miss it later.
• Get help through email at support@penfriend.ai – Sometimes apps have issues, and these guys reply fast, even when they are supposed to enjoy their weekends (Thanks, John!)

Recap

Well, there you have it. If you’re looking for a streamlined first draft machine, you just stumbled upon one of the coolest new apps on the market that won’t break the bank.

If you are already hiring freelance writers, you owe it to yourself to try this out (and the first three articles are free). The best part is it will only get better from here. I can only imagine how many iterations the core prompt is going to go through in the coming months.

Out of the box, Penfriend AI is smooth like butter and while it won’t wholesale repalce my current AI apps, it is a welcomed addition to my content pipeline.

Change List

Since this is a new tool, the features may change quickly. I will document the main changes to my review in this section. Check back often to stay up to date.

February 2, 2024 – Penfriend allows 3 free credits now.

December 12, 2023 – Credits are now consumed when the Generate Outline button is pressed and not when the Generate Article is pressed.

This is the year of the custom GPT intelligent chatbots. Chatbase is yet another entry to what appears to be hundreds of GPT chat programs using natural language processing. But can it do what the others have not been able to achieve – a chatbot that can give answers with precision?

In this article, I break down everything you need to know about Chatbase and whether it is for you.

Key Takeaways

What is Chatbase.co?

Chatbase is a custom AI chatbot tool that lets you train it using your own dataset. Instead of training it on the entirety of the internet, you can use Chatbase to train just on the pages that are important to your website or brand.

There are a variety of ways to input source data, including Q&A style (similar to Frequently Asked Questions) uploading PDFs, .docx among other file types.

What makes this tool particularly powerful is the ability to find out the pain points of your customers quickly. This is something that some brands never actually achieve and leads to a lack of product market fit.

When programmed properly, this tool can cut out hours a week of repetitive customer service ticketing so you and your team can focus on revenue-generating tasks while improving the conversational experiences for your website visitors.

Who can benefit from a Chatbot?

While you could make a case that everyone could benefit, I do find that brands, and in particular those with a SaaS or e-commerce product, benefit the most.

By using a chatbot builder, you can tailor the answers to drive traffic directly to your category or landing pages. This helps your prospect bypass the pain of having to learn your site navigation. What’s easier than just asking and getting a direct and useful answer to where you want to go?

Now, with natural language and the ability to custom program your chatbot, there isn’t a better time to add a service like Chatbase to your tech stack. And when I say “program,” I’m not talking about coding. These are still under the category of no-code tools, so there are no excuses for not making the user experience better with an NLP chatbot.

Ok, let’s get started with the main features.

Primary Features

• Upload multiple sources of (training) data: Chatbase accepts quite a few file types, including .txt, .pdf, .doc and .docx files. They also have a dedicated Question and Answer feature, which makes training very simple and straightforward.
• Embed chatbot on an unlimited number of websites: If you have several websites for different regions, you can use the same bot on all of them without having to train them all separately.
• Suggested Messages: This option allows you to create “buttons” on the chatbot that simulate pre-filled chat options like “How to Contact Customer Service.” Typically, you will put your most asked question here or a promotional offer.
• Ability to lead capture: Whether you want a visitor’s contact info so you can reply back to them or you want to capture a potential prospect, the lead capture form can capture Name, Phone number, and Email out of the box.
• Ability to revise answers to spec: Your chatbot will need time to refine its answers. You can speed this along by manually revising it so it never gets that question wrong again. Add hyperlinks to get the answer to your customer quickly.
• Customizable base prompt for more control: While the Chatbase base prompt is good enough to get you going, you may want to update it to customize it for your brand or style. Chatbase allows you to get access to the primary prompt that runs the chatbot.
• ChatGPT API Keys: Bring your own ChatGPT API key on the paid tier if you need to scale past the message limits on your tier.
• Use GPT-3.5 or GPT-4 models: For Standard plans and higher, you can use GPT-4 models for an even deeper chat experience. Note, the costs are higher with GPT-4, and you’ll need to have a paid ChatGPT subscription to have access to it.

Pricing

With flexible pricing models, you’ll be sure to find a suitable tier for your business.

How does it work

Set up your Chatbase chatbot in 3 easy steps.

1. Upload data

One of the key features of Chatbase is the ability to upload multiple types of data.

• Files: Upload a wide range of files in either .pdf, .doc, .docx. or .txt and if the document can be copied or the bot won’t be able to train it. I found it particularly useful to upload company PDF Files that were created from the product team to help with the learning.
• Text: this field allows you to just enter information in a plain text field.
• Website: add your sitemap or URL and the bot will start to crawl the website to find pages. Then you can select which pages to include in the learning.
• Q&A: The bulk of your training time will be spent here. Every time you “revise” an answer, it will show up in the Q&A. You can also manually input Q&A as a starting point for the chatbot.
• Notion: A newer feature, the chat bot can now use Notion DB as a learning source.

With all of these options, you can create a robust knowledge base to ensure there will be a high chance of customer engagement with the chatbot.

2. Adjust Chatbot Settings

In the Settings tab, you can configure the following

• Set the base prompt
• Adjust your model, GPT 3.5 Turbo or GPT 4.0
• Customize the Chat Widget: You can change the colors of the chat bubble, chat messages, and the chat icon (profile picture)
• Whitelist specific domains
• Look up your list of leads
• Setup notifications
• Setup custom domains

3. Embed the chatbot

Chatbase gives you two options to embed your chatbot. You can either add it to a specific page or to the entire website using their snippet codes.

That’s it for setup. From this point on, you just want to visit your Dashboard section on a regular basis to revise the answers – see below.

Customer Service from Chatbase team

You can reach customer support at Chatbase at support@chatbase.co. They seem to have a couple of different members answering questions, and I get replies within 24 hours. The only way I could see this being better is if they had someone on live chat during core hours.

For everything else, you can just go directly to their Chatbase Chatbot which of course, is completely trained on their own dataset. I’ve done this a couple of times to ask about basic account and setup questions and have found it to be useful.

To date, I haven’t had a single negative experience, and they’ve been able to help me with some really nuanced problems for the chatbot I setup for work.

If you need more support they have some documentation here:

• Developer docs
• Chatbase guide

Practical Use Cases for Businesses

Customer Support Chatbot (For most Businesses)

The most obvious use case is to use Chatbase to support all your customer service requests to improve customer satisfaction. By having it handle all of your frequently asked questions, you can significantly cut the time your CSRs spend answering really mundane questions. Their time is better used to solve more complex problems or follow up with past customers for an overall better user experience with your brand.

Businesses that book appointments

With Chatbase’s Suggested Messages feature, you can lead customers down specific user paths that your visitors are looking for most often.

Let’s say you own a salon, and your customers just want to see your availability or store hours. Simply program these Suggested Messages in the Settings tab and cut down the time it takes to find this information.

E-Commerce Businesses

For e-commerce businesses, you can promote specific items using the Suggested Messages feature or train the entire dataset of your product lines so that customers can ask specific questions about ingredients, sourcing, shipping, pricing, etc.

Make sure to go back into your past customer’s questions and add them to the training dataset to get a comprehensive chat experience.

My Personal Experience

I first came across this by chance when I was browsing the AppSumo directory. I had seen that a few other companies have had customizable GPT chatbots but none of them seemed that interesting.

The main reason I purchased Chatbase was to make a prototype for work. I currently work in the Education Tech sector and during the summer we don’t have as many customer service staff, and I personally thought it would be more beneficial to the company if we could reserve some of their time working on more difficult problems than “How can I reset my password” type of queries.

So I picked up the Tier 2 license from AppSumo in case any other clients wanted to try this out, and I started programming it.

Right away, you could see some interesting results. The bot was very easy to train, and I had a trained, working prototype within the day.

I took the questions I found in our HubSpot database as well as our own website’s FAQ and added all the data to the bot’s “brain.”

Then, I started testing it for chatbot performance. It is a good idea to spend a few hours testing the bot to make sure it doesn’t make up information or give incorrect information. Every brand will have a different level of tolerance for non-perfect outputs, so do use caution if you are implementing this for work or clients.

For example, one of the test questions I asked was, “Who was the CEO of our company?” and it completely got the founder incorrect. The reason was obvious: most of the data kept talking about another notable educator instead of our CEO, so it naturally “decided” – incorrectly so – this was the founder.

After I had a good working prototype (as in, it got most of the answers correctly), I pitched the chatbot to our CEO and he was onboard to greenlight the project.

We slowly rolled it out to only a few web pages at a time so we could get questions from the public. Each day I would revise the answers and add hyperlinks to our product pages to further streamline the chatbot.

The maintenance for the chatbot is pretty low, depending on the volume of chats. I can check and revise about 10 chats quickly within a half hour. Over time, I expect this maintenance to go down as we will have answered the majority of the questions.

Downsides

While I was quite happy with my Chatbase experience, it wasn’t all roses, and other customers have expressed some of the same issues I’ve seen.

I like the basic idea of Chatbase but I get so frustrated with the bot answering questions it shouldn’t and making up links that don’t exist on the website

comment from appsumo member

Unfortunately, this is a problem with GPT in general and not only Chatbase. The chatbot will make up answers when not trained thoroughly. See troubleshooting tips below to fix this.

Other complaints have been the lack of customer service response, but this was mainly only during the initial launch of the lifetime deal. I personally have not had this problem with the company.

I’m also not a fan of the current UI for the dashboard. It has a two tiered menu which is confusing and their training section is under the header of “dashboard” which is not intuitive. The integration documentation is lacking as they talk only about how to integrate the bot into other tools, but no documentation on how to actually use it once it’s integrated.

The quality of your output is going to be directly related to how long you train your chatbot. If you spend only the time to complete the setup, but no time training, you can expect to get poor output. The key to great output is increase the training time and match your inputs with what a visitor would type in – this takes testing time and slowly roll out the chatbot.

Troubleshooting Tips

1. Spend time to train the bot

This step alone will solve the majority of your problems.

Spend the extra hour or two when you launch to test your bot for obvious questions and then the more nuanced questions.

Use your own customer service data and ask your team to write the most frequently asked questions and answer them all in the Q&A section of the Chatbase dashboard.

Chatbots do well when there is a large dataset, less conflicting information and clear question and answers. If you leave it with less information, it will either not know how to answer it, or worse, make up answers.

2. Chatbot keeps saying the same output even after revision

I ran into this problem where the chatbot would not give the proper hyperlink even though I repeatedly revised the answer to do so.

To solve this, update your base prompt to the following:

I want you to act as a document that I am having a conversation with. Your name is "AI Assistant."

You will provide me with answers from the given info. If the answer is not included, say exactly "Hmm, I am not sure." and stop after that. 

When I provide you with information in [] then (), embed the link in the () to the text in [] and provide the text that is in [] as a hyperlink.

Refuse to answer any question not about the info. Never break character.

This allows your links to output consistently.

3. Roll out your chatbot slowly

I wouldn’t recommend launching your brand new chatbot to the entire market or website you plan to use it on. Instead, have it show up on some pages that get a reasonable amount of traffic (not your highest traffic pages).

Check and revise the answers everyday. When you revise your answers they go into the Q&A section of the tool so you can always go back and delete duplicates or answers that are conflicting.

3. How to add hyperlinks

The syntax for adding hyperlinks is as follows:

Type your answers in the answer box and then put the [anchor text in brackets like this](https://google.com). Then, follow it up immediately without a space with the URL in parenthesis.

Then make sure you test the question and answer to ensure the link will show up.

4. Prompt to get consistent output

Sometimes, you want the output to always be the same. For example anything around customer service, you may want the output to say the anchor text and URL exactly the same way.

To do this, follow this sample prompt.

Say exactly "To contact customer service go to the [Contact Page](https://yourwebsite.com/contactpage )" and stop after that.

The key is to wrap the prompt in quotes with the instructions to say exactly the output.

Final Thoughts

While Chatbase still has a long way to go to be considered a full custom chatbot assistant, it has some serious potential.

In its current state, with a simple setup and the ability to quickly train the bot and maintain it over time, there is no excuse for saving your customer service rep’s time with this tool.

I would love to see better integrations in the future where support tickets could be automated based on certain language triggers or sequences. And the ability to integrate right into a CMS without software like Zapier would be a plus as well.

If you haven’t tried any AI product before, this is a fun one to play around with since it has a Free tier option.

FAQ

How does the Slack integration work?

The installation for the Slack integration guide can be found here, but what you may be wondering is what does the integration actually do?

By integrating Chatbase with Slack, it allows you to “talk” to the chatbot directly from Chat.

Once you finish integrating Chatbase, you will need to invite the Chatbase app to your channel. Do this by selecting the correct channel and do the “/invite” command. Select the “Add apps to this channel” and select Chatbase.

If Chatbase is not on the list of apps, then it was not integrated properly. Return back to the help doc and try again.

After you add Chatbase, you can chat with it by mentioning it. For example, “@chatbase how are you doing” will start a dialog with the chatbot.

If you have trained your chatbot on your internal documentation, then you could use this Slack integration to quickly find the right documents, saving your organization time.

How many more tokens does GPT-4 cost compared to GPT-3.5 Turbo?

According to Chatbase it cost 20 times more credits. For example, 1 GPT-4 message = 20 message credits vs 1 GPT-3.5 message = 1 message credit.

gpt-4 is much better at following the base prompt and not hallucinating, but it’s slower and more expensive than gpt-3.5-turbo (1 message using gpt-3.5-turbo costs 1 message credit. 1 message using gpt-4 costs 20 message credits)

help text in chatbase around message credits

Learning how to write quality content that will rank in Google search engines can seem like black magic sometimes. With Frase.io, they lift the curtain to equip you with the tools to consistently rank content.

In this article, I’ll share with you my experience using the Frase ai writing tool, the key features, pricing, and some pros and cons.

By the end of this Frase review, you’ll have a good idea of whether this tool is right for you.

Let’s get started.

What is Frase.io

Frase is a powerful tool that helps you conduct research, create outlines, write and optimize the articles to achieve high-quality SEO content. With the optional pro add on, you can also use their proprietary ai technology to assist you in your writing and content generation through native Frase and community driven ai templates.

Who is Frase for?

Frase is perfect for anyone who does content marketing as part of their daily workflow.

Freelancer writers can use Frase to quickly create relevant content in minutes. Generate question ideas during the Research step to ensure common questions are answered.

Agencies can write detailed content briefs to share with their clients and other writers. As a business owner, they would purchase this as part of their in house marketing software stack.

Content Strategists can use Frase to create SEO friendly content by carefully analyzing search intent in the Research step and cover any topic gaps through competitor analysis.

What Sets Frase Apart From Competitors

There are three main features that set Frase apart.

First, their Research feature is incredibly in depth and well laid out to quickly write briefs and outlines. This feature alone should save you 2 hours or more per article.

Secondly, their ai has now migrated over to OpenAI’s Davinci model which puts them on par with all of the other leading ai writers on the market. Here’s what Frase’s co-founder posted in the Frase community group.

Frase is currently using OpenAl’s Davinci as the default model, the so-called “GPT 3.5” (i.e. their
latest model). We are still evaluating performance but SO far the response has been overwhelmingly
positive

-Tommy Rgo in the Frase Facebook Community

Finally, the community managed ai templates are off the charts. The output I’ve been able to achieve is truly unbelievable. I haven’t tried every single ai writer on the market, but from my research, there are no other ai writers that allow community members to modify the ai’s input via templates.

Why I’m Qualified to Review Frase

Frase is currently the only software I use to assist me in writing these blog posts. I also use SEO tools like ahrefs, but all of my content is created inside the Frase editor.

I research, outline, write and optimize everything in Frase and then assemble it in WordPress before I publish it.

This doesn’t mean I won’t recommend other software in the future (I’m looking to try Jasper soon), but I’ve watched this software grow from the beginning and have been impressed with the progress.

I also have the pro add on for the unlimited ai writer because the community templates (discussed later) are so incredibly powerful and unique and I just don’t have a need to add another tool to my stack at this time.

With that, let’s get started with a quick dashboard overview of Frase.

Dashboard Overview

Frase has a simple dashboard setup with the primary options on the left. You’ll spend the majority of your time in the Documents part of this app as this is where the research, outlines, writing, and optimization happens.

In the workspace area on the right, you can sort by User and Status which you can toggle between Researching, Writing, Optimizing, Completed and Published. 

From here, start by selecting the New Document option to go to the content brief flow.

Writing Content Briefs in Minutes

Whether you’re writing for yourself or outsourcing the content creation process, writing a content brief is going to be the most efficient way to get a writer up to speed on what your content goals are for a particular article.

Frase wastes no time in getting you results. In the New Document flow, you’ll be presented with two main workflows: Create new content or Optimize existing content.

Create New Content

When creating new content it’s important to set the country correctly because search results will vary from country to country. You also have the option at this point to add it to a specific folder to group your articles together.

Target Search Query is where you type in the target keywords that you want to rank for and not necessarily the parent keyword. For example, use “how to affiliate market on Pinterest” instead of “affiliate marketing Pinterest”

For the rest of the review, I will use this Pinterest example using the new content workflow.

Optimize Existing Content

I use this option in two different ways. To optimize my already published articles or to look at the topic optimization score for a client’s article to suggest edits to freshen up the content.

If you select this option you will need to supply the Import URL so Frase will know where to pull the content from to complete the analysis.

Research Feature

This is where the massive time savings begins. After selecting a new document and supplying the query (I’ll use “how to affiliate market on Pinterest” for this demo), Frase prompts you to start your research.

Pressing Let’s Go will start pulling and summarizing search result topics via their ai. Headers along with other topical information are presented in a nice ordered list for the top 20 search results on Google for the query you entered.

At this point, you know the average word count is 3.6k with 8 headers and 7 links within the article. This helps you get a rough idea of how long and in depth your article should be if you want to have a chance at ranking.

Headings Dropdown in Research

If you change the dropdown to Headings, you can view all the heading tags for the top 20 results. This turns out to be a really long list of headers, but Frase takes care of this in another feature called Outline which is covered further down in this article.

Questions Dropdown in Research

To round out your articles you typically want to include a frequently asked question heading in your articles. This is where the Questions option comes in.

In this dropdown you have the option to select a question and answer source like People Also Ask (from Google), Quora, the popular question answer site, and Reddit.

In the example below, someone on Quora is asking if you can “use affiliate marketing without a blog.” This is a very specific question around this topic that I would not have come up with on my own unless I was deep in that niche.

This is just one example of the power with which Frase can surface very nuanced topics to make sure you are covering all your content bases.

Automate Content Brief

Automated content briefs are useful for quickly jotting down the summary of the ai research and beginning to piece together your outline.

Planning ahead of time, helps your article stay on track without writing a bunch of fluff that no one is searching about.

I recommend using this alongside a manual outline which I discuss in the next section. 

To create an automated brief follow these steps:

1. Note that your automated brief will generate in the Content Brief tab regardless of where you initiate this action. So be sure to write your content in the My Content tab to ensure you don’t overwrite content on accident with the ai.
2. Next, select SERP Overview in the dropdown
3. Check the brief options you want to include in the auto generation. I like to include questions and statistics sometimes depending on the article’s goal.
4. Click the Insert Brief into Editor button
5. Your automated brief will end up in the workspace on the left.

From this point, you can either create a new tab in the editor to add your manual outline or continue to manually add it at the bottom of the Content Brief tab.

Outline Feature

The Outline builder feature presents you with a horizontal view of all of the top 20 headers. From the Explore tab, you simply click to add the header from the left workspace into your content outline on the right.

Run out of content ideas? Keep scrolling right to see more headers to get more ideas. If you see a trend in the topics, it is likely a topic that is important to cover in your blog article.

Below I have selected my main H1, followed by my main H2 topic, and a mix of H3s from the Shopify article and the article beside it.

I would then continue to the right and add any relevant headers to fill in my content gaps and complete my outline.

What would normally take several hours to piece together between multiple tools can now be achieved in minutes.

Another handy feature in Frase is the ability to easily drag and drop headers up and down to reorder them into a more logical order for your article.

You can even drag them left and right to promote or demote your headers between H1-H5 (see below)

Once you save your outline it’s time to decide if you want the AI to assist in the outline summary.

Ai Topic Summary

In the Summarize tab, click the Summarize button to have the ai insert the key topics from its research results. 

This step can take several minutes if you have a lot of headers, but Frase gives you an estimated time frame.

Once your summary is ready, your editor will show you the results of the key topics summarized from all the top results. You can simply paste this summary into your editor and start writing content or you can take this one step further and have the ai write the entire first draft.

Writing an Automated Draft with Ai

Before automating drafts on multiple articles, try it on one article to see if you like the results. Using this feature will use ai writing credits.

Whether you are on the Solo, Basic, or Team premium tier, 20,000 characters per month are included.

Now back to this amazing feature. If you want to basically eliminate writer’s block forever, follow these steps:

1. Reminder again, your ai generated draft will show up in the Content Brief tab.
2. Review the topics that are summarized to make sure you are happy with them. You can always remove individual sections
3. If you don’t like any of them you can click re-summarize to get a whole new batch of topic summaries.
4. If you’re happy with the topics, click Write Draft and watch the magic start.

Within a few minutes, you will have an entire draft written by the ai based on the topic keywords and your article title. Of course, this works better with some topics compared to others.

Typically I find that articles that have a lot of content online will have better ai written results.

I do want to stress however, you should never publish the article that the ai wrote without editing it for several reasons.

First, the ai does not fact check any of your content. While it could potentially pull the correct stat, that would be the exception to the rule.

Also, using ai generated content is simply against Google’s terms of service, so while Google’s algorithms are getting smarter at detecting this, you can bet they will release an update in the future that could potentially wipe out all your hard work if you just published ai drafts verbatim.

Instead, use it to aid you in your introductions, conclusions, and transitions from topic to topic.

For example, here is some ai content for the Pinterest demo:

Affiliates are people who earn commissions by promoting products and services online. This includes affiliate marketing, where you recommend products and services to others. In return, you receive a commission when someone buys something because of your recommendation.

I would rewrite this to pass any checks by Google in the future and to know for certain it is 100% unique, plagiarism-free content:

Affiliate marketers are content creators who earn commissions by promoting other people’s products and services. By essentially referring web traffic from the content, if the reader ends up purchasing due to the recommendation, the affiliate would receive a commission.

(my rewritten version)

To come up with that type of topic flow might be easy for a subject matter expert, but what if you were someone that didn’t even work in marketing? This type of paragraph would be difficult to write without ai assistance or a lot of revisions.

Content Writing

Now that you have researched your topic, and filled in any gaps with a proper outline, it’s time to fire up the content editor and start writing content.

The vast majority of the time spent creating long-form content is during this phase where you will use the ai to assist you in your writing which we will cover next.

AI Write Feature

Next, we have the ai writing feature. If you want the ai to generate meaningful long-form content, you have to learn how to use this feature properly.

This feature has all of the Frase team’s ai templates and community generated templates combined.

My main gripe with this feature is how it is currently laid out. You have to know who in the community creates good templates and basically memorize a few acronyms which are very counterintuitive and not user friendly if you are brand new to Frase.

What is an Ai Template

The ai templates in Frase were created to help steer the ai to a specific outcome based on the input. As such, each template will have an objective.

For example, the template below by community member Matt Zimmerman has one specific objective which is to create paragraphs from some background input. 

The use case could be to rewrite existing paragraphs from a real blog. To continue with our Pinterest demo, I’ve added a new H2 and pasted actual content in the editor. After selecting the Paragraphs from Background template, it is now asking for input.

The templates are now smarter and will add your Title and Heading fields for you (usually the H2 above your cursor).

That leaves you to populate the Background Knowledge field. In this example, I’ve pasted the real article content right into that field and clicked Run Tool.

Here’s what the ai generated:

Pinterest is a great platform for passive income because it’s a place where people spend lots of time browsing through pictures. People are constantly searching for new ideas, products, and services. As a result, there’s a lot of opportunity to drive traffic to your website.

Pins have a longer shelf life than tweets, so if you’ve already created a pin that’s relevant to a certain topic, it can still generate traffic for you. Plus, since Pinterest is a visual search engine, it makes sense that people would click on a picture before reading any accompanying text.

So if you want to make money on Pinterest, you need to optimize your pins. Start by choosing a niche that interests you. Then find out what terms people are typing into Google to find related topics. Once you know those keywords, write a description that answers questions people might ask themselves when they land on your pin.

Then, once you’ve written a compelling description, add a link to your website. Finally, include a call to action button that encourages people to visit your site.

Once you’ve done all that, you’re ready to start earning money on Pinterest.

I would rate that output 8 out of 10. It’s pretty good and even suggested using google to find what people were searching which wasn’t even part of the input, so the ai came up with that concept on its own.

From here I would rewrite that entire section in my own words while fact checking what it said to ensure 100% fresh content.

From this point, it’s about trying and learning which template works for your writing style. You will find that the quality of your input is what really drives the output.

If you find you’re getting consistently bad results, it’s likely because the input is not good enough or the ai is not trained sufficiently on the topic and you’ll have to come up with content the old fashion way.

Optimize Feature

By this point, you should have a well researched article, with an outline and a decent amount of content written either with or without the ai.

The final part of this feature review is the Optimize feature.

Content optimization is where Frase scores the topics and headers that are written about across all the top 20 search results. Then it summarizes it in a very simple paint by numbers counter.

Going back to our Pinterest demo article, I am currently at a score of 26%. The average is the line to the right which is around the 47% range and each Long Tail keyword has a color coded status

• Green = the keyword was mentioned a sufficient number of times – don’t add or subtract this keyword in any more content.
• Yellow = the keyword is almost mentioned enough times – mention this keyword more in the content.
• Red = the keyword is mentioned too many times – remove some content with this keyword in it or reword content.
• White = the keyword was not mentioned at all – add more content with this keyword in it.

Based on the data Frase is giving me, I need to add quite a few more keywords (add more headers and written content to make that board go green) before I can meet or exceed the Frase content optimization score target of 46%.

I would then continue to write and check this real-time score until I am happy with the score.

So I know what you’re thinking, this is really gimmicky. At least that’s how I felt when I started using this feature. It feels so unnatural because it is.

And this isn’t just a Frase feature, the other leading ai-powered tool Surfer SEO uses the exact same feature.

But this is currently one of the factors of how Google determines the topical relevance of content so until Frase finds a better way, we just follow the tool because it has proven to work once your content ranks (you’ll only know if you try to edit your content for a better score yourself).

In order to prevent the scoring mechanism from affecting my writing, I only optimize when I’ve finished an entire draft of an article. If you use it from the beginning, it’s too easy to get swayed by trying to achieve a 100% rating and taking the focus off of what matters – quality content. Don’t fall into that trap!

Publish Your Article

By the end of the optimization phase, give it one last review or even better, take a break and come back a day later if you have the time and just give it one last edit, then publish your well researched and optimized content.

Congrats, you just published your first ai article!

Pricing

I’ll start with the bad news. Frase does not have a free tier option, however, they do have a 5-day free trial for $1.

Here’s the breakdown for the pricing plans.

The most budget option is $14.99 per month followed by the Basic plan at $44.99 all the way to $114.99 for the Team plan. With each tier, you are increasing the number of articles you can write and optimize per month.

Each tier comes with 20,000 characters of ai generated content so that is consistent across the board.

I would recommend if you’re just starting out, just start with Solo as you can always upgrade later.

If you have 1-2 clients, then consider the Basic (middle) tier as 4 documents per month likely won’t be enough.

The main challenge I see with the Basic tier is that there is still only a single-user license. If this was increased to 2, perhaps it would be more attractive. You could use one seat for yourself and one for your outsourced writer for example.

For unlimited document credits, you’ll need to upgrade to the Team level.

The $35 Pro Add On

Whether you select the Solo, Basic or Team plan, you still have the option to add the Pro Add on which gives you:

• Keyword Search Volume – monthly search volume in the Outline step
• SERP data – domain authority and backlink data to search results
• Frase Ai Writer – unlimited access to the ai writer. The primary reason to upgrade

If you are signing up for the trial, pay particular attention to whether you are adding this $35 add on. You may receive a larger bill than expected if you are just clicking through quickly.

At $1 for 5 days of taking the tool for a spin, there’s really no reason why you shouldn’t try it out.

Tutorials

These are the resources I recommend to quickly get started.

• Frase runs weekly webinars that anyone can join. Register Here.
• Frase internal wiki – Every feature is broken down in the wiki curated by the Frase team. May topics include video tutorials as well as written ones.
• Youtube – Unlimited supply of tutorials, but I would recommend starting with Ranking Tactics channel to quickly get going and stay up to date.
• If you are still stuck and just want to chat with me 1:1, you can contact me here. I’ll be happy to help get you started.

Integrations

Frase currently has 3 free integration options if you don’t want to use the Frase content editor.

Google Search Console

If you have a google search console integration set up, you can take advantage of their content analytics feature.

With this feature, it will pull in your search console data to show you where your content opportunities lie.

You can also view your other metrics like impressions, clicks, top-ranking pages, and average CTR.

Google Docs Add On

If you work exclusively in google docs, you can take the Frase content score feature from into your Google doc now.

Just download the add on from the marketplace and you will be able to see your topic score and competitor content in the right sidebar. Here is the Frase help guide to install this integration.

WordPress Plugin

Similar to the Google Doc plugin, if you work mostly in WordPress during your content creation phase, you can install this optional plugin for free.

It also gives you the topic score right inside your WordPress Gutenberg editor.

In order to get the topic score to show up in WordPress you’ll need to generate a share document from Frase.

Wrapping it Up

When it comes to content optimization tools, it’s hard not to consider Frase as one of the top content writers on the market.

From complete writer’s block to SEO-optimized content and unlimited documents, there is so much packed into this tool suite which starts at $14.99.

Many of you will be coming from the Jasper + Surfer SEO combo which starts at $24 per month for Jasper and $49 for the basic plan for Surfer (unless you stay on the free tier).

Either way, Frase is going to be a less expensive option to get you started for a similar feature set of content optimization + ai writing credits.

Frequently Asked Questions

How much of this article was generated by ai?

The final written content was 0% generated by ai, but the outline and draft were 100% written by ai, and then manually rewritten to be 100% unique which is how I recommend you use any ai tool, not just Frase.

Is there a Roadmap for Frase?

Frase has recently conducted a survey to consider its 2023 roadmap. Consider joining the Frase Facebook community to stay up to date. This is where the roadmap will likely be posted.

What integrations are available?

Yes, there are currently 3 integrations available. Google Search Console, WordPress Plugin, and a Google Doc plugin. These are covered in this section above.

What languages are supported?

The Frase content editor supports English, French, Spanish, German, Danish, Dutch, Portuguese, Italian, Norwegian, Finnish, Swedish, Romanian, Polish, Russian, Hindi, Indonesian, Malay, Thai, Japanese and Korean

Language toggles are discussed in this help article. 

Is the Pro add on worth it?

Compared to other similar products I would recommend adding the Pro Add on if you are not an experienced writer and need to generate 4 articles a month.

Note, even without the $35 Pro Add on, all paid tiers already come with 20,000 ai generated characters. The best way to measure if you need the Pro Add on is to see how many articles you can get through before upgrading.

If you are using Jasper ai and the Surfer SEO content optimizer, this would set you back a minimum of $73.

Compared to the Solo plan from Frase which is $14.99 + $35 for the unlimited Pro ai add on, this would still be less expensive and you would only need to use a single tool versus two separate tools to achieve similar results.

What is the Frase Answer Engine?

Frase’s Answer assistant is their own version of the generic chatbot. By initiating an initial crawl of your website data, Frase Answer will create an intelligent knowledge base.

Then when you train and turn on the chatbot, it will learn to serve up the optimal information to your website visitors. 

When it fails to produce a good result, you can configure it to defer to live chat based on specific scenarios.

To learn more about Frase Answer, check out this help article.

Where can I check my Ai Credits?

To check how many characters your account has produced per month, go to Account > Subscription page.

Does Frase have a LTD (lifetime deal)?

Currently, there are no plans to have a lifetime deal for Frase. A while back, they were listed on App Sumo (and possibly other LTD deal sites) so it is possible to buy in the secondary market if anyone is willing to sell it.

Does Frase have an affiliate program?

Yes, Frase has an affiliate program where they pay 30% on recurring subscription revenue among other perks like 60-day cookies and swipe files. Find out more about their affiliate program here.

When they run into problems they want to know help is just one step away and that’s where a chat box can help.

For years, chatbots have been all the hype and while they are certainly useful, nothing can beat the human touch.

Enter Crisp Chat.

Why Do I Need Live Chat?

Depending on where the customer is in their buying journey, adding live chat can be the difference between a conversion on your eCommerce site or purchase of your course.

Often when I’m shopping online, I have only one or two major objections, and had I been able to get those questions answered (ideally right on the spot), I would have purchased.

Without questions answered, customers go away and get distracted and often don’t come back.

Another popular use case for chat is simply to engage with prospects. If you’re just starting out and need to validate an idea or product, who better to speak with than directly to people who visit your website or landing page?

Getting Started on Crisp Chat

Crisp makes it incredibly easy to get signed up. Four screens and you’re done in under 5 minutes.

Start by going to https://crisp.chat

On their landing page, click either Get Started or Use Crisp for Free

In Step 1, fill in your basic details

In Step 2, they ask for the Company Name and Website.

Both of these fields can be updated after you get through the setup in case you need to update it in the future.

If you provide the website URL it will pull the image from the website, but this can be overridden in the next step.

Click Continue to proceed.

For Step 3, you can customize the look of your chat box to your brand colors.

On the final page, you’ll see all of the documentation to implement the chat widget code. If you don’t have a code snippet manager, I would recommend downloading one at this time.

After you install the code on your website, open another tab and test using your website. You should see the chat bubble in the corner.

From this point on you can start chatting with your customers and the Crisp dashboard will track all of your chat threads.