JobQuest Assistant: How I Built a Smart Companion for Stress-Free Job Hunting With Coze | HackerNoon

Job hunting is so unbelievably draining.

Yes! That’s not a cliché — It’s something I experienced.

Moreover, a survey by ResumeGenius revealed that around 70% of job seekers feel more stressed about job hunting.

70% of job seekers feel more stressed about job hunting.

Although there have been a lot of articles on how to make job hunting less stressful (Organize yourself, see it as a purpose, take breaks, create a portfolio, be selective, tailor your application, and so on), the truth is that this advice doesn’t address the fundamental problem.

The real problem is that job hunting is complex, boring, and repetitive, which makes it overwhelming. So, the right solution, I believe, is breaking down the application process to make it less mentally draining and just making sure job seekers only focus on the necessary tasks.

With the recent trends in artificial intelligence (AI), there are possibilities for building chatbots to solve these problems. And that’s how JobQuest came to be.

JobQuest is an AI chatbot that can streamline your job-hunting processes. It scrapes job boards for job openings that fit your preference, offers tips to improve your resume for different roles, crafts personalized cover letters, and helps you prepare for interviews.

In this article, I will walk you through the chatbot’s development phase. But before that, let’s check out how the chatbot works.

How Does JobQuest Work?

Simply put, JobQuest is your job-hunting companion. It’s a chatbot that helps streamline job hunting by doing all the repetitive and boring tasks. This lets you focus on the most important processes — developing intellectual and personalized documents to attract the employer’s attention.

The job market is fragmented — different companies have ways they call positions they’re hiring for, even though they’re looking for someone to do the same work.

For example, a company can be looking for a content writer while another is looking for a copywriter. Both companies have very similar job descriptions with different names. For a job seeker to know if this is appropriate for them, they need to go through each job description — which is time-consuming and overwhelming.

JobQuest lets you provide your job preferences. Based on this information, it searches different job platforms for recent jobs related to your preferences.

Here is how JobQuest Works:

The AI chatbot contains four features:

• Job Search
• Resume Updating and Personalization
• Cover letter Generation
• Interview Practice

A typical job hunter would love to check job boards for job openings in different companies. So, instead of manually scanning different job boards and filtering out jobs based on your preference, JobQuest handles the task and provides you with job recommendations.

When you send a prompt telling the chatbot that you’re looking for a job, the chatbot asks you to submit your preferences (industry, location, job types, position, and your resume (not compulsory).

Once you submit your preferences, the chatbot searches and returns a list of jobs with links to apply or learn more. (To avoid being overwhelmed, I ensured that the chatbots returned less than 15 jobs per session. If you need a list of more jobs, you can prompt the chatbot).

Once you find jobs you’re interested in, your next move, as a job seeker, is to submit your application (your resume and cover letter). Let’s check out a few facts about submitting job applications:

• You’re not the only person applying for that job.
• You’re probably just one of the over 500 people applying. The HR will not have the time and mental ability to read each job application. So, they’ll deploy Application Tracking Software (ATS) (a software that scans and filters job applications based on certain criteria like keywords used), to help them select the best candidate.
• You need to submit a personalized cover letter and resume to increase your odds of getting an interview.

Based on the points above, I believe no job seeker can keep up without getting drained and exhausted. Instead of personally writing every unique cover letter and updating your resume for each job you applied for, an AI chatbot can help with the task.

So, if you need help improving your resume and writing a cover letter, all you need to do is prompt the chatbot. The chatbot will ask you to upload your resume and paste the job description (JD).

The chatbot then generates a personalized cover letter, which is optimized with keywords from the job descriptions.

The chatbot also provides insights, recommendations, and things to add, remove, or update in your resume about the job description.

Apart from reducing the stress you encounter as a job hunter, a personalized cover letter and resume increase your chance of hearing back from a company.

You can also use JobQuest to practice interviews and get tips on how to get a call back from the hiring manager. JobQuest can help you ace your interview by helping you practice likely questions and offering feedback for performance improvement. This chatbot will simulate the interviewing stage, taking the position of an interviewer.

Once you send your prompt requesting an interview, the chatbot asks you to provide the necessary information related to the interview process. Information like duration of the interview, industry, type of job you’re applying for, kind of interview to focus on (behavioral, situational, or technical), and also the job description.

Once this information is provided, the chatbot asks you questions, waits for your answer, and provides feedback and examples. At the end of the interview section (when the duration has elapsed), the chatbot gives feedback and provides tips on improving yourself to increase your chances of success.

JobQuest also provides you with online information on how to ace your interview. Apart from the chatbot’s features, this chatbot also updates you with daily alerts to the latest jobs.

Watch this demo video on how this chatbot works.

You can also try the chatbot out:

How Did I Build This AI Chatbot?

In this section, I’ll walk you through how I built this chatbot in less than three days.

Yes! That wasn’t a typographical error. I built this job seeker’s companion in less than 72 hours — thanks to Coze, the no-code AI Chatbot development platform.

What’s Coze?

Coze is an AI Chatbot creation platform that enables you to bring your chatbot idea to life without writing a single line of code. You build your chatbot by writing prompts (instructions to guide how the chatbot works). Coze uses Natural Language Processing (NLP) to communicate.

Natural language processing (NLP) enables computers to understand and communicate with humans in human language. Computers normally communicate in zeros and ones but with the NLP algorithm, computers can read instructions provided by humans and process the information.

Process of Building The Chatbot

With Coze, I idealize, implement, and publish the chatbot without writing a single line of code. Follow along as I walk you through my process of building this Chatbot.

Setting Up The Prerequisites

This seems to be the most significant part of my chatbot development process. Why? It helps me develop a clear thought process, visualize the problem, and provide appropriate solutions.

Here is a checklist of things that helped me visualize the chatbot.

• Understanding the Problem I Want to Solve

I want to build a chatbot that relieves job seekers of their stress during job hunting. I had to analyze each checkpoint that a job seeker needs to go through for every application.

A job seeker searches for jobs, applies for those jobs with personalized cover letters and resumes, and prepares for interviews.

Having a clear understanding of what I want to build helps me:

• Make decisions about a multi-agent model or a single-agent model. Based on the checkpoint above, I’ll be building a multi-agent chatbot.
• Decide the types of plugins I need for the project. For this project, I need to search for jobs online. So, I need a plugin to connect with job boards (or a job aggregation site) to retrieve job information.  I also need a document reader and parser to read and parse the resume that users upload. For this project, I was lucky to find a plugin in the Coze plugin store.

Once I make these prerequisites available, it’s time to get my hands dirty. Let’s get working!

Building the Backend

Now that I have the prerequisite, let’s build a chatbot.

Setting Up the Workspace

The first step is setting up your workspace.

To set up your workspace:

• Visit Coze.com
• Click the ‘+’ at the top left corner of the home page, and a page will bring up the chatbot creation wizard, which provides you with two options (the standard and the AI-generated option) to create your chatbot.

The standard option lets you set up the agent name and functions, and choose the workspace, and the chatbot icons manually.

The AI-generation option requires you to input a rough idea of what you want and the name, description, icon, and prompt are generated.

Assuming I want to develop a JobQuest chatbot. I’ll just input this summary:

“a chatbot that streamlines the job hunting process. It helps with searching for jobs, reviewing resumes, generating cover letters, and helping practice the interview process.”

The AI will generate the prompt, skills, icon, name, description, and prompt.

However, as someone with a clear idea of what I want to develop, I prefer to manually input the agent name, function, and icon, which I did.

Building The Features of My Chatbot

When you create a chatbot, you’ll be taken to the single-agent mode workspace by default. You can switch to the multi-agent mode by clicking the mode-switching icon on the top left side of the screen.

In multi-agent mode, you can tweak your chatbot on both the global level (which affects the overall chatbot’s behavior) and on the local level (changing how each node in the multi-agent chatbot works).

Global Features

The global feature settings are located at the far left of your screen. It’s called the “Arrangement” section. Here is where you generate the Persona and Prompt, add the skills, tweak the memory settings, tweak the chat experience, and add roles to the chatbot.

For now, I’ll focus on the Persona and Prompt section of the global settings because it aligns with the backend settings. We’ll talk about the others in the latter section of the article.

Adding the Persona and Prompt to my Chatbot

This section lets me set my chatbot’s character, what it can/can’t do, the features and skills, the workflow, the content, and constraints.

One thing I want you to know is that building a chatbot on a no-code platform like Coze requires you to be proficient in prompt engineering.

For your chatbot to be effective and accurate, you need to know how to structure your instructions in ways that the generative AI agent will understand.

However, if you don’t know prompt engineering, don’t worry, Coze has provided a way to get around prompting.

Before learning how to get around prompting, let’s learn a few tips about good prompting.

1. Keep your instruction concise. Don’t ramble. The more unclear your instructions are, the more difficult it becomes for the chatbot to understand and return a correct answer.

2. Use natural and basic language. Generative AI can process natural languages—so use natural language. Write grammatically correct prompts and punctuate your instructions correctly.

3. Assign roles and specify your output. See the generative AI as a student, treat it like one. Assign roles, and instruct it on how you want your outputs to be.

4. Apply good structure. Be detailed. Include the tone, length, structure, style, and how you want your chatbot to respond to user queries. Experiment and iterate.

5. Give instructions and try new things with your chatbot. By this, you’ll learn how to build efficient, accurate, and useful chatbots.

Let’s learn Coze’s way around writing prompts from scratch.

On the single-agent mode workspace, select the optimize button at the top right of the arrangement session.  A box will pop up where you have three options. You can pick between:

• Auto-optimizing directly
• Generating prompts based on the debugging results.
• Writing out the chatbot idea.

When you pick any of these options, the AI generates a prompt and persona for your chatbot. And you can edit it to fit your idea.

You can also choose between different Persona and Prompt templates.

After generating the draft, all you need to do is to switch to multi-agent mode.

For this project, my Persona and Prompt section consists of the role section, where you let the chatbot assign a role.

You’re …

Role

You are JobQuest Assistant, a job search companion dedicated to providing personalized job recommendations, expert tips on resumes and 
cover letters, and tailored interview preparation to help users navigate their job search and secure their dream positions.

Features, where I talk about the different skills. They are sometimes called skills if you’re not designing a complex chatbot. In my chatbot, I have three features — the job search feature, the cover letter and resume tips feature, and the interview preparation assistant features.

And these features have sub-skills.

Check this out:

Features

### Feature 1: Streamline and Personalized Job Search and Recommendation

- Fetch job opportunities based on the user's preferences (desired role, industry, location, experience level and so on)

- Display results in an organized and categorized format with quick apply links.


### Feature 2: Cover Letter and Resume Tips

**Skill 1: Resume Review and Tips**

- Identify strengths, weaknesses, and formatting errors in uploaded resume.

- Suggest industry-specific keywords for better ATS compatibility.

- Give recommedations for improving and personalizing resuming based on job description.

**Skill 2: Cover Letter Writing**

- Provide a guided wizard to draft personalized cover letters.

- Suggest phrases and formats based on the job role and company.

- Generate a coverletter template based on the job description.


### Feature 3: Interview Preparation Assistance

**Skill 1: Mock Interviews**

- Simulate interview scenarios based on the user role. (you ask a question like an interviewer and the user answer the question like the interviewee).

- Provide behavioral, technical, and situational questions.

- Provide an overall overview after the interviewing section is over.

**Skill 2: Preparation Resources**

- Provide resources like FAQs, STAR method examples, and confidence-building tips.

- Example:

🎤 Question: "Tell me about a time you overcame a significant challenge."

💡 Tip: Focus on STAR (Situation, Task, Action, Result).

Workflow is where I gave instructions on how my Chatbot should execute processes. For a complex chatbot like what we’re working on, you need to start with:

• Collect preferences to search for jobs
• Improve their resume and cover letters Generation
• Create an interactive interview session to help the job seeker improve their question-answering skills.

Content is where I give instructions on how answers should be. Here you tell the chatbot the right tone and how you want it to be. For my Chatbot, I don’t want it to be harsh on the job seeker, it should be a friend, a companion that relieves them of their stress.

I also wrote prompts for the chatbot on how I wanted it to structure answers to user questions.

Check this out:

Content

**Example Prompts:**

- **Job Search:** "Let’s find your dream job! What role and location are you targeting?"

- **Cover Letter Tips:** "Here’s a professional tone for your cover letter. How would you like to personalize it further?"

- **Mock Interview:** "You’re applying for a project manager role. How would you handle conflicting priorities in a project?"

The Constraints serve as a guardrail to make the chatbot focus on the problem it’s solving.

Check this out:

Constraint

- Only focus on job hunting topics

- Only offer feedback at the end of interviewing session

For a multi-agent chatbot, this section allows you to give an overview of what you want the chatbot to be. This session is like designing the wireframe when designing a website.

Just because it’s an overview doesn’t mean it should be vague and confusing. Like what you saw above, this is the foundation of building an efficient chatbot. Find the right balance that would contribute to the overall efficiency and accuracy of your chatbot.

Once we create the chatbot’s prompt, it’s time to focus on building the local features. Let’s do this!

Local Features

Unlike the global feature, the Local features affect the efficiency of each chatbot’s node. I discovered while working on this project that the efficiency of your chatbot depends on the independence and interdependence of each node.

So, let’s make our chatbot both independent and interdependent.

The ‘Develop’ section is where we tweak the local features of our chatbot. The drag-and-drop feature makes it easier to build your chatbot. Once you switch to a multi-agent mode, there is a start button in green connected to a node as your default session.

You can add more nodes to the workspace by either double-clicking any part of the developed workspace and adding the agent or clicking the add nodes button (in blue) at the bottom of the section.

Here, you can create as many nodes as you want

For this project, I’ll be creating four nodes, which are listed below:

• Personalized job recommendations
• Resume tips
• Cover letter generator
• Interview preparation assistant

Each node contains the scenario, agent prompt, skill, and auto-suggestion section.

The scenario. This is where I add the functions and supported scenarios that apply to the chatbot. This section helps other chatbot nodes know when and to which node to assign tasks they can’t handle.

Here is how my scenario looks for the ‘Personalized Job Recommendations’ node:

Helps you streamline the job search process. 
It helps user solve the problems related to finding relevant job opportunities, 
ensuring the desired role align with their preferences.

The agent’s prompt section is where you provide the operational logic and the steps for handling problems.

Unlike the Persona and prompt section of the global settings, I was more detailed here. I added the role of the node, the operational logic (gives instructions on how I want my nodes to operate), how I want the chatbot to format answers, how it would handle errors in case of any, and the constraints.

Here is an example:

Role

You're a job hunter. your task is to help users search and retrieve jobs that align to their preferences


### **Operational Logic**


#### **Prompt User for their Preferences**

- Ask for the role, industry, location, and experience level.

Example: *"What type of job are you looking for? Please specify the role, industry, and preferred location."*


#### **Save Searches & Alerts**

- Ask the user if they'd like to save the current search for future alerts.

Example: *"Would you like to save this search for future updates?"*


...


## Output Format Example

========================== 
🎯 Top Recommendation: <Job Title> at <Company Name>

 🏢 Location: <City, State>

 💼 Job Type: Full-Time/Part-Time/Remote

 💰 Salary Range: <Salary Information>

 📅 Posted on: <Date>


**Note:**

Provide an action button for each job listed to apply or view more details.


## **Problem Handling Steps**

#### **Scenario 1: Missing or Incomplete User Input**

- **Problem:** The user skips providing some preferences or does not upload a resume.

- **Solution:**

Prompt politely for the missing information.

Example: *"I noticed you haven’t uploaded a resume. Would you like to proceed without it or upload one for tailored recommendations?"*

Proceed with default options if the user chooses to skip.

### **Scenario 2: No Matching Job Listings Found**

- **Problem:** The `Job_search()` plugin returns no results.

- **Solution:**

Apologize and offer alternative suggestions.

Example: *"I couldn’t find any jobs matching your preferences. Would you like to expand your search criteria or explore similar roles?"*

Suggest nearby industries, roles, or locations.


---

## **Constraints**

1. Only job-search related tasks are handled by this agent.

2. User data, including preferences, must be securely stored and not shared without consent.

...

As you can see, the prompts above are well more detailed

The skills section is where you add the plugins, knowledge, or workflow to enhance the node’s capability. With skills, your node/agent will work like an individual on steroids.

To add skills,

• Click the ‘+,’ and follow the instructions to select the skills you like to add.

Toggling on ‘Auto-suggestion‘ means that after each answer, the chatbot returns three additional questions you can ask.

OTHER THINGS TO NOTE:

• Adding the Global Jump Condition

The global jump condition helps the node decide what to do in certain scenarios. Adding global jump conditions to each of my nodes increases their effectiveness. When a user needs help with another function, the jump condition helps them make the switch decisions faster.

The global jump condition also makes it easier for the chatbot to assign tasks to any of the nodes as fast as possible.

One thing I want you to know is that you don’t just connect any node. Your AI chatbot’s efficiency also depends on the relationship between each node. You don’t connect nodes that are unrelated together.

Connecting one node to another requires a deep understanding of how users think and the problem you help them solve.

For this project, a job seeker will most likely start by searching the web for jobs. So, I placed the ‘personalized job recommendation’ node as the base node, which is the primary checkpoint for every new user.

Once they receive a job list, the next stage of the workflow is sending out applications. Here, they’d need help with their resume and cover letter. This is why I connected the ‘Resume Tip‘ node to the ‘Personalized Job Recommendation’ Node.

The ‘Resume Tip‘ node is connected to the ‘Cover Letter Generator‘ node. Whenever a chatbot user asks to generate a cover letter, the node checks if their resume has been uploaded before, if not, it asks the user to upload their resume. I did this to avoid redundancy.

The ‘Interview Preparation Assistant’ node is different from the others. It’s linked to two different nodes — the ‘Personalized Job Recommendations‘ node and the ‘Resume Tip‘ node. Why?

To simulate a personalized interview section, you need the job description and the resume. So, instead of being only a sub-node to the ‘Resume Tip‘ node, I decided to also add it to the base node — the ‘Personalized Job Recommendations‘ node.

We’ve developed the most tedious part of the project, the backend processes. It’s time to improve the chatbot’s user experience.

Designing The Frontend

The Frontend is as important as the Backend. It helps humanize the chatbot and makes it easier to communicate with the chatbot. In this section, we’ll be tweaking some settings to improve the performance of our chatbot and help with user experience. Just as with the backend above, we also need to tweak both the local (to improve individual nodes) and the global settings (to work on the chatbot’s efficiency).

Without ado, let’s get started!

Tweaking the Local Settings

Here, we have to work on tweaking the model settings, the node switching settings, and “Which nodes to handle every round of conversation?”

The Model Settings

Here, I choose the Generative AI model appropriate for the chatbot, select the generation diversity, and set the input and output settings.

For the model settings,

• Click the three dots at the top right corner of the node
• Select the model settings.

The model settings let me choose from any of the available generative models. There are several versions of the OpenAI GPT models, the Anthropic Claude, and Google Gemini. The type of model you choose depends on your need for efficiency and cost.

For my project, I went for the OpenAI GPT-4o mini model because it works well for my project.

The best thing with a multi-agent model like this is that each setting only affects each node.

Generation Diversity focuses on tweaking the model generation diversity. You can choose between precise, balanced, creative, or even custom settings.

For this project, I left the model setting as-is.

The Node Switching Settings

This is where we tweak how the current node handles scenarios where it can’t answer users’ questions.

Here, you can select what to do when the node can’t solve the user’s problem and who is responsible for deciding to switch answering nodes— either an independent model from the current node or the model answering the questions.

When to decide to switch nodes. After user input, after the model answer, or both.

Overall, this setting helps build a smooth transition between each node in a multi-agent system.

Which Node Should The New Round of Conversation Be Sent To?

These settings ask to decide how every new round of conversation should be handled.

Should the last node that answers the previous question continue? Or should the start node answer the question every time? As a default setting, Every new conversation will be sent to the last node that answers the question, which is the best for this chatbot development project.

The Global Settings

The global settings encompass the overall functionalities of the AI Chatbot.

For the global settings, go back to the arrangement section on the left side of the screen. Here, you can tweak the memory, improve the chat experience, and add voice functionality.

Memory Functionality

The memory section consists:

• The variable lets you create a container to store user information.
• The database is an organized table storage that lets you store user data in a structured table format.
• Long-term memory works like the human brain. It helps the AI agent formulate a memory of the user based on the interaction. Activating this makes it easier to improve user experience. It remembers previous chats, stores important personalized information, and refines answers based on past chat experiences to provide personalized information.

For my project, I assigned variables to store user preferences (Desired Job role, location, and experience level).

I also toggle on the long-term memory to improve user experience and help the chatbot personalize chats for each user.

The Chat Experience

While all the global settings are important, this chat experience setting is more important. Why? It can directly impact user experience.

The opening questions, shortcuts, and background image can help set the modes and keep the conversation going as long as possible.

This is where you introduce your chatbot to first-time users and give them a list of questions they can ask.

For JobQuest, I make sure the questions are related to questions my users would likely ask. Even if it’s not what they want, it can help them set the mode for using the chatbot.

To create the introduction and opening questions,

• Click the ‘+’ at the top of the section, and the AI will generate some for you.
• Tweak and edit to fit your preference.

These are buttons above the search input. They make it easier for users to initiate preset questions.

To alleviate stress during the job search process, I decided to add shortcut buttons that users can easily click for their queries.

To create shortcuts,

• Click the ‘+’ button at the top of the shortcut section and fill in the information on the pop-up page.

Here is a resource on how to add a shortcut to your chatbot.

Add a related background image that increases the user experience.

Testing and Debugging

The testing and debugging process is as important as the rest of the chatbot development phase. Skipping this section is like developing an application and deploying it without testing if it works or not.

When developing a chatbot, the best practice is to continue to test and debug as you build. This helps discover errors early on in the project development phase.

The typical development process I followed for this chatbot is,

• Write the prompt
• Test if it works
• If it doesn’t work  as intended, edit the prompt
• If it works as intended, move on

I Repeat the process for all nodes.

After the overall development phase (both frontend and backend), I assume the position of a user and simulate how they think. This helped me write prompts to test how efficient my chatbot is and how effective it is at switching nodes.

Here is my checklist for the Testing and Debugging phase. I check if,

• nodes answer questions as intended
• the number of tokens consumed and answering speed
• the chatbot invoked plugins at the appropriate time
• nodes are switched when they should

The conventional understanding is to debug when there is a problem. However, I believe you shouldn’t only debug when errors occur; it’s also important to understand what’s happening in the backend.

Here is how I learn about what’s going on at the backend of my chatbot:

• Click the debug symbol (the wrench image at the bottom of the generated answer), and the debug detail will be shown.

The debug information provides you access  to a lot of information including:

• the RunTree( the process the chatbot took to answer your question)
• the node details (latency, start/end time, token used, etc.)
• the input (that’s the user input in JSON format)
• the output (the generated output in markdown format)

If you think the model consumes too a lot of tokens, you can learn where to optimize the process for efficiency.

All you have to do is go over to the Runtree section:

• Hover your cursor over the list
• And you’d be able to access the latency information and the token consumption of each invocation.

In the image above, for example, we can see that GPT4 mini consumes 3253 tokens and has a latency of 5034ms.

With this information, you can decide how to improve your chatbot’s efficiency and response speed.

Testing and Debugging isn’t a one-time process, it’s iterative. You continue to do this till you are satisfied with the result.

Publishing

To publish your chatbot, click the blue publish button at the top right corner of the screen.

Choose the platform where you want to publish your chatbot. You have the option to publish on various platforms, as an API (which can be accessed by other programs or applications), or a web SDK (to integrate this chatbot’s functionalities into web applications).

Wherever or however you want to deploy your chatbot, all you need to do is select as many platforms as you like and configure them to get them authorized.

For this chatbot, I’ll be deploying it on the Coze agent store, Cici AI, and Telegram.

The first two platforms (Coze Agent Store and Cici AI) have already been authorized, so I only need to select them. To deploy the chatbot to the Telegram platform, I need to configure it.

Learn how to configure Telegram for Coze Chatbot deployment.

Once you have completed the configuration, click the publish button and wait for your submission to be approved and published.

Future Upgrades

The job-hunting process is a very complex task. So, building a tool to relieve job seekers of stress is a way to go. Although I had a full-fledged idea of what problems I wanted the JobQuest chatbot to solve, the more I went through the development process, the more I uncovered problems that needed solving.

So, although I’ve published this chatbot for use, there are still more features I want to add as time goes by.

Here are some of the features I’ll look forward to adding in the not-so-distant future:

1. I’d love to integrate a spreadsheet application into my Chatbot. This will ensure that every job is added to the spreadsheet for better organization.
2. Integrating a task and calendar application for users to schedule interviews and get constant reminders about them.

Overall, shortly, I’ll be working to add features to organize job search processes.

Conclusion

In this article, I’ve detailed how the JobQuest Chatbot works, including its features and development journey using the Coze no-code platform. Whether you’re an experienced developer or a tech enthusiast, you’ll find insights to inspire your next project.