Introduction
Summary of this page
This page helps you understand what to start with and in what order when using the evaluation version.
- Recommended routes based on your evaluation goal 60 min / 90 min / using your own files / DC integration
- What you should do first and what you should avoid
- A checklist for organizing your evaluation results
By reading this page, you can start your evaluation without hesitation.
Overview
This page explains what to check first and in what order to try things for users who are starting to use the Agentiqs evaluation version.
Agentiqs is not just a tool for chatting with AI.
By combining AI agents, knowledge, workflows, file operations, and external tool integration, you can incorporate AI into tasks such as design, reviews, test design, meeting minutes creation, and inquiry handling.
On the other hand, right after obtaining the evaluation version, you may feel things such as the following.
- You do not know what to try first
- You do not know how to create AI agents or workflows
- You are not sure which sample data to use
- You do not know how to map it to your own work
- It is hard to understand how it differs from ordinary AI chat
- You want to quickly confirm the value of integration with Next Design or Lightning Review
This start guide explains how to proceed so that you can confirm the value of Agentiqs in a short time using the evaluation version.
What you can learn on this page
This page explains the following.
- What to check first in the evaluation version
- How to proceed with the evaluation
- Which samples to try first
- Recommended evaluation routes for 60 minutes, 90 minutes, and half a day
- How to replace the sample flow with your own business files
- How to check integration with Next Design and Lightning Review
- What points to focus on during evaluation
- Common stumbling points during evaluation
- Which pages to read next
We provide sample data that lets you quickly experience how
Agentiqs works.For details, see "Sample Data".
This evaluation guide shows you how to deepen your understanding of Agentiqs using sample data.
What we want you to check in the evaluation version
In the evaluation version, please check not only whether "the AI can answer," but also how Agentiqs can help your work.
In particular, we recommend checking the following four points.
| Checkpoint | Details |
|---|---|
| Value of AI agents | Whether you can use AI tailored to your business purpose |
| Value of knowledge utilization | Whether it can answer by referring to business documents such as manuals and specifications |
| Value of workflows | Whether multiple processes can be combined and executed as routine work |
| Value of tool integration | Whether it can integrate with business tools such as Word, Excel, Next Design, and Lightning Review |
When evaluating Agentiqs, it is important to look not only at the intelligence of the AI itself, but also at whether AI can be incorporated into business processes.
Goals of the evaluation version
In the evaluation version, aim for the following state.
Goals for the first 60 minutes
- You can launch Agentiqs
- You can chat with AI
- You can run one sample data set
- You can imagine what kinds of work support Agentiqs can provide
Goals for the first 90 minutes
- You can try 2 to 3 representative samples
- You can understand the differences among knowledge utilization, review support, and deliverable generation
- You can identify use cases close to your own work
- You can explain how it differs from ordinary AI chat
Goals for a half-day evaluation
- You can try it with your own files
- You can find one theme that seems usable in your own work
- You can understand adjustment points for AI agents and workflows
- You can explain the evaluation results to your team or manager
What to prepare before evaluation
Before starting the evaluation, prepare the following.
| Preparation Item | Details |
|---|---|
| Agentiqs itself | Install Agentiqs and make sure it can be launched |
| AI provider settings | Configure it so you can chat with AI |
| Sample data | Download sample data and place it in the specified folder |
| Evaluation theme | Decide what you want to check |
| Your own files | If possible, prepare 1 to 3 business documents you want to try |
| External tools | If you want to try Next Design or Lightning Review, make sure those tools can be launched |
Recommended environment for evaluation
For the initial evaluation, we recommend an environment like the following.
| Item | Recommendation |
|---|---|
| Input data | Sample data, or copies of your own files prepared for evaluation |
| Number of files | 1 to 3 files |
| Document volume | About 10 to 30 pages |
| Target work | Focus on one theme |
| Output destination | Evaluation folder, evaluation Excel file, evaluation review |
| External tools | Use a sample or copied project instead of production |
| Evaluator | A person who understands the work and can judge the validity of the output |
Before outputting directly to production files or production projects, always try it with evaluation data first.
What you should avoid at first
To make the evaluation go smoothly, avoid the following in the initial trial.
| What to avoid | Reason |
|---|---|
| Registering a large number of documents at once | If the results are poor, it becomes difficult to isolate the cause |
| Trying multiple business tasks at the same time | Evaluation points become scattered and it becomes hard to judge the effect |
| Outputting directly to production files | Unexpected results may remain |
| Creating a workflow from scratch at the beginning | It is faster to first understand the processing flow with samples |
| Evaluating with materials whose correct answer is unknown | You cannot judge whether the AI output is valid |
| Starting directly with external tool integration | There are many prerequisites, so it is easy to get stuck in the initial evaluation |
First, confirm that it works normally with the sample data.
After that, replacing it with a small theme close to your own work will make the evaluation easier to proceed.
Evaluation Route Details
Evaluation Route List
Choose from the following routes according to your evaluation objective.
| Evaluation objective | Recommended route |
|---|---|
| I want to quickly get the feel of it first | 60-Minute Quick Evaluation Route |
| I want to check the value of Agentiqs overall | 90-Minute Standard Evaluation Route |
| I want to see whether it can be used in my own work | Own Business File Evaluation Route |
| I want to use it for design reviews | Review Task Evaluation Route |
| I want to integrate it with Next Design or Lightning Review | DENSO CREATE Product Integration Evaluation Route |
| I want to consider team rollout | Half-Day Evaluation Route |
60-Minute Quick Evaluation Route
If you are using Agentiqs for the first time, we recommend this route first.
- Check the quick start
- Chat with AI
- Place the sample data
- Try "Search Across Massive Documents"
- Try "Automatically Generate Meeting Minutes"
- Check the evaluation points
What you can check with this route
| What you can check | Details |
|---|---|
| Basic operations | Whether you can chat with AI |
| Knowledge utilization | Whether it can answer by referring to documents such as manuals |
| Deliverable generation | Whether it can create deliverables such as meeting minutes from notes |
| Image of business application | Whether it seems possible to replace it with your own documents or meeting notes |
Recommended samples
| Order | Sample | Purpose |
|---|---|---|
| 1 | Search Across Massive Documents | Experience knowledge utilization |
| 2 | Automatically Generate Meeting Minutes | Experience deliverable generation from file input |
Points to look at in 60 minutes
| Perspective | What to check |
|---|---|
| Ease of understanding | Whether you can understand the operation flow |
| Answer quality | Whether the answer is based on the input data |
| Business applicability | Whether it seems possible to replace it with your own documents or meeting notes |
| Difference from AI chat | Whether it uses business data rather than being just a conversation |
90-Minute Standard Evaluation Route
If you want to check the value of Agentiqs overall, we recommend this route.
- Check the quick start
- Check the sample data list
- Try a knowledge utilization sample
- Try a meeting minutes generation sample
- Try a design document review sample
- Choose a theme close to your own work
Recommended samples
| Order | Sample | Purpose |
|---|---|---|
| 1 | Search Across Massive Documents | Experience knowledge utilization |
| 2 | Automatically Generate Meeting Minutes | Experience deliverable generation |
| 3 | Review a Design Document Based on Review Criteria | Experience design document review support |
What you can check with this route
| What you can check | Details |
|---|---|
| Knowledge utilization | It can answer by referring to documents |
| Document generation | It can generate deliverables from notes |
| Review support | It can check design documents based on review criteria |
| Output format | It can expand results into tabular format or file output |
| Application to your work | You can find ways of using it close to your own work |
Own Business File Evaluation Route
This route is for replacing the sample data with your own files after confirming operation with the sample data.
- Choose a sample close to your own work
- Run the sample as-is with the bundled data
- Select one of your own files
- Replace only the input file
- Check the output result
- Adjust the questions or criteria as needed
Samples that are easy to replace with your own files
| Sample | Data to replace | Work that is easy to evaluate |
|---|---|---|
| Search Across Massive Documents | Your own manuals, specifications, FAQs | Document search, inquiry handling |
| Automatically Generate Meeting Minutes | Your own meeting notes | Meeting minutes creation, ToDo organization |
| Review a Design Document Based on Review Criteria | Your own design documents, your own review criteria | Design document review |
| Generate Test Design from a Design Document | Your own design documents, test criteria | Test design |
Recommended conditions for evaluating your own files
| Item | Recommendation |
|---|---|
| Number of files | 1 file |
| Document volume | About 5 to 20 pages |
| Target work | 1 theme |
| Output destination | Evaluation file |
| Evaluation method | Compare with existing deliverables or reviewer feedback |
| Scope of use | Evaluation staff or a small team |
Review Task Evaluation Route
If you want to use Agentiqs for design document reviews or specification reviews, we recommend this route.
- Try extracting review criteria from review findings
- Try reviewing a design document based on review criteria
- Strictly review the design document by criterion
- Try Lightning Review integration as needed
Recommended samples
| Order | Sample | Purpose |
|---|---|---|
| 1 | Extract Review Criteria from Review Findings | Create review criteria from past findings |
| 2 | Review a Design Document Based on Review Criteria | Experience basic design document review |
| 3 | Strictly Review a Design Document by Criterion | Review comprehensively by chapter × criterion |
| 4 | Automatically Register Review Findings in Lightning Review | Register findings in a review management tool |
What you can check with this route
| What you can check | Details |
|---|---|
| Creating criteria | It can convert past findings into review criteria |
| Extracting findings | It can extract review findings from design documents |
| Coverage | It can reduce omissions by chapter × criterion |
| Creating deliverables | It can output to a findings list or Lightning Review |
| Application to work | You can judge whether it can be used for self-checks before review or for primary reviews |
DENSO CREATE Product Integration Evaluation Route
If you want to check integration with Next Design or Lightning Review, we recommend this route.
- Check the basics with document review samples
- Try a Next Design integration sample
- Try a Lightning Review integration sample
- Check how to replace it with your own project
Recommended samples
| Order | Sample | Purpose |
|---|---|---|
| 1 | Generate Test Design from a Design Document | Generate test design from documents |
| 2 | Generate Test Cases for Each Requirement Model | Generate test cases from Next Design requirement models |
| 3 | Review Next Design Models by Criterion | Review Next Design models with AI |
| 4 | Automatically Register Review Findings in Lightning Review | Register review findings in Lightning Review |
What you can check with this route
| What you can check | Details |
|---|---|
| Next Design integration | AI can retrieve and utilize design models |
| Model utilization | It can use design models rather than documents as input |
| Model generation | It can generate deliverables such as test case models |
| Traceability | It can handle relationships between requirements and test cases |
| Lightning Review integration | It can register review findings extracted by AI |
Half-Day Evaluation Route
If you want to reach a state where you can explain the evaluation results internally in about half a day, we recommend the following approach.
- Check the quick start
- Run three representative samples
- Replace one with your own file
- Have the person in charge check the output result
- Organize themes that seem applicable to business tasks
- Decide which AI agent or workflow to create next
Recommended configuration for a half-day evaluation
| Order | Content | What to check |
|---|---|---|
| 1 | Knowledge utilization sample | Whether it can answer by referring to documents |
| 2 | Meeting minutes generation sample | Whether it can generate deliverables from business notes |
| 3 | Design document review sample | Whether it can point out issues based on review criteria |
| 4 | Replacement with your own file | Whether it seems usable in actual work |
| 5 | Organizing evaluation results | Whether you can explain the benefits and issues of introduction |
Points to check during evaluation
During evaluation, please check from the following perspectives.
| Evaluation perspective | What to check |
|---|---|
| Usability | Whether the evaluator can execute it without getting lost |
| Use of input data | Whether the AI can handle documents, models, review criteria, and so on |
| Output quality | Whether the output has a level of detail usable for work |
| Groundedness | Whether the answers and findings are based on the input data |
| Reusability | Whether it seems usable repeatedly for the same work |
| Ease of correction | Whether it is in a format that people can easily review and revise |
| Ease of replacement | Whether it can be easily replaced with your own files and criteria |
| Workflow capability | Whether it seems possible to execute it as routine work |
| Tool integration | Whether it seems possible to reflect deliverables in external tools |
| Business effect | Whether it seems likely to reduce time, improve quality, and prevent omissions |
Points to compare with ordinary AI chat
When evaluating Agentiqs, also check how it differs from general AI chat.
| Comparison point | Ordinary AI chat | Agentiqs |
|---|---|---|
| Use of business documents | Tends to depend on what is pasted on the spot | Can refer to knowledge and files |
| Role by business task | Must be specified by prompt each time | Can define the role as an AI agent |
| Routine processing | Must be requested manually each time | Can be turned into a workflow |
| Deliverable output | Mainly chat responses | Can output to Excel, Word, Markdown, and external tools |
| Integration with development tools | Usually requires manual work | Can integrate with Next Design, Lightning Review, and others |
| Team rollout | Tends to remain personal use | AI agents and workflows can be shared more easily |
How to organize evaluation results
After evaluation, organizing the results as follows makes them easier to use for internal explanation and introduction planning.
| Item | Example entry |
|---|---|
| Samples tried | Search Across Massive Documents, design document review |
| Your own data tried | Product manual, functional design document |
| Good points | Document search is fast, usable as a first draft of review findings |
| Issues | Review criteria need adjustment, output format needs to be adapted for your company |
| Areas that seem applicable to work | Design document review, FAQ response, meeting minutes creation |
| Additional things to verify | Lightning Review integration, Next Design model review |
| Preparation needed for introduction | Organizing knowledge, creating AI agents, preparing usage rules |
Evaluation checklist
Use the following checklist during evaluation.
[ ] Agentiqs can be launched
[ ] AI provider settings are complete
[ ] You can chat with AI
[ ] Sample data has been placed
[ ] Sample settings can be imported
[ ] A knowledge utilization sample could be executed
[ ] A meeting minutes generation sample could be executed
[ ] A design document review sample could be executed
[ ] Output results could be checked
[ ] A sample close to your own work could be selected
[ ] You understand how to replace it with your own files
[ ] You can explain the difference from ordinary AI chat
[ ] You have decided the next work theme to try
How to move on to your own work
Once you have confirmed the value in the evaluation version, the next step is to replace it with your own work.
Proceeding in the following order will make failure less likely.
- Confirm normal operation with a sample
- Choose a sample close to your own work
- Replace only the input file
- Check the output result
- Adjust the review criteria and request text
- Create an AI agent for your company
- Turn it into a workflow as needed
- Share it with the team
You do not need to create a perfect AI agent or workflow from the start.
First, check whether it seems usable with a small work theme.
Common stumbling points in evaluation
| Stumbling point | Common cause | Countermeasure |
|---|---|---|
| You do not know what to try first | There are multiple samples and it is hard to choose | First run the 60-Minute Quick Evaluation Route |
| The answer becomes too general | Knowledge or input files are not specified | Clearly specify the target knowledge or files |
| The output is different from expectations | The question or criteria are vague | Make the target, conditions, and output format more specific |
| Too many findings are produced | There are too many criteria or the target scope is too broad | Narrow down the target chapters or criteria |
| It does not work well with your own files | The document structure or amount of information is insufficient | First try with 1 file, 1 function, and 3 to 5 criteria |
| You get stuck with external tool integration | Prerequisite settings or launch order are not correct | Check the sample steps and prerequisites |
| It is hard to explain the evaluation results | Evaluation perspectives have not been decided | Organize them according to the evaluation results table |
Common evaluation themes
In the evaluation version, we recommend starting with themes such as the following.
| Theme | Sample to try | Value to check |
|---|---|---|
| Manual search | Search Across Massive Documents | Improved efficiency of document investigation |
| Meeting minutes creation | Automatically Generate Meeting Minutes | Improved efficiency of deliverable creation |
| Design document review | Review a Design Document Based on Review Criteria | Creating a first draft of review findings |
| Strict review | Strictly Review a Design Document by Criterion | Comprehensive checking by chapter × criterion |
| Test design | Generate Test Design from a Design Document | Generation of downstream deliverables |
| Model utilization | Generate Test Cases for Each Requirement Model | Next Design integration |
| Findings registration | Automatically Register Review Findings in Lightning Review | Integration with review management tools |
Next steps after evaluation
Once you have confirmed the value in the evaluation version, consider the following next.
| Next step | Details |
|---|---|
| Decide your business theme | Decide which work Agentiqs will be applied to |
| Organize knowledge | Organize manuals, design standards, FAQs, and so on |
| Create AI agents | Create AI agents tailored to business purposes |
| Create workflows | Automate routine work |
| Decide usage rules | Decide responsibility for checking AI output, usage scope, and handling of confidential information |
| Try it with the team | Use it with a small group and identify improvement points |
| Incorporate it into business processes | Incorporate it into reviews, inquiry handling, test design, and so on |
Notes
- In the evaluation version, first confirm operation with the sample data.
- If you try it with your own files, use copied files prepared for evaluation.
- If you handle confidential information, personal information, or customer information, follow your internal rules and security policies.
- Treat AI output not as the final deliverable, but as a first draft for the person in charge to review.
- If you output or register to external tools, check the destination and update target in advance.
- Before outputting directly to production files, production projects, or production reviews, always verify with evaluation data first.
Pages to move on to next
Move on to the following pages according to your purpose.
| Purpose | Next page |
|---|---|
| I want to check how to start the evaluation version | Evaluation Version Start Guide |
| I want to check basic operations | Quick Start |
| I want to check sample data | Sample Data List |
| I want to experience the value of Agentiqs in 90 minutes | Experience the Value of Agentiqs |
| I want to try it with my own business files | Try with Your Own Business Files |
| I want to use knowledge | Use Knowledge |
| I want to create an AI agent | Create an AI Agent |
| I want to create a review-specialized AI agent | Create a Review-Specialized AI Agent |
| I want to create a workflow | Create a Workflow |
| I want to create a workflow for comprehensive review | Create a Workflow for Comprehensive Review |
| I want to check the tool list | Tool List |