You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/research-agent/research-intro.mdx
+57-92Lines changed: 57 additions & 92 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,7 +4,7 @@ sidebar_position: 1
4
4
5
5
# Introduction
6
6
7
-
AI agent for pre-implementation research that suggests solutions after verifying against your unique system configuration.
7
+
SAP Research Agent is built for pre-implementation research that suggests solutions after verifying against your unique system configuration.
8
8
9
9
This means:
10
10
@@ -16,130 +16,95 @@ This means:
16
16
Check out the [free and open version](https://research.getadri.ai) where the results are validated against the 18 million standard objects in ECC 6.0 EHP 7 and 27 million standard objects in S/4HANA 2023.
17
17
:::
18
18
19
-
## Connected systems
20
-
21
-
If you are on Adri AI's Enterprise Plan, you can connect SAP Research Agent to your SAP environment to enable system-aware recommendations.
22
-
23
-
You can also check out the [free and open version](https://research.getadri.ai) where the results are validated against the 18 million standard objects in ECC 6.0 EHP 7 and 27 million standard objects in S/4HANA 2023.
24
-
25
-
### Installed Components
26
-
27
-
You can view the installed components in the attached ERP system.
28
-
29
-
importuseBaseUrlfrom'@docusaurus/useBaseUrl';
30
-
31
-
<figurestyle={{ textAlign: 'center' }}>
32
-
<imgsrc={useBaseUrl('/img/installed.png')}alt="Installed Components in ECC 6.0 EHP 7 system" />
33
-
<figcaption>Installed Components in ECC 6.0 EHP 7 system</figcaption>
34
-
</figure>
35
-
36
-
You may search the components in the attached system using the SAP Research Agent or by writing SQL-like queries (using ChromaSQL). More on ChromaSQL [here](search-index.mdx).
37
-
38
19
## Key Use Cases
39
20
40
-
### 1. Check Configuration Options
21
+
1.**Check Configuration Options**
41
22
42
23
Find standard SAP capabilities that will satisfy your requirement, thus avoiding unnecessary custom code development
43
24
44
-
```bash title="Sample prompt from a question asked in SAP Community"
45
-
I want to implement the requirement shared below in the SAP ECC 6.0 EHP 7 system. Does the implementation require the creation of custom objects?
25
+
2.**Write Functional and Technical Specification Documents**
46
26
47
-
- My goal is to minimize custom development as much as possible by using standard objects, wherever possible.
48
-
- Do not use the objects in the $TMP package
49
-
50
-
## Requirement
27
+
Create comprehensive functional and technical specifications for configuration changes and custom code.
51
28
52
-
`Update Positive Pay File with BofA to include Vendor name to prevent check fraud`
53
-
```
29
+
3.**Find SAP Objects**
54
30
55
-
### 2. Write Functional and Technical Specification Documents
31
+
Instantly locate relevant objects from your ERP system. Search across programs, function modules, tables, data elements, and more with results filtered to your system's active components.
56
32
57
-
Create comprehensive functional and technical specifications for configuration changes and custom code.
33
+
## How does SAP Research Agent work?
58
34
59
-
```bash title="Sample prompt"
60
-
Create a functional specification formy Change Request. Fillin the template shared below as applicable. If a heading or subheading is not relevant for this Change Request, drop it from your output.
35
+
### Core Architecture
61
36
62
-
Do not include requirement IDs, codes, or reference numbers.
37
+
-**AI Backend**: SAP Research Agent uses over 40+ Large Language Models with specialized configuration for SAP research tasks. More details [here](https://getadriai.github.io/code-agent-docs/faqs#which-models-are-being-used)
38
+
-**Vector Database**: The attached ERP system has been indexed into a ChromaDB vector database. It contains 18 million standard objects of ECC 6.0 EHP 7 and 27 million standard objects of S/4HANA 2023 like:
63
39
64
-
## Change Request
40
+
- Transaction codes, programs, classes
41
+
- Tables, views, data elements
42
+
- BAdIs, enhancements, function modules
43
+
- Menu paths, documentation, and more
65
44
66
-
SAP-to-JPMorgan Chase Payment Automation
45
+
### Which models are being used?
67
46
68
-
The current SAP payment process to JPMorgan Chase requires extensive manual intervention including
47
+
SAP Research Agent uses a combination of:
69
48
70
-
- manual execution of F110 payment runs,
71
-
- physical printing and signing of payment approval documents,
72
-
- manual generation and transmission of payment files via Z-code,
73
-
- manual download and processing of bank statements and receipts, and
74
-
- manual reconciliation of payment statuses in SAP.
49
+
- Internally fine-tuned models
50
+
- Public frontier models
75
51
76
-
This results in:
52
+
Depending on the subtask, an appropriate combination of models is chosen dynamically.
77
53
78
-
- processing delays of several hours to days per payment batch,
79
-
- high error rates in manual file generation and data entry,
80
-
- limited visibility into real-time payment status,
81
-
- compliance risks due to incomplete audit trails, and
82
-
- significant resource inefficiency with the Finance team spending 15-20 hours per week on repetitive manual tasks that could be automated.
54
+
| Model Provider | No. of models used |
55
+
| -------------- | ------------------ |
56
+
| Adri AI | 4 |
57
+
| Google | 9 |
58
+
| Mistral | 6 |
59
+
| Meta | 8 |
60
+
| OpenAI | 7 |
61
+
| Anthropic | 4 |
62
+
| Cohere | 4 |
63
+
| Total | 42 |
83
64
84
-
## Template
65
+
Note: Counts are as of November 2025
85
66
86
-
1. Executive Summary
87
-
1.1 Business Context
88
-
1.2 Objectives
89
-
1.3 Scope
67
+
### Key Capabilities
90
68
91
-
2. Current State Analysis
92
-
2.1 As-Is Process Flow
93
-
2.2 Pain Points & Quantified Impacts
94
-
2.3 Current System Landscape [if complex integrations]
69
+
1.**System Validation**
95
70
96
-
3. Future State Solution
97
-
3.1 Solution Overview
98
-
3.2 To-Be Process Flow
99
-
3.3 Key Changes Summary
71
+
SAP Research Agent queries the attached SAP system by writing ChromaSQL queries. ChromaSQL is a SQL-inspired domain specific language created by Adri AI.
100
72
101
-
4. Functional Requirements
102
-
4.1 Process Requirements (by business process)
103
-
4.2 Data Requirements (Master Data, Transaction Data, Configuration)
104
-
4.3 Business Rules & Validation Logic
73
+
2.**Web Search**
105
74
106
-
5. Integration & Interface Requirements
107
-
5.1 Integration Architecture
108
-
5.2 Interface Specifications (per system)
109
-
5.3 Data Mapping
75
+
SAP Research Agent also has the capability to search the internet for SAP solutions, best practices, and similar use cases.
110
76
111
-
6. User Experience Requirements
112
-
6.1 User Interfaces (Screens/Forms/Dashboards)
113
-
6.2 Reports & Analytics
114
-
6.3 Workflow & Approvals
77
+
3.**Document Processing**
115
78
116
-
7. Non-Functional Requirements
117
-
7.1 Performance & Scalability
118
-
7.2 Security & Access Control
119
-
7.3 Compliance & Audit
120
-
7.4 Usability & Accessibility
79
+
SAP Research Agent can read the uploaded attachments and optionally index them for future reference.
121
80
122
-
```
81
+
### Typical Workflow
123
82
124
-
### 3. Find SAP Objects
83
+
1. User asks a research question (text + optional attachments)
84
+
2. Depending on the question, the SAP Research Agent may think of a solution from its own knowledge base or search the web for relevant resources.
85
+
3. The solution is then validated against your SAP system's capabilities using the vector database.
86
+
4. The agent creates detailed implementation plans with specific T-codes, ABAP objects, and configuration steps.
87
+
5. The agent provides alternatives and visual diagrams when helpful.
125
88
126
-
Instantly locate relevant objects from your ERP system. Search across programs, functionmodules, tables, data elements, and more with results filtered to your system's active components.
89
+
## Connected systems
127
90
128
-
```bash title="Sample prompt from a question asked in SAP Community"
129
-
I created an internal IHC bank statement using transaction F9N7 (Single Run).
91
+
If you are on Adri AI's Enterprise Plan, you can connect SAP Research Agent to your SAP environment to enable system-aware recommendations.
130
92
131
-
When I attempted to post this statement using transaction FEBP, the posting did not execute. FEBP is the transaction we normally use to post internal IHC bank statements created via F9N1 (Mass Run).
93
+
You can also check out the [free and open version](https://research.getadri.ai) where the results are validated against the 18 million standard objects in ECC 6.0 EHP 7 and 27 million standard objects in S/4HANA 2023.
132
94
133
-
Which transaction should I use to post bank statements created with F9N7 (Single Run)?
134
-
```
95
+
### Installed Components
135
96
136
-
## How does SAP Research Agent work?
97
+
You can view the installed components in the attached ERP system.
137
98
138
-
When you ask Adri AI's Research Agent to perform any kind of research, you'll notice that it does one or all of the following activities:
99
+
importuseBaseUrlfrom'@docusaurus/useBaseUrl';
139
100
140
-
- Searching data in the attached system (say, ECC 6.0 EHP 7)
141
-
- Searching web
101
+
<figurestyle={{ textAlign: 'center' }}>
102
+
<imgsrc={useBaseUrl('/img/installed.png')}alt="Installed Components in ECC 6.0 EHP 7 system" />
103
+
<figcaption>Installed Components in ECC 6.0 EHP 7 system</figcaption>
104
+
</figure>
142
105
143
-
In order to search data in the attached system, the agent writes SQL-like queries. To be specific, these queries are written in ChromaSQL, a SQL-inspired domain specific language to query ChromaDB vector database. ChromaSQL has been created by the team at Adri AI.
106
+
You may search the components in the attached system using the SAP Research Agent or by writing SQL-like queries (using ChromaSQL). More on ChromaSQL [here](search-index.mdx).
144
107
145
-
You may search the components in the attached system by creating or writing your own ChromaSQL queries. Learn more [here](search-index.mdx).
108
+
:::info
109
+
Have any other questions? Check our [FAQs](../faqs.md).
0 commit comments