Developer Needed – Multi-AI Comparison & Consensus Platform MVP
Job Description
I am looking for an experienced developer to build a working MVP for a new AI comparison platform.
The concept is simple:
A user enters one question or prompt once.
The platform then sends exactly the same prompt to multiple AI providers/models and displays the responses side-by-side.
The user should then be able to ask the platform to compare the different responses and produce a consolidated view showing:
- Where the AI models agree
- Where they disagree
- Important points identified by only one model
- Potential factual conflicts
- A combined or consensus answer
- An overall confidence/consensus indication
This is initially a validation MVP, not a fully developed commercial SaaS platform.
I want the first version to be simple, reliable and inexpensive enough to test with real users.
Core MVP Functionality1. User Prompt
Simple text box where the user enters a question.
Example:
Assess this company as an acquisition target. Identify the principal risks, opportunities and questions that should be asked before proceeding.
User clicks:
ASK THE AI PANEL
2. AI Selection
The user should be able to select which AI models/providers they want to query.
For example:
- OpenAI / ChatGPT models
- Anthropic / Claude
- Google / Gemini
- Perplexity
- Grok / xAI
- Other providers where suitable API access is available
The architecture should make it reasonably straightforward to add or remove providers later.
I am not expecting the developer to circumvent provider restrictions or use consumer subscriptions. The platform should use legitimate API access.
3. Parallel Responses
The exact same prompt should be sent to the selected models.
Results should appear in separate panels or columns.
For example:
ChatGPT
Response...
Claude
Response...
Gemini
Response...
The user must be able to read each complete answer separately.
4. Consensus / Comparison Engine
There should be a button such as:
COMPARE RESPONSES
This should create a separate summary containing:
Areas of Agreement
What most/all models agree upon.
Areas of Disagreement
Where conclusions differ.
Unique Insights
Important issues raised by only one model.
Conflicting Claims
Where models appear to make materially different factual statements.
Combined Conclusion
A concise synthesis of the strongest points.
Suggested Follow-Up Questions
Questions the user should ask next based upon gaps or disagreements between the models.
This comparison can itself be performed by one nominated AI model in Version 1.
5. Blind Comparison Mode
I would like a simple optional feature where the model names can initially be hidden.
The responses would appear as:
Response A
Response B
Response C
Response D
The user can then select:
Which response did you find most useful?
Only after voting should the model identities be revealed.
The result should be stored.
This will eventually allow us to establish which AI models perform best for different categories of question.
6. File Upload
Users should be able to upload documents for analysis.
Initially:
- PDF
- Word documents
- Text
- Ideally Excel / CSV if practical within the MVP budget
The same document/content should be made available to each selected AI as consistently as possible.
Example use case:
A user uploads company accounts and asks:
Analyse this company. Identify maintainable earnings, financial risks, opportunities, unusual items and information I should request before considering an acquisition.
7. Prompt History
Basic user history containing:
- Date
- Original question
- AI models used
- Responses
- Consensus result
- User's preferred response, if selected
The user should be able to reopen an earlier comparison.
8. User Accounts
Simple authentication:
- Register
- Login
- Forgotten password
- Individual account history
No complicated enterprise permissions are required for Phase 1.
9. Administration
Simple administrator area allowing me to:
- View registered users
- View number of queries
- See which models are being used
- See user rankings/votes
- Enable or disable individual AI providers
- View approximate API usage/costs
10. API Cost Protection
This is important.
The system should prevent uncontrolled API spending.
Ideally the administrator should be able to specify:
- Maximum number of models per request
- Maximum response/token size
- Daily user query limit
- Monthly user query limit
- Global API expenditure limit or alerts
The platform should log approximate API cost per request wherever practical.
What Is NOT Required in Phase 1
I do not presently need:
- Native mobile apps
- Complex corporate dashboards
- Team collaboration
- Enterprise SSO
- Full subscription billing
- Advanced analytics
- AI training/fine-tuning
- Our own large language model
- Elaborate branding
- Complex integrations
- Large-scale infrastructure
The objective is to create a working product that can be tested quickly.
Preferred User Experience
The interface should be extremely simple.
Conceptually:
QUESTION
⬇
SELECT AI PANEL
ChatGPT ☑
Claude ☑
Gemini ☑
Other ☐
⬇
ASK
⬇
AI 1 | AI 2 | AI 3
⬇
COMPARE RESPONSES
⬇
AGREEMENT | DISAGREEMENT | UNIQUE INSIGHTS | CONSENSUS
The platform should feel like a serious business/research tool, rather than an AI novelty site.
Important Technical Requirement
I do not want the product hard-coded around individual providers.
Please create an abstraction/interface for AI providers so additional models can be added later without rebuilding the entire application.
Conceptually:
User Interface
↓
Application
↓
AI Provider Layer
↙ ↓ ↘
OpenAI | Anthropic | Google | Other
↓
Comparison Engine
↓
Results Database
This modular architecture is important.
Data & Confidentiality
Users may eventually upload confidential business information.
The developer should therefore consider:
- Secure transmission
- Secure file handling
- Encryption where appropriate
- Avoiding unnecessary permanent storage of uploaded documents
- Clear deletion process
- API keys stored securely server-side
- No API keys exposed in browser code
- Basic logging without unnecessarily recording confidential source material
Please explain the security approach in your proposal.
Deliverables
I expect the completed MVP to include:
- Working web application
- Source code
- Database
- AI API integrations
- User registration/login
- Prompt interface
- Parallel AI responses
- Consensus/comparison function
- Blind comparison/voting
- Basic document uploads
- Query history
- Admin interface
- API usage controls
- Deployment to an agreed hosting environment
- Basic documentation explaining how the system works
- Instructions for adding additional AI providers
- All source-code/IP rights transferred to me on payment
Intellectual Property
All code developed specifically for this project, database structure, interface design and associated project materials must become my property following final payment.
The developer may obviously retain ownership of pre-existing libraries, frameworks and open-source components subject to their normal licences.
Please identify any significant third-party paid components before using them.
Proposal Requirements
Please do not send a generic AI-generated proposal.
I would like your proposal to answer these questions:
- Have you personally integrated two or more LLM/API providers into the same application?
- Which providers have you integrated?
- How would you architect this application?
- What technology stack would you recommend?
- How would you handle document uploads across multiple AI models?
- How would you prevent excessive API expenditure?
- How would you protect confidential user data?
- What parts of the specification would you simplify for the MVP?
- How long would development take?
- What is your fixed price for the MVP?
- Please provide examples of relevant work.
Please begin your proposal with the words "AI PANEL" so I know you have read the specification.
...
Show more