PROJECT 02 · PROJECT TEAM
Arun Pandian J
Solution Architect — Bridge AI
Contributes to the solution architecture, technical structure, and implementation approach.
Phase · Prototype development Delivery target: Not set
PROJECT 02 · FOUNDER-ORIGINATED
DomainBridge AI
workflows shaped around real work.
DomainBridge AI is in development as a focused AI-assisted workflow project shaped around a team’s domain, context, and operating data. Its current prototype explores one practical business problem—purchase-order and supplier follow-up.
Prototype walkthrough · Not a live production service · No subscription checkout BUSINESS VIEW · PROCUREMENT OPERATIONS
Keep supplier follow-up
clear, timely, and reviewable.
The current business prototype focuses on repetitive coordination around purchase orders—not autonomous procurement or automatic supplier messaging.
ILLUSTRATIVE BUSINESS WORKFLOW
SAMPLE DATAPurchase-order follow-up
PO status, items, delivery date, contact preferences, and business priority.
Check for missing acknowledgement, delivery delay, urgency, and escalation rules.
Inspect the decision and draft message; a person decides what to send or change.
This is a product-flow illustration, not a live dashboard. The prototype does not send email, WhatsApp, or purchase orders.
Less repeated checking.
Bring order facts, supplier preferences, and reminder rules into one review flow.
Earlier visibility into risk.
Surface acknowledgement gaps and delivery concerns for a human to assess.
More consistent follow-up.
Prepare clear, context-aware messages without claiming the system contacts suppliers itself.
PROTOTYPE WALKTHROUGH
How to use
the current demo.
The working prototype presents one dashboard with configuration panels, file input, and an agent result preview.
- 01
Bring in the purchase-order context
Enter order details manually or upload a CSV/text file with purchase-order and supplier information.
- 02
Set supplier and follow-up rules
Review the supplier contact, preferred channel and language, reminder timing, escalation setting, and human-approval preference.
- 03
Choose the demo or model setup
Use the built-in demo behavior, or configure an OpenAI-compatible provider and model for a connected prototype run.
- 04
Run and review the result
Inspect the follow-up decision, risk reasons, and suggested supplier message before anyone takes action.
Use synthetic or approved test data in a demo. A configured model provider may process the data you submit under that provider’s terms; no security review, production deployment, or data-retention controls are represented here.
RECOMMENDED TECHNICAL BLUEPRINT
Deterministic where it matters.
AI-assisted where it helps.
This is a proposed production architecture, not a claim about what the current prototype already delivers.
Business workspace
Role-aware web app for procurement, operations, approvers, and administrators; guided CSV import with schema validation and clear error reports.
Workflow orchestration
Versioned workflow definitions, deterministic rules for reminders and risk signals, idempotent jobs, and explicit human-approval gates.
Provider-neutral model gateway
We plan to publish a free, general-purpose MakeLifeEasy model for the Free Pack; it is not released or integrated yet. Domain-specific workflows would add their own data, rules, and evaluation, with provider adapters as appropriate.
Tenant-scoped records
Separate customer workspaces for purchase orders, supplier preferences, workflow runs, policy settings, and audit events.
Reusable, not one-size-fits-all
Each domain gets its own data schema, approved sources, workflow rules, evaluation set, and explicit capability boundaries.
Controlled integrations
Start with CSV import/export. Add ERP, email, or messaging connectors only after consent, permissions, delivery logs, and approval behavior are defined.
Observable and testable
Run-level logs, redacted traces, quality evaluations, latency and failure monitoring, and a way to replay representative test cases.
SUGGESTED REFERENCE STACK · CHOOSE AFTER PILOT NEEDS ARE KNOWN
Keep the first release boring, modular, and supportable.
- Web workspace
- Next.js + TypeScript; server-side sessions and role-aware routes.
- Workflow API
- FastAPI service with typed request/response schemas and versioned rules.
- Business records
- PostgreSQL with tenant scoping; enforce isolation in the data layer, not just the UI.
- Background work
- Managed queue and workers for imports, long-running model calls, retries, and scheduled reminders.
- Files and secrets
- Private object storage for temporary uploads; a managed secrets vault for provider credentials.
- Quality and operations
- Structured logs, metrics, traces, and an evaluation set for each workflow/domain pack.
SECURITY BASELINE BEFORE CUSTOMER DATA
Trust is designed into the workflow.
- Encrypt traffic in transit and stored business data at rest.
- Keep provider credentials in a server-side secret store; never expose keys in browser code or logs.
- Document each model provider’s data-retention terms and prefer no-training settings where available.
- Enforce workspace isolation, least-privilege roles, and auditable administrative actions.
- Set documented retention, export, and deletion controls before onboarding real records.
- Require human approval before any external message or consequential business action.
- Evaluate prompt injection, malformed uploads, hallucinations, and provider failures before release.
Suggested implementation path: validate one procurement workflow with synthetic data, then add authentication and tenant boundaries, persisted runs and audit events, provider secret management, evaluations, and only then carefully scoped integrations.
PROPOSED COMMERCIAL MODEL · NOT LIVE
Start small. Price for
useful work, not hype.
Suggested India-first monthly pricing for discussion and validation. Prices and limits are recommendations—not current offers or implemented entitlements.
Free Pack
₹0 / month
For individuals exploring the workflow with synthetic or sample data.
- 1 user · 1 workspace
- Planned access to our free, general-purpose model—not released yet
- Proposed fair-use workflow limit, to be validated before launch
- Manual entry and CSV import/export
- No domain-specific model, live integrations, or automated messaging
- 7-day run history
Builder
₹1,499 / workspace / month
For a small team validating one repeatable operational workflow.
- Up to 3 users · 1 workspace
- Up to 500 workflow runs / month
- Up to 3 configured workflows
- Bring your own model-provider key
- 30-day run history and basic exports
- Human review queue and email support
Business
₹5,999 / workspace / month
For teams coordinating supplier follow-ups across a shared operation.
- Up to 10 users · 1 workspace
- Up to 3,000 workflow runs / month
- Up to 10 configured workflows
- Role-based access and approval policies
- 90-day history, audit export, and usage view
- Priority support during business hours
Scale
Custom / scoped proposal
For larger deployments that need integration, governance, and rollout planning.
- Custom users, workspaces, and run volumes
- Integration and deployment discovery
- Security and data-processing review
- Retention, support, and service targets agreed in writing
- Pricing after technical discovery and scope
Pricing assumptions to validate: Indicative INR list prices, billed monthly per workspace; applicable taxes extra. The Free Pack is intended to use a general-purpose model published by MakeLifeEasy, subject to a fair-use limit; publication, hosting, license, and quotas are future work, not available today. Paid tiers may use a customer-provided model-provider key. No automatic overage: pause new runs at the plan limit and let the workspace owner upgrade or wait for reset. Annual discounts, free trials, and final run definitions should be tested with pilot users before launch. If validated, begin with month-to-month billing and owner-controlled cancellation; publish the renewal and cancellation terms before collecting payment.
“Workflow run” proposal: One accepted business record processed through one published workflow, with one final reviewed result. Retries caused by platform errors should not count against the allowance.
Prototype reality: The current UI only exposes “Free Demo” and “Premium Mock” toggles and mentions “Ultra” for custom tools. It has no live billing, subscription entitlements, user accounts, usage metering, or plan-limit enforcement.
HONEST SCOPE
A focused prototype,
not a finished platform.
The current prototype demonstrates procurement follow-up decisions and can call a configured OpenAI-compatible model. The business workflow is to review order and supplier context, identify follow-up triggers, and prepare a suggested message for human review.
There is no production billing, supplier messaging integration, persistent business workspace, or proof of performance claims in this prototype. The architecture and subscription tiers above are recommendations for future validation—not implemented capabilities. The planned free general-purpose model has not been published or integrated. A future deployment would need a clearly scoped use case, appropriate data controls, and human review.
IN DEVELOPMENTHave a similar workflow
Meet MakeLifeEasy Have a similar workflow
worth exploring?
We are open to conversations about a small, clearly scoped pilot with a real operational problem.