The goal
In two weeks, build and validate the first working loop from supplier contract to AI analysis to human approval to supplier communication to contract closure. At the same time, map Qtech's real data and systems and produce the exact 10-week execution backlog for the build that follows.
The contract closing loop
- Contract
- Read
- Extract
- Check mandate
- Draft response
- Human approve
- Send
- Supplier response
- AI review
- Human approve
- Finalize
- Close
Communication
Qtech selects the channel. Additional channels are not part of the 2-week scope.
End deliverables
- Full Travel OS plan document
- Working approval-based Contract Closing Agent
- Up to 100 contracts processed during discovery
- Mandate / rule framework
- Email + one messaging channel in the prototype
- Agent architecture and deployment setup
Two-week execution
Discover, connect, set up
- Map the actual contracting workflow: receive → review → negotiate → approve → close
- Define AI vs. human actions, approval gates and escalation rules
- Identify sources: live system, DB/API, document repository, supplier data, historical contracts, commercial parameters
- Prepare up to 5 machines / agent workspaces
- Load and process up to 100 representative contracts
- Identify contract structures: rates, seasons, supplements, cancellation, release, allotment, taxes, payment terms and special clauses
Build the contract closing loop
Contract → read → extract → check mandate → draft response → human approve → send → supplier response → AI review → human approve → finalize → close
- Structured contract parameters and confidence / exception handling
- AI-generated supplier responses
- Human approve / edit / reject before external communication
- One selected messaging channel + email
- Supplier responses returned to the agent workflow
- Final contract state and closure recorded
10-week expected action plan
Discovery produces the exact backlog. This is the baseline build sequence.
Contract intelligence
Production ingestion, classification, extraction, structured contract model, confidence scoring, versioning, review UI.
Commercial intelligence
Rate normalization, historical comparison, system data, mandate/rules, accept / negotiate / reject recommendation.
Negotiation + communication
Conversation state, supplier context, response generation, approval thresholds, follow-ups, audit trail.
Live data integration
Connect required Qtech systems, databases, APIs, contract repositories, rates and supplier master data.
Contract loading
Move approved contract parameters into the operational workflow; validation and testing.
Monitoring
Expiry, release deadlines, rate changes, missing data, supplier follow-ups and escalations.
Production pilot
Run real workflow at larger volume, measure accuracy, turnaround and exceptions; harden and hand over.
Scope guardrails
- 2 calendar weeks / 40 human hours
- Up to 100 representative contracts
- Up to 5 agent environments / machines
- One messaging channel + email
- Prototype is approval-based; no autonomous contractual commitment
- Qtech provides representative contracts, required stakeholders and available test/sandbox access
- Production-scale deployment, migration and additional channels are part of the subsequent implementation
“Success is a working loop, not a slide deck.”