Services
One way to start. Two ways to continue. A pod for the firms that sell AI programmes.
Every engagement begins with the two-week sprint, so you see your own agent working on your own data before committing to anything larger. Read the offers as a specification sheet — that is how we write proposals too.
The specification sheet
Scope, duration, acceptance, price, exclusions — in one table.
| spec | start hereAgent Production-Readiness SprintTwo weeks, fixed price. One use case, one data source, one channel. | Build to ProductionSix to ten weeks, fixed scope, priced from the sprint. | Run and ImproveMonthly retainer after go-live, three-month minimum. |
|---|---|---|---|
| price | $7,500₹4 lakh for Indian clientscredited against a build signed within 30 days | $35k–60ktypical range; exact price in the sprint readout | from $8,000per month; half to one engineer-equivalent |
| duration | 10 working days from data access | 6–10 weeks | Monthly, 3-month minimum |
| scope | One agent use case; one data source; one channel (Slack, Teams, web or API) | The agent in production in your cloud, with the evaluation suite gating every release | Monitoring, eval regression, prompt/model/tool updates, new integrations and channels |
| acceptance | Evaluation score at or above the agreed target on 50–100 real cases; readiness scorecard delivered; readout held | Evaluation score at or above target on production traffic; 30-day warranty | Monthly business review and a response SLA agreed in the order |
| deliverables |
|
|
|
| not included | Production deployment, integrations beyond the one channel, model fine-tuning | Ongoing operation (see Run and Improve), new use cases | Net-new use cases (scoped as a sprint) |
| next step | Book the sprint call | Ask about this | Ask about this |
For consultancies and integrators: dedicated agent-engineering pods on a four-week minimum, white-label welcome, with our evaluation harness included. Ask about a two-week trial pod.
The sprint, day by day
Ten working days, one readout, no surprises.
The clock starts when we have data access, not when the SOW is signed. The deliverable is the Agent Readiness Gate's Evidence Report.
- Days 1–2
Scope and access
One use case with a measurable target; data access and environment set up; the evaluation cases agreed.
- Days 3–7
Thin slice
A working agent on your data, evaluated daily against the case set; cost per task measured from day one; the attack pack run.
- Days 8–9
Evidence Report
The 8-area readiness scorecard, the Gate's findings mapped to OWASP and NIST, the gaps that would stall production, and the fixed-price plan to close them.
- Day 10
Readout
Sixty minutes with your decision-makers: the demo on your data, the score, the plan, the price. You decide.
How we work
Five rules in every statement of work.
- 01Scope in numbers
One use case, one measurable target, one data source, one channel. The acceptance criteria are agreed before the clock starts.
- 02Inside your cloud
Access to your environment, not copies of your data. Engines and reports run and stay where your data already lives.
- 03Evaluate from day one
A real-case evaluation set and a cost-per-task meter exist before the first prompt is tuned. Every change is scored against them.
- 04Readout, not report-out
Sixty minutes with the decision-makers: the demo on your data, the score, the gaps, the fixed-price plan. You decide.
- 05Hand over, every time
Harness, runbook, training and the right to walk away are in every SOW. Retainers earn renewal; they do not hold you.
What we build
Agents first. Everything they need around them.
- AI agents and copilots — multi-agent workflows, tool use, MCP integrations, human-in-the-loop
- GenAI applications — RAG and knowledge assistants, document intelligence, chat and voice, summarisation
- LLM engineering — evaluation harnesses, guardrails, prompt and model optimisation, fine-tuning, cost control
- ML and data — forecasting, classification, recommendations, anomaly detection, pipelines, analytics
- MLOps and platforms — AWS, Azure, GCP, Kubernetes, CI/CD, monitoring, security and compliance
- Integration — SAP, Salesforce, ServiceNow, Microsoft 365, Zendesk, Intercom and your internal systems
For consultancies and integrators
Engineering pods: senior agent engineers plus the toolkit.
A delivery partner for the AI programmes you have sold. Gartner expects 70% of enterprises to abandon agentic AI built by vendors' forward-deployed engineers by 2028 for the lack of knowledge transfer, IP rights and a planned exit — so ours are contractual.
Engineers plus toolkit
Every pod ships with the Gate's harness, the Governor, MCPSentry attestation and the recorder — which is what separates a specialist bench from an offshore one.
White-label welcome
Evidence Reports under your brand with Rainkernel as the named assessor, or unbranded. Your client relationship stays yours.
Four-week minimum, two-week trial
Pods start at a four-week minimum. Ask about a two-week trial pod on a real deliverable before committing.
Your standards, your stack
Senior agent engineers working in your repos, your CI and your review process, in overlap with US, UK and EU hours.
FAQ
Questions about working with us
Why does every engagement start with the sprint?
Because in two weeks you see your own agent evaluated on your own data, with a cost number and a mapped findings list, before committing to anything larger. The sprint fee is credited against a Build signed within 30 days, so nobody pays twice for the same scoping.
Who owns the harness and the code?
You do, on full payment, as every statement of work says. Rainkernel keeps its pre-existing tools and the open-source Readiness Kit, licensed to you as embedded.
Do you work on-site?
Yes, in Hyderabad the same week; elsewhere by arrangement. Most engagements run remotely inside your cloud with daily written updates and a weekly call.
How do you invoice internationally?
In USD, by wire or card, from Rainkernel Technologies Private Limited in India. Indian clients are invoiced in INR with GST. Payment terms are in the order; the sprint is paid on signature.
What if the sprint shows the use case is not ready for an agent at all?
Then the readout says so, with the reasons and the numbers, and the recommendation might be a workflow change or a smaller automation. We would rather lose a Build than deliver one that stalls.
Start with the sprint.
Two weeks, $7,500, fixed. You see your agent evaluated on your own data and leave with the Evidence Report and a fixed-price plan — whether or not you continue with us.