Services
Seven capabilities. One way of working.
Capability 01
Custom software & SaaS
We design and build web applications, internal tools and SaaS MVPs for teams whose core process has outgrown spreadsheets or off-the-shelf software. Each build starts from the workflow it has to support, and ships in reviewable increments.
The problem
Off-the-shelf tools fit the average business. When your process isn’t average, people bend around the software instead — with workarounds, side spreadsheets and manual exports that only one person understands.
We validate
- Roles and permissions that match who actually does the work
- Input and webhook validation at every boundary
- Separate development, staging and production environments
- Secrets kept in secret management, never in code
Works with: Next.js · React · Tailwind CSS · Node.js · Express · Go · Supabase · PostgreSQL
Capability 02
AI-enabled applications
We build retrieval-backed (RAG) assistants, knowledge-base chat, internal copilots and bounded agentic workflows over your own documents and data — with retrieval quality, grounding, evaluations, access controls and human escalation designed in from the start.
The problem
Your team’s knowledge is spread across documents, SOPs, recordings and old conversations. Finding the right answer takes too long — and an assistant that gives a confident wrong answer is worse than no assistant at all.
We validate
- Retrieval quality — the right passages, ranked well
- Groundedness — answers traceable to their sources
- Evaluations for correctness, refusal and escalation
- Access controls and tenant isolation
Works with: OpenAI · Pinecone · LangChain · LangGraph · MCP · Supabase · PostgreSQL
Capability 03
Voice AI & telephony
We build inbound and outbound voice agents, IVR and call routing on telephony and speech platforms, with call logging and a clean handoff to a person. Before anything answers a real phone line we validate latency, interruptions, consent, language performance, fallback and cost per minute.
The problem
Calls arrive when nobody can answer them. Routine questions tie up skilled staff. And what was said on a call rarely reaches the system that needs it.
We validate
- Latency across the whole conversational turn
- Interruption (barge-in) handling
- Consent and recording notices
- Language performance for each language in use
Works with: Twilio · Retell · Vapi · Bland · Deepgram · ElevenLabs · OpenAI
Capability 04
Workflow automation
We automate the handoffs between the tools you already use — CRM and lead workflows, integrations, approvals and alerts — with exception handling designed in, so a failure reaches a named person instead of disappearing.
The problem
Leads sit uncontacted, approvals stall in inboxes, and someone copies data between systems every morning. Nothing is broken, exactly. It is just slow, manual and fragile.
We validate
- Idempotency — a retry never sends or charges twice
- Timeouts, retries and rate limits on every external call
- Webhook signature validation
- Exception queues with a named owner
Works with: n8n · Make · Gumloop · Activepieces · MCP · Node.js
Capability 05
Healthcare & privacy-conscious software
We build privacy-conscious software for clinical and therapy settings — EHR/EMR-adjacent workflows, session notes and secure records — designed with least-privilege access, audit trails, documented retention and plainly stated limits.
The problem
Clinical and therapy teams often handle sensitive records with tools that were not designed for them — or avoid software altogether because the privacy trade-offs are unclear.
We validate
- Least-privilege access and individual accounts
- Collecting only the data the workflow needs
- Documented retention, deletion and export
- Encryption choices — and their limits — written down
Works with: Next.js · React · Supabase · PostgreSQL · Node.js
Capability 06
Websites & commerce
We build conversion-focused websites and stores on WordPress, Shopify or custom stacks — with forms that route every enquiry to an owner, CRM automation from the first message, and sound SEO foundations.
The problem
A website that looks fine but sends enquiries nowhere useful, ranks for nothing, and needs a developer to change a headline.
We validate
- Page speed on real mobile connections
- Accessible forms with clear, specific errors
- Where each enquiry goes and who owns it
- Structured data and crawlability
Works with: WordPress · Shopify · Next.js · Tailwind CSS · n8n · Make
Capability 07
Mobile & local software
We build mobile apps for work that happens away from a desk, local and self-hosted AI deployments for data that should stay on site, and software concepts for hospitality operations such as hotels and motels.
The problem
Some work happens on a phone in a corridor or at a front desk. Some data should not leave the building. Most software assumes neither.
We validate
- Offline behaviour and sync-conflict rules
- Where data is stored and processed
- Hardware and model-size trade-offs for local AI
- An update and support path after launch
Works with: React · Node.js · Go · PostgreSQL · Supabase
Delivery principles
The same six rules, whatever we build.
- R.01
Understand the workflow first
We map trigger to completion — people, systems, data, handoffs and exceptions — before choosing a tool.
- R.02
Keep MVP scope testable
A first release should answer a question with real users, not try to answer every question at once.
- R.03
Design observability and recovery
Logs, alerts and a way back when something fails are part of the build, not an afterthought.
- R.04
Clarify data ownership and retention
Who owns the data, how long it is kept and how it is exported or deleted are agreed before we build.
- R.05
Keep people over consequential actions
Anything that moves money, changes records or speaks for you gets human approval where it matters.
- R.06
Write down scope and exclusions
You know what is included, what is not, and what we assumed — before work starts.
FAQ
Questions we are asked first.
Straight answers about how projects start, how AI accuracy is handled and who owns what.
Q.01How does a project with Revot start?
With the process that costs you the most time or money, not a feature list. You fill in a short project brief; we follow up with a discovery conversation that maps the workflow from trigger to completion — who is involved, which systems, what data, where it breaks. That becomes a proposal with the problem, proposed workflow, deliverables, exclusions, milestones and acceptance criteria written down.
Q.02Do you only build AI systems?
No. A lot of what we build is conventional software, integrations and automation. We use AI — retrieval-backed assistants, voice agents, model steps inside a workflow — only where it genuinely improves the process and can be evaluated. Sometimes the right answer is a clear owner and a better form.
Q.03How do you handle AI accuracy?
We never promise zero errors. We design accuracy into the workflow instead: measuring retrieval quality, grounding answers in cited sources, running evaluations for correctness, refusal and escalation, testing prompt injection, and re-running those checks after any model or prompt change. Uncertain cases are routed to a person, and consequential actions keep a human approval step.
Q.04Who owns the data and the work you build?
Data ownership, retention, export and deletion are clarified in the proposal before we build, along with which assets are transferred to you. At handover we transfer the agreed assets, revoke access we no longer need, and meet any agreed export or deletion obligations.
Q.05Can you work with the tools we already use?
Usually, yes. Much of our work connects existing systems — CRMs, inboxes, spreadsheets, telephony, helpdesks and accounting tools — through their APIs or automation runtimes such as n8n or Make. Part of discovery is confirming what each tool allows and where its limits are.
Q.06Is the software you build HIPAA or GDPR compliant?
Compliance depends on your organisation, the system’s full scope and a formal assessment — not on the tools or hosting provider used. We design privacy-conscious systems with access controls, audit trails, documented retention and human review, and we say plainly what has and has not been assessed. We do not claim compliance on a client’s behalf.
Q.07What happens after launch?
Deployment, known limitations and support arrangements are documented as part of handover, and support terms are agreed in the proposal. If something goes wrong in production, we triage, contain, communicate, restore or roll back, and write up the root cause and how to prevent it.
Q.08How much does a project cost, and how long does it take?
It depends on the workflow and the scope, so we do not publish fixed prices. The proposal sets out milestones, payment terms and what is excluded. We keep first releases deliberately small and testable, so you can judge the value before committing to more.
Next step
Start with the process that costs you the most.
Seven short questions about how the work happens today. No sales deck — the first conversation starts from your answers.