Skip to content
revot
revot

Work

What we have built, described without the names.

Our clients have not yet approved public case studies, so this page describes the shape of each engagement — the problem, what was built and how it fits the workflow — and nothing that would identify them.

Read this first

Every engagement on this page is anonymised: no client names, figures or locations. Statuses describe the kind of work, not a verified production outcome. Fuller case studies are shared on request and published as clients approve them.

Status: Build
Software Revot designed and built for a client.
Status: MVP
A minimum viable product, scoped to be tested with real users.
Status: Pilot
A bounded trial with real users, designed to inform a decision.
Status: Demo
A working demonstration, not deployed for production use.
Status: BuildCustomer operations

Bilingual voice system for a customer-operations team

Problem shape

Calls needed to be handled conversationally in two languages, without building and maintaining a separate call flow for each. For a voice agent to feel usable, the pause between a caller finishing and hearing a reply has to be short — so speed was a design constraint, not a nice-to-have.

What was built

  • A voice pipeline combining telephony, streaming speech-to-text, a language model and synthesised speech
  • Conversation handling in either of two languages
  • Latency work across each stage of a conversational turn

How it fits the workflow

The system sits on the phone line in front of the team. Each turn was broken into stages — end-of-speech detection, transcription, reasoning, speech synthesis and playback — so each could be measured and improved separately rather than guessed at as a whole.

Fig. W1Anatomy of one conversational turn, in either language
Anatomy of one conversational turn, in either languageOne conversational turn: the caller speaks, the end of their speech is detected, streaming speech-to-text produces a transcript in the caller's language, a language model reasons over it, speech is synthesised in the same language and played back. Each stage is measured separately because latency is the gap between the caller finishing and hearing a reply.next turn — caller may interruptCaller speakslanguage a or bEnd of speechdetectedTranscriptstreaming sttReasoningllmSpeechsame languagePlaybackcaller hearsA each stage timed on its own: the engineering work went into the gaps between them
Main pathModel step
Built with:
Twilio · Deepgram · OpenAI · ElevenLabs
Status: BuildVoice-agent operations

Operations dashboard for a voice-agent service

Problem shape

Running voice agents for several customers means juggling accounts, usage and call history across a platform designed for one. An operator needed a single white-label view over all of it.

What was built

  • Sub-account management per end customer
  • A credits and usage ledger
  • Searchable call logs
  • Assisted prompt enhancement with a review step before changes go live

How it fits the workflow

The dashboard sits between the operator and the underlying voice platform’s API. Prompt changes are drafted with assistance but published only after a person has reviewed them.

Fig. W2One operator view over many voice-agent accounts
One operator view over many voice-agent accountsAn operator uses one dashboard to manage sub-accounts, a credits ledger and call logs. Prompt changes are drafted with assisted enhancement, reviewed by a person, and only then published to the underlying voice platform.publishusageOperatorwhite-labelDashboardone viewSub-accountsper customerCreditsusage ledgerCall logssearchablePrompt editorassisted draftsReviewbefore publishVoice platformvia its api
Main pathHuman gateModel step
Built with:
Retell
Status: MVPPhone-based services

Voice-agent MVP for a phone-based service

Problem shape

Before committing to a full build, the question was simple: would a voice agent handle this call flow well enough with real callers? An MVP was the cheapest honest way to find out.

What was built

  • A single, testable call flow on a hosted voice-agent platform
  • A review loop: test calls, transcript review, prompt and flow adjustments

How it fits the workflow

The MVP deliberately avoided deep integrations. Its job was to produce evidence — transcripts and observed behaviour — for a build-further decision.

Fig. W3An MVP loop: test the call flow before committing to the build
An MVP loop: test the call flow before committing to the buildThe MVP is scoped to one testable call flow, built on a hosted voice-agent platform, and exercised with test calls. People review the transcripts, prompts and flow are adjusted, and the loop repeats until there is enough evidence to decide whether to build further.evidenceissuesScopeone call flowBuildhosted platformTest callsscripted + realReviewread transcriptsDecidebuild further?Adjustprompts · flow
Main pathHuman gate
Built with:
Retell
Status: DemoInbound enquiries

Inbound voice-agent demonstration

Problem shape

A business wanted to hear, rather than read about, how an inbound voice agent would handle its calls before deciding whether to go further.

What was built

  • A working voice agent on a test number
  • Scripted scenarios covering typical calls
  • A simulated handoff to a person

How it fits the workflow

The demo was explicit about its boundary: live integrations, production telephony and monitoring were left for a later build if the project proceeds.

Fig. W4What a demo proves — and what it deliberately leaves out
What a demo proves — and what it deliberately leaves outInside the demo boundary: an inbound call is answered by a voice agent, which handles a set of scripted scenarios and shows a simulated handoff. Outside the boundary, left for a later build: live integrations with business systems and production telephony and monitoring.IN THE DEMOLATER, IF IT PROCEEDSInbound calltest numberVoice agenthosted platformScenariosscriptedHandoffsimulatedLive integrationsbusiness systemsProduction telephonymonitoring
Main pathException pathHuman gateModel step
Built with:
Vapi
Status: BuildTherapy practice

Privacy-conscious notes application for therapy practitioners

Problem shape

Session notes are among the most sensitive records a practitioner keeps. General-purpose note tools are not designed around who should — and should not — be able to read them.

What was built

  • A notes workflow for writing and reviewing session notes
  • A privacy-first data design, oriented towards limiting what the hosted service can see
  • Access-controlled storage with an access log

How it fits the workflow

The design is organised around a trust boundary between the practitioner and the hosted service, with only the minimum necessary data crossing it. We describe this as privacy-conscious design, not a guarantee: the precise properties depend on the threat model, and we share them in detail on request.

Fig. W5A notes workflow designed around a trust boundary
A notes workflow designed around a trust boundaryInside the practitioner's boundary, the practitioner writes and reviews a session note in a note editor. Only the minimum necessary data crosses to the hosted service, which stores protected records and keeps an access log. The precise security properties depend on the threat model and are shared on request.PRACTITIONER SIDEHOSTED SERVICEminimum data crossesPractitionersigned inNote editorwrite + reviewProtected recordsaccess-controlledAccess logwho · whenA design properties depend on the threat model — shared in detail on request
Main pathHuman gate
Status: PilotEducation

Education pilot for a learning institution

Problem shape

An education provider wanted to test whether a digital workflow could support day-to-day work — without the risk of a large rollout before anyone knew it would fit.

What was built

  • A bounded pilot of a digital workflow with a small group of real users

How it fits the workflow

The pilot was scoped to learn: one workflow, a small group, a bounded period, and an explicit decision point to extend, adjust or stop.

Fig. W6A pilot sized to learn, not to impress
A pilot sized to learn, not to impressOne workflow is selected, a small group of real users runs it for a bounded pilot period, feedback is gathered and reviewed with the people involved, and a decision is made to extend, adjust or stop.adjust and run againOne workflowchosen togetherSmall groupreal usersPilot periodboundedFeedbackreviewed togetherDecideextend · adjust · stop
Main pathHuman gate

Next step

Want the fuller story behind one of these?

Fuller case studies are shared on request, within what each client has agreed. Or tell us about your own process and we will talk about what applies.