Skip to content
Let’s talk

Client success stories

Launched in weeks. Run by a fraction of the usual team.

26 companies I have worked with as fractional CTO: what we built, what it saved and what got better, in numbers. Start with the story closest to your company. The short version is open to everyone. The full case, with the architecture, the costs and the team plan, I send personally as a PDF.

ResultsThe strongest figure from each of eight stories

Stories

02Metalworking and parts manufacturingMetizPromServiceA factory saves $300,000+ a year and halved its IT costs after two engineers I hired replaced its contractors$300k+cut from the software and licence bill, every year50%+less spent on IT, the team included, than the contractors cost03Job search and AI CVsCV RocketOne engineer and a fractional CTO took CV Rocket from zero to sales in five weeks and to 1,000+ customers4 weeksfrom zero to a working product, where quarters are the norm1,000+customers since the first sale in week 504AI call centers and telephonyCallbrainAn AI operator closes 92% of calls without a human, in 57 languages, and costs 3 to 5 times less a minute92%of calls closed by the AI operator, with no human on the line3–5×cheaper a minute than a human operator, at $0.16 billed per second05B2B SaaS, inbound leadsLeadPendingTwo people shipped a lead inbox in 24 hours: leads get a reply in 10 minutes, not the market's 40+ hours10 minto answer a lead, where the market average is 40+ hours24 hfrom the first commit to production, where weeks are the norm06Security for AI agentsSallyport CloudTwo people did a security team's work: AI agents make 100,000+ calls a day to 120 services and never hold a key100,000+agent calls a day through the gateway, and not one key has leaked<20 msof gateway overhead per call, so security does not slow agents down07AI software developmentKoder.aiTwo people shipped an AI app builder in six weeks, not quarters, and 70,000+ people in 100+ countries use it6 weeksfrom start to a working platform, where such products take quarters70,000+users in 100+ countries, served by a team of two08AI model gatewayAIRouterIn five weeks, one person launched an OpenRouter-class AI gateway that has since processed 168 billion+ tokens168B+tokens processed in 29.5M+ API requests, counted live on its site5 weeksto launch with a team of one, where such platforms take quarters09Community for women foundersFemSFOne platform replaced four for 357 Silicon Valley women founders and cut application review from 15 minutes to 115 → 1minutes to review an application, with the screening already done1,000+applications pre-screened automatically before a person reads them10Video Q&A for public figuresAskOnCamThree people launched a US video Q&A service on its own streaming in six weeks, where others staff whole teams6 weeksfrom start to a launched MVP, with two engineers and me<1 sstreaming latency on a stack the company owns instead of renting11Legacy code migration with AICodeHeroLegacy code moved to Go, Rust and TypeScript 5 to 10 times faster and 70% cheaper, by 5 people instead of 265–10×faster than a manual rewrite of the same legacy system−70%cost against a usual migration of the same system12CNC machines and manufacturing automationEast CNCFrom zero engineers to a CNC product that earns $500,000+ a year and cuts part design time by 70%$500k+a year from a product built by a team of six−70%part design time at customers, with our own AI and CAD13Rescuing AI-built codeFixMyMess100+ broken AI-built apps rescued in 48 hours on average, 5 to 10 times cheaper than a rewrite from scratch5–10×cheaper than rewriting the app from scratch99%of rescues succeed, and the founder keeps the product and users14Backlink mining and SEOSEOBoostyThree of us built a product that reads up to 1B pages a minute, holds 10B+ links and earns $350k+ a year1Bpages a minute, read by a product three people run$350k+a year in revenue from a team of three15Software development companySaaS ProductionTwice the clients and margin up from 20% to 60%: I took a software company from almost 80 people to 12 with AI20% → 60%project margin, with a team of 12 working on AI tools2×as many clients, served by a sixth of the old team16Personal data removalREMOVEThree people built a privacy product with 2M+ removals from 500+ data brokers, 95%+ with no human involved95%+of removals run with no human involved, so a team of three is enough2M+removals completed, each one costing the company cents17Email validation APIVeriMailTwo people run an email validation API with 1M+ checks a day and 99.99% uptime, and no ops team99.99%actual uptime against a 99.9% SLA, with two people and no ops team1M+email checks a day from businesses that put every sign-up through it18Hotel technology and guest registrationEMISFour people now run a platform that registers 29 million hotel guests a year in the CIS, a job that took 2529Mguests registered a year, across every CIS country25 → 4people run the whole platform now, and it keeps growingEnterprise ATS19Applicant tracking and hiringEnterprise ATSSix people rebuild a US hiring platform for 1,000+ companies and 2.5M candidates with no pause in hiring2.5Mcandidates in a system six people are rebuilding from legacy1,000+client companies, defense and healthcare among them, hire on itHotel Platform20AI hotel management systemHotel PlatformEight engineers built an AI hotel platform with 1,000+ integrations, where competitors need 50 to 100 people8engineers do the work competitors need 50 to 100 people for1,000+integrations, so a hotel keeps the locks, Wi-Fi and systems it has21AI chat on abliterated modelsSubmissive AIFour people took an AI product from idea to launch in 9 weeks, with inference 60%+ cheaper than API providers9 weeksfrom idea to launch, for one of the hardest kinds of AI product60%+cheaper inference than API providers, on our own engine

More clients

Their full cases are not public. I send each one myself, as a PDF, with the numbers behind the result.

22QueueStoneERP for mental-health providersTechnology consultantA clinic ERP now ships more often and costs less to build and test, after I put AI into its deliveryQueueStone makes an ERP for mental-health providers, and clinics run on it every day, so every release has to be safe. I brought AI tools into its delivery cycle and changed how the team works. It now releases more often and spends less on development and QA, without the usual army of testers.Request the full case
23FluxoPayPaymentsAI transformationA payments company cut its engineering team and overhead and lost none of its delivery speedFluxoPay is a payments company. I automated its engineering workflow from end to end and moved it to a smaller team model with less overhead. It ships as fast and as well as before on a smaller payroll, and the company still has room to scale.Request the full case
24Activity ControlWorkforce analyticsProduct architectureA workforce analytics platform enterprises run at scale, on an architecture I designed and a team I ledActivity Control does employee monitoring, activity tracking and analytics for enterprise customers. I designed the product architecture that carries their scale, led the engineering team, set its standards and made the core technical decisions, so the product keeps up with its largest customers.Request the full case
25StrazhAccess control and identityArchitectureRestricted sites clear visitors automatically, from the ID document to the face check, with no manual checksStrazh lets restricted sites clear visitors automatically, work that usually takes staff checking documents by hand. I designed its architecture: document recognition, face authentication and identity checks, sized for the load these sites carry and the regulations they answer to.Request the full case
26Face Recognition SystemMachine visionCTOFace recognition that stays accurate on real cameras and in hard installations, where many systems failFace Recognition System recognises faces from real cameras, including installations where conditions are hard. As CTO I owned the product architecture, led the engineering and made the core technology decisions that keep recognition accurate in the field, where systems tuned in a lab lose their accuracy.Request the full case

The full case

The numbers behind each result are in the PDF

Architecture, costs and team numbers do not go on an open page. I send them to people who are working on a similar problem and want the same result.

  1. 01

    Choose a story

    Pick the company closest to your situation, or ask for several.

  2. 02

    Leave your work email

    Your name, your company and a line about what you are working on.

  3. 03

    I send the PDF myself

    Usually within one business day, with a note on what applies to you.

oleg.isSent on request

Full case · PDF

Client success stories

Launched in weeks. Run by a fraction of the usual team.

  1. 01Aula
  2. 02MetizPromService
  3. 03CV Rocket
  4. 04Callbrain
  5. 05LeadPending
  6. 06Sallyport Cloud
  7. and 20 more
Prepared by Oleg SotnikovPDF

Get the full case as a PDF

Pick a story and leave your work email. I send the PDF myself, usually within one business day.

I send every case myself and use your email for nothing else.

Want the next story to be about your company?

In a 30-minute call we pick the first task to hand to AI and estimate what it will save you.

Book a callThe call is free, and you talk to me directly.