The practice, and what it does
What is QuikA Consulting?
QuikA Consulting is my technology consultancy and software engineering practice. I'm Jamie Thomson, an AI consultant with 25 years' experience in HR and L&D, including senior roles at EdApp (SafetyCulture), Degreed, GP Strategies, Capita and Babcock.
The work divides into three things: connecting the systems a business already runs, building the software no product on the market provides, and advising on the people and process side of technology decisions. Those are covered in depth on Integration, Development and HR and L&D.
What technical capabilities does the practice cover?
Six areas, which sit underneath the three services rather than replacing them:
Custom software and mobile applications, built as cross-platform web and native iOS and Android apps from a single codebase.
AI and agentic systems, integrating frontier models into production software rather than bolting a chat box onto it.
Systems integration and data engineering, connecting disconnected business software into one source of truth, with reporting on top of it.
EdTech and e-learning engineering, including SCORM packaging, standalone course players and interactive learning engines.
Answer Engine Optimisation and technical SEO, which is optimising a site for AI crawlers and semantic search rather than for keyword ranking alone.
Lightweight content management and internal tools, including admin panels and custom CRMs built on git-backed deployment workflows.
Software architecture and the stack
What technology stack is used for custom software?
The front end is React with TypeScript, Tailwind CSS and Shadcn UI, with Zustand for global state and Framer Motion for interaction detail.
Mobile and progressive web apps use Capacitor to take a web application to native iOS and Android and into the App Store and Google Play, with service workers, offline-first architecture and push notifications through Supabase Edge Functions on FCM and APNs.
The back end runs on Supabase for PostgreSQL, authentication, storage and edge functions, with Railway for Python and media transcoding services, Trigger.dev for background job orchestration, and Vercel for hosting.
Third-party services in regular use include Stripe for billing and subscriptions, Sentry for error tracking, AssemblyAI for transcription, Replicate for generative media, and Google Analytics 4.
How are large language models integrated into production software?
Google Gemini and OpenAI models generate curriculum outlines, quizzes and expanded content. Anthropic Claude drives agentic auditing workflows, including AI readiness scoring, automated change branches, preview deployments and automated verification of the result.
The engineering that matters here is the reliability layer around the model. Structured output validation and JSON repair pipelines stop a malformed response from a model breaking a client-facing application, because a model that returns something almost valid is the normal case rather than the exception.
Some processing runs in the browser rather than on a server: FFmpeg compiled to WebAssembly for video manipulation, and Transformers.js for on-device language tasks, which keeps the data on the user's machine.
Integration, data and reporting
How is systems integration approached?
By designing one central source of truth rather than a web of point-to-point connections, each of which is one more thing to break. The method that gets there is set out on How we work.
Typical work connects a CRM such as Salesforce or HubSpot to an accounting platform such as Xero, QuickBooks or Sage, and puts custom Power BI dashboards over the result for operational, financial and sales reporting. Where an off-the-shelf CRM is heavier than the business needs, a lightweight custom one is often the better answer.
Systems connected in live businesses include Xero, QuickBooks, Sage, Shopify, Salesforce, HubSpot, Calendly, Microsoft 365, Teams, Slack, Zoom, Google Workspace, Google Analytics, Google Ads, Azure, Vercel, Railway, Supabase, WordPress, Wix, Squarespace, Claude, ChatGPT, Gemini, ElevenLabs, and bespoke systems with no public API.
What e-learning and EdTech standards are supported?
Full engineering compliance for SCORM 1.2 and SCORM 2004, including IMS manifest generation, JSZip packaging and the communication shims a course needs to talk to an external LMS.
That is not a capability described from the outside. QuikAuthor, my own product, is a SCORM creation tool: an AI-enabled course authoring and microlearning platform, built end to end by me, web app and native apps both. The SCORM engineering above is what it is made of.
Beyond packaging: custom standalone course players, and interactive learning engines with branching logic, feedback loops, structured pathways, prerequisite locking and event-driven progress tracking. The game formats include a Jeopardy style quiz board, a Bullseye style round, memory games and swipe games.
AI search visibility and technical audit
How is a site optimised for AI crawlers?
The premise is that an AI crawler does not run JavaScript, so anything that only appears after hydration does not exist to it. That drives static builds, prose that is readable without a browser, and no content that depends on a script to arrive.
On top of that: authoring llms.txt and llms-full.txt so a model can read the site in one fetch, interconnected Schema.org JSON-LD describing the practice, the person behind it, the services and this FAQ, tied by sameAs links to the same entities elsewhere, and removing duplicate content so a page is not competing with itself.
The content side matters as much as the markup. Self-contained answers with high information density, written so a passage lifted out of the page still makes sense on its own, because that is the unit an answer engine actually quotes.
What technical web standards and compliance work is covered?
Accessibility remediation to WCAG 2.2 AA, covering colour contrast, focus states and heading hierarchy. GDPR and Google Consent Mode v2 implementation. Email authentication through DMARC, SPF and DKIM. Deployment and performance work on Astro and Vercel builds.
Working together
How does QuikA Consulting charge?
The rate is £75 an hour, and work is quoted by the project rather than billed by the hour, so the price is fixed before anything starts. Published examples: a quick project such as a tool or a single integration is £1,125 over two days; an average project, meaning a connected system, is £3,375 over six days; a complex project touching core operations is £7,300 in under three weeks.
Where scope is genuinely exploratory and cannot be pinned down in advance, an agreed-outcome or day-rate arrangement is used instead.
Who does the work?
I do. You deal with me directly, and the person on the first call is the person who designs and builds what you need. When a project needs another specialist, I bring in someone I trust and tell you who they are.
Everything I build is documented and handed over, so you are never dependent on me: the code, the documentation and the access are yours.
What happens first?
A free scoping call. After that every project runs the same four stages: Analyse, Design, Build, Support. Analyse produces a written summary with milestones, responsibilities, pricing and a timetable, so the decision to continue is made against something concrete.
You own the code, the documentation and the access at the end of it. Start at Contact.
Book a free scoping call
A call with no fee and no sales deck. Share what you are aiming to achieve and where current technology is holding you back, and you will receive an honest evaluation of what is achievable, the timeline required, and the expected commercial value. If a project is not worth undertaking, I will tell you directly.