Straight answers, no hedging

For a mid-sized business, a first production automation typically costs £15,000–£60,000 depending on process complexity and integrations, with discovery-and-strategy engagements from £8,000. We quote fixed scope, not open-ended day rates, and every proposal includes the projected annual saving so the payback maths is explicit before you commit.

Three components: model/API usage (typically £100–£1,000 a month at mid-market volumes), hosting, and an optional support retainer if you want us to monitor and extend the system. Everything we build is handed over with source code and documentation, so running it in-house is always a real option — there is no licence ransom.

Success metrics are fixed in writing during the strategy phase — hours saved, error rates, turnaround times — and measured against a baseline we capture before building. Our track record across 50+ deployments is 85% average time saved on automated processes and over £2M in client savings. If discovery shows the ROI case doesn't hold, we tell you and stop there; that discipline is why our case-study numbers are real.

It is the engagement structure, not a slogan: 10 days of discovery, 10 of strategy and architecture, roughly seven weeks of weekly shipped increments, then production hardening and a measured ROI report. Our recent case studies ran 9–12 weeks from first workshop to production. Multi-process programmes take longer — but each individual process still ships inside its own 90-day arc.

Plan for 2–4 hours a week from a project owner and short weekly sessions with the people who run the process being automated. The heaviest ask is early: discovery workshops in weeks one and two. From week three your involvement is mostly testing real increments and telling us where they're wrong — which is exactly what makes the end system fit your operation.

Typically within two to three weeks of a signed proposal. The free strategy call comes first: 30 minutes to pressure-test whether your process is worth automating at all. If it isn't, we say so on the call and you've spent half an hour, not a procurement cycle.

Your data stays in your controlled environment or a dedicated deployment with agreed residency (UK/EU for British clients, India options for Indian clients). We do not use client data to train shared models, and commercial API providers are configured with training opt-outs. Data-flow diagrams and processing agreements are part of every engagement's paperwork.

CortexIntel is a trading name of Infosec Solution Limited — we spent over ten years in cybersecurity, managing 200+ enterprise sites, before building our AI practice on top of it. Least-privilege access, audit logging and failure containment are our defaults, not add-ons, which is why regulated clients in legal, healthcare and financial services trust us with production systems.

Every system we ship assumes mistakes will happen and engineers for them: confidence thresholds route uncertain cases to human review, consequential actions require sign-off, and every automated decision is logged so errors are traceable and fixable. Accuracy is audited weekly against human-checked samples — autonomy is earned with evidence, never assumed.

Yes — it's routine. Our deployments run inside GDPR, UK financial-services compliance expectations and healthcare data-governance policies, with audit trails designed for exactly those reviews. India-side work accounts for the DPDP Act. Where your compliance team has specific requirements, they join the strategy phase and their requirements become acceptance criteria.

Our production case studies span legal, retail and e-commerce, healthcare, financial services and education, and the underlying patterns — document processing, order and enquiry handling, reporting, phone automation — transfer across most mid-sized operations. If your industry isn't listed, the honest test is volume: thousands of documents, orders, calls or applications a month means the economics almost certainly apply.

Our sweet spot is 50–500 people — big enough to have real process volume, small enough to ship in 90 days without a steering committee. Smaller firms with one high-volume process (a clinic's phones, an agency's inbox) are often great fits too. If you need a 2,000-seat global transformation programme, we're the wrong firm and will happily say so.

One team, two offices. Engagements are led from the office nearest you, with engineering shared across both — which is how we offer senior UK-market judgement and deep build capacity at mid-market prices. Clients talk directly to the people building their system in either geography; there is no offshore wall.

Our in-person AI school in Kochi, with three tracks: school students aged 14–18 (from Summer 2026), college students, and a four-week intensive for working professionals. Courses are taught by the same practitioners who build our commercial systems, class sizes are capped, and every student ships a real project. Over 500 students have trained with us so far.

Contact us through the form or hello@cortexintel.ai with the track you're interested in — school (14–18), college, or professional — and we'll send current cohort dates, fees and the syllabus. Seats are limited (18–24 per cohort depending on track) and need-based scholarships are available in every cohort.