The specialist for the systemnobody understands anymore.
The senior picks up. Every time. No account managers, no juniors-in-training.I'm Darren. 25+ years untangling legacy software for banking, healthtech, wealth management, PropTech and the NHS. Established mid-sized companies bring me in for the critical system that's become a risk, and I leave the team stronger than I found it.
Built by people who've long since moved on. Releases are a held breath, changes take days, and the knowledge to fix it fast walked out with them.
ii.
Every change costs more
The team is capable, but the estate makes them slow. Small changes cross five systems, releases keep slipping, and nobody can explain where the time really goes.
iii.
The six-figure rebuild
An agency has quoted six months and a team of four, and the board wants the number justified. There's usually a sharper, staged route that doesn't bet the year on a big bang.
And there's usually a date attached: a PE deal on the horizon, an FCA resilience deadline, a board asking for a costed modernisation plan, or a resignation letter from the only person who understands the system.
Free · No email required
Software Health Check
Five questions on where the risk actually sits, then an honest first read from a senior pair of eyes. No email, no follow-up unless you ask.
Where does the real risk sit in your software estate?
1 / 5
How we work together
One clear first step. Everything after it is optional.
The assessment is the product, not a disguised sales pitch. Use the roadmap with me, take it in-house, or hand it to somebody else. A pilot or retainer only follows if it earns its place.
Start here
£7,950/ fixed · 3 weeks · ex VAT
Estate Assessment
Three weeks in your codebase, read-only, ending in a written report: what you've actually got, where the risk sits, and a sensible order for dealing with it. Detailed enough for the team that has to live with it, clear enough for the board that has to fund it.
Estate inventory and risk map: systems, dependencies, key-person exposure
Phased modernisation roadmap with effort bands, written for a board decision
AI-readiness: where it genuinely helps, where it bites, guardrails included
Quick wins your own team can action straight away
Written report plus a 90-minute leadership walkthrough
Read a sample report before you spend a pound, or take the one-page summary to your board. The fit call confirms the estate can be covered properly at the fixed price before either side commits.
£19,500/ fixed · 4–6 weeks · ex VAT
Modernisation Pilot
One subsystem from the roadmap, delivered, with before-and-after numbers you can put in front of a board. A characterisation test harness pins down existing behaviour before anything changes, and it stays with you afterwards.
First roadmap phase shipped on one subsystem
Characterisation test harness built first, yours to keep
Metrics agreed in week one, measured at the end
Your developers shadow the work; the method is handed over
Fixed price preferred; day rate agreed at scoping where it fits better
Take the one-page summary to your board. Broader engagements (legacy rescue, architecture, performance) are scoped the same way: fixed wherever the work can be pinned down.
from£4,000/ month · ex VAT
Fractional Principal
The senior on tap. A standing share of my time each month, used flexibly: architecture, code and vendor review, technical due diligence, the hard calls. For teams who want the expertise close without a permanent hire.
Three days a month, shaped to you
Architecture, code review, technical due diligence
A full day a week runs from £6,000 a month. Lighter-touch advisory from £2,500. We agree shape and scope together.
How it works
Weeks, not months. Visible, not vapour.
You'll see progress every week. I send a short Friday note: what shipped, what's next, what I'm stuck on. No sprint theatre, no status decks.
"If I can't show you something real by Friday of week two, I'm doing it wrong."
Day 0
Take the Health Check
Two minutes, no email required. You get a plain-English read on where you stand. If something rings true, you tell me a bit more.
Weeks 1–3
Assessment → roadmap
Read-only, in your code and around your people. Automated inventory, three stakeholder interviews, AI tooling tried against a copy of your code. Ends with the written report and a 90-minute walkthrough.
Weeks 4–9
Pilot the first phase
One subsystem from the roadmap, delivered with before-and-after numbers. Characterisation tests first, your developers shadowing throughout, the working method handed over at the end.
Beyond
Keep shipping, or stop
Carry on to the next phase, keep a fractional retainer, or take the roadmap in-house. Either way your team leaves stronger than we found it.
What this looks like in practice
The kinds of problem we take on. Three shapes.
Real engagements, described by sector rather than client. Different shapes, same through-line: the software ends up better understood, and the in-house team ends up stronger than it started. Where a programme ran long, it ran as staged phases with working software from the early weeks, never a big bang.
01
PropTech · Legacy monolith
~16 months
UntangledThe codebasestored procedures to clean architecture
The codebase everyone tip-toed around.
Before
A .NET monolith that had become, in practice, a wall of stored procedures and triggers. No tests, no documentation, no-one still there who fully understood the logic. Every release was a held breath.
What we did
Stabilise the worst performance pain first to buy breathing room. Wrap the system in black-box tests against sanitised production data so behaviour is pinned down before anything moves. Decompose into a modular architecture, piece by piece, with the business running on the old version throughout.
Outcome
A codebase the in-house team can now change without fear, a DAL that performs under load, and a clear route for them to keep eating the elephant on their own after we're gone.
KeptThe clientswho were ready to walk out the door
The architecture that kept clients from leaving.
Before
A flagship product with long-standing performance issues, an AngularJS frontend showing its age, and significant clients threatening to leave. A product team that needed a credible next-generation story, fast.
What we did
Define the target architecture end-to-end, hardware integration, frontend, federated GraphQL, microservices, infrastructure, reporting, and write a vertical slice through it so nothing stays theoretical. Deliver an MVP of the at-risk product to a hard deadline. Recruit and shape the team that will own it.
Outcome
The at-risk clients stayed. A modern architecture in place with the team structure and ways of working to run it. The in-house team leaves this engagement with more capability than they started, not less.
ZeroDefects at go-liveone chance, right first time
The migration that had to be right first time.
Before
A legacy Lotus Notes estate that had to move to SharePoint inside a non-negotiable window. No second attempt, if it failed, it failed in front of the whole business.
What we did
Build a purpose-made C# migration utility, not a script. Model every document type and edge case the old system had accreted over the years. Dry-run against sanitised data until the report is boring, then run it for real.
Outcome
Zero defects. An uneventful go-live, the highest compliment a migration can get. Everything the team needed to run similar moves themselves afterwards.
C# / .NETLotus NotesSharePointData migration
In their words
The sort of thing colleagues say when you're not in the room.
He doesn't just tell you what a solution should be; he actually builds it as a proof of concept. You always get an honest, real-world view of the challenge.
Senior healthcare technology colleague · public LinkedIn recommendation
He is the one to go to if you want to solve problems. A key reference for everybody in the team. Having him in the team means success.
Senior healthcare technology colleague · public LinkedIn recommendation
Both are public recommendations, and there are more of them: read them on LinkedIn. Names and full recommendations are visible there. Client references are available before you commit.
A different problem, a separate route
Need AI governance or team adoption instead?
Leadership briefings, hands-on team workshops and multi-week embed programmes now have their own buying path. Same senior, practical approach; no toy demos and no need to pretend it is legacy modernisation.
Because it's a proper three-week engagement, not a discovery call. I come in read-only, run the inventory, sit with your people and write the report. The fixed £7,950 means you know the cost before we start, and I'm not incentivised to find work that isn't there. The report is yours to keep: use it with me, without me, or with someone else. There's a sample report linked above so you can see exactly what you're buying.
A free assessment is a sales instrument: it's funded by a vendor, done by whoever is on the bench, and it tends to conclude that you need what that vendor sells, usually a migration. This one is independent and senior-only. Nobody behind it is paid on what you do next, the report is the product rather than the pitch, and it will happily conclude 'leave it alone' where that's the honest answer.
Established mid-sized companies with one defined software estate that matters: the system the business runs on, or the one a regulator cares about. Usually a CTO, engineering lead or MD who needs a specific hard problem solved, not another agency to manage. You get a senior specialist directly, with trusted senior devs on tap when a job needs more hands. No account managers, no bench of juniors learning on your time.
Often yes, just in a different way. AI makes first drafts and mechanical analysis faster; it does not decide which behaviour matters, where authority belongs, what evidence is sufficient or whether a production result is safe. I pair with the in-house team on those harder judgements, the architecture and the verification, while helping them make good use of the faster tools. The aim is not to replace your developers. It is to leave them faster and more capable without lowering the bar.
No. Most clients come to me with a frustration, not a spec. The assessment is where we turn 'something's wrong' into 'here's what we'd do about it.'
Some of it. Plenty of it isn't. AI is already very good at first-pass analysis, drafts and repetitive implementation. The useful question is where that speed survives review and produces a better business outcome. The seasoned developer's role shifts towards framing the job, supplying context, designing the checks, controlling authority, integrating the result and owning what reaches production. Usually the right starting point is one or two bounded jobs, not fifty.
You do. All the bespoke work Epona writes for you is yours outright, assigned to you on delivery, clean, documented, in your repos. We lean on permissive open-source libraries (MIT, Apache 2.0, BSD) so you can fork, audit or walk away whenever you like. Some of those libraries Darren maintains himself, which means faster fixes, not vendor lock-in. Licences flagged up front. No black boxes.
Your message and email address come to me and nobody else. No mailing list, no sequence, no sharing with third parties, and no follow-up beyond answering you unless you ask for it. Health check answers are only sent if you choose to send them, and they're used to write you a reply, nothing more. The privacy notice sets out the detail.
TypeScript, React, Node, .NET/C#, Python, Rust. Darren's 25+ years span most eras of software, including the inherited-from-someone-else kind.
Let's untangle it.
Tell me what's stuck and I'll reply within a working day. If you would rather talk it through, request a confidential fit call. It is a check on whether I can genuinely help, not a sales presentation.
