Most AI consultancies won't quote a number until you're on a discovery call. That costs everyone time. Here is what we build, what it costs, and when it ships. Every project fixed-scope, delivered in weeks not quarters, and still running properly in year two. AI software rots faster than traditional software (data sources retire, model APIs deprecate, fallbacks lie); someone needs to watch it. That's the moat.
Six months of watching buyers on our own cost-comparison sites (monitoringcost.com, egresscost.com, siemcostcalculator.com and others) taught us something obvious in hindsight: vendors who publish list prices win shortlists. Vendors who hide behind "custom pricing" get skipped, even when their number would have been cheaper.
Buyers do not want to book a sales call for every candidate. They want to build the model, get a rough shape, then decide who to actually talk to. So we publish the shape. If the fit looks right, the scoping call converts a rough number into a written fixed quote inside 24 hours.
Prices are ranges because scope varies. The scoping call locks a specific number, in writing. All prices exclude VAT.
A one-week paid pilot. We take one specific feature or workflow you're evaluating, build a working prototype, and ship a written note on what full delivery would look like. You get real code and a real quote at the end. Perfect if you're not ready to commit to £25k+ without seeing us work.
A production voice agent on your own inbound line. Twilio SIP + LiveKit + Deepgram + LLM + ElevenLabs. One workflow, dashboards, transcripts to email, synthetic-caller test harness so you can prove it holds under adversarial calls.
A custom admin, back-office or portal app with AI woven into the workflow. Auth, roles, database, integrations, admin UI. LLM sits where it actually helps: draft generation, categorisation, extraction, summarisation.
A chat-style AI assistant scoped to one industry (accounting, legal, property, healthcare, care, HVAC). Real integrations, real data. Ships as an embeddable widget or a standalone app.
A cost-comparison, calculator or research site in your category, built to rank. Same pattern we use across our own 450-site portfolio. Vendor-verified data, comparison tables, calculators, monetisation via affiliate, sponsorship or captured intent.
Find one workflow inside your business where AI would compress a week into a session, and ship the compression. Concrete deliverable, not a slide deck. Often the fastest way to prove AI's value before committing to a bigger build.
The bit most agencies leave out. AI software rots faster than traditional software (data sources retire, model APIs deprecate, fallbacks lie). The retainer keeps your shipped project healthy, monitored, and evolving.
Same voice-agent MVP scope. Rough numbers based on publicly available agency rates in London.
| Line item | Traditional dev shop | Digital Signet |
|---|---|---|
| Timeline | 14-20 weeks | 6 weeks |
| Team size | 3-5 engineers + PM | 1 senior + AI agents |
| Build cost | £80,000-140,000 | £25,000-40,000 |
| First-year support | £15-30k, retainer only | £12-36k, cancellable |
| Silent-failure monitoring included | Rarely | Yes |
There are dozens of shops now promising "6 weeks, AI-native, fixed price". Some are good; many aren't. Here is the honest side-by-side of what usually differentiates us.
| What matters | Typical AI-native shop | Digital Signet |
|---|---|---|
| Time to ship | "6 weeks" (often 10-12) | 6 weeks, and we publish the record |
| What happens at go-live | Hands you the repo, waves goodbye | Retainer covers year 2, when AI code actually starts to rot |
| Silent-failure monitoring | Rarely | Standard on every ship |
| Test coverage on shipped code | Not usually discussed | Discipline built in; ask for the last project's test-count delta |
| Who directs the AI agents | Junior devs learning on your bill | Twenty years of delivery experience, directing not directed by |
| Production proof | Marketing deck screenshots | Our own live voice product (bottie.bot), 450-site AI-native portfolio, AI receptionist on our own US line |
| How you get a price | Book a discovery call | Read the price on this page |
| De-risk before committing | Free scoping call (they're selling you) | £3k paid spike week, credits against the full build if you continue |
The build cost of AI-native software has collapsed. The scarce thing now is picking who's still around to fix it in month 14.
Because AI-native delivery compresses the build hours. What a 4-person team used to ship in 3 months, one senior engineer directing AI agents can ship in 6 weeks. The saved hours become saved cost, and we pass the saving through instead of banking it. Every SKU on this page is priced on the compressed timeline, not on 2020's hourly rates.
Honest answer: you might be fine, especially for a throwaway MVP or a prototype your team will rewrite. Where that path tends to break: month 3, when the code you shipped stops working because a data source retired, a model API deprecated, or a fallback lied silently. The £5k dev is long gone. Retaining them costs the same as us, and they usually can't debug their own code six months later because AI wrote most of it. Our published price includes the discipline (tests, monitoring, silent-failure detection) that makes the code survivable in year two, plus a retainer path so someone is actually watching. If your build is genuinely throwaway, the Cursor-dev route is fine; if it needs to still work in a year, our number is realistic.
To de-risk you before you commit to a £25-60k engagement. In one week for £3,000 we take one feature you're evaluating, build a working prototype, and hand you real code plus a written firm quote for the full build. You get to see how we work, what our shipped code looks like, and how we scope. If you then buy the follow-on build, the £3k credits against it. If you don't, you've spent £3k to test-drive rather than £30k to find out.
Book a scoping call. We only publish the SKUs we build repeatedly. Custom work happens, it just needs an explicit scope, timeline, and price agreed in writing before we start. We do not do open-ended time-and-materials engagements.
Rarely. If you are pre-revenue and the fit is genuinely exceptional, we will consider a small equity component alongside a reduced cash price. But the default answer is cash, because cash flow keeps a small consultancy alive.
You do. Every fixed-scope engagement includes a full IP transfer at ship. Repo, deploy access, model prompts, integration credentials, everything. You can walk away with the whole thing on day one if you want to.
Silent-failure monitoring (the data sources your app depends on retire quietly; the model APIs deprecate quarterly; fallbacks lie), bug fixes, and small feature evolution. Not open-ended new-feature builds. Bigger changes are quoted separately as new fixed-scope engagements.
UK, London. We work with clients in the UK, US, and EU. All prices in GBP, US clients typically pay in USD at the day's rate.
Thirty minutes on a call. We agree the scope. You get a written fixed quote within 24 hours, usually the same day. Free.
Prices last verified August 2026. All prices exclude VAT.
