Services · AI Platform Development
£60,000 to £120,000 a year in subscriptions. People doing rules-based work an agent could do in minutes. A £5K diagnostic tells you what to keep, what to cut, and what an agent can replace. Your team owns the layer at the end.
Three buyers. Three tool menus. One platform underneath. Commercial Leaders, Marketing Leaders, and Business Leadership. Built around your workflow, your regulatory environment, and your commercial model. Not bolted onto assumptions built for someone else's industry.
Built for your team. Run by your staff.
Before the menus
Half of every AI conversation is people meaning different things by the same word. A sixty-second level-set before we walk through what we'd build for you.
Option 1
“Let's get the team on ChatGPT.”
ChatGPT, Copilot, Gemini. General-purpose chat models. Genuinely useful for personal productivity. Lock you into someone else's product roadmap. Limited control over which data leaves your environment.
Option 2
“Our CRM has AI now.”
HubSpot AI, Salesforce AI, Veeva AI. Generic AI bolted onto software built for everyone. Faster than no AI. Still per-seat pricing. Still not built for your regulatory environment, your commercial workflow, or your data.
Option 3 · what we build
“Software built around our workflow.”
Bespoke software built around YOUR commercial workflow. Trained on YOUR data. Owned by YOU. Replaces the specific subscriptions and rules-based work an agent can do in minutes. This is what we build.
Pick your audience
Most AI buys are for one function. We build for the whole business. Pick the track that matches the team you're building for.
Track 1 · Commercial Leaders
Six tools a Commercial Director would commission to lift the team's output without hiring another head. Each one replaces a specific slice of admin or agency retail cost.
“My reps spend three weeks qualifying prospects against ICS and Trust priorities before they know if an account is worth pursuing.”
Ingests ICS, NHS Trust, and JCVI strategy documents and ranks prospects by fit with live national priorities. Output is a one-page qualification brief per account.
£12K-£18K build
“We don't have a consolidated view of who influences what across the NHS and private market. Every new rep rebuilds the same map from scratch.”
Scrapes publications, committee rosters, and public filings to build a shared map of clinical decision-makers and their affiliations, updated weekly.
£18K-£28K build
“Every proposal ties up a senior commercial lead for eight to twelve hours, and the output varies in quality depending on who drafted it.”
Drafts proposals and SOWs in your house voice using your frameworks. Senior time shifts to review and judgement, not production. This is Salesy, the agent we run AGENCY on.
£15K-£25K build
“We buy BI data. Nobody has time to turn it into a ranked prospect list. It sits in a folder until it goes stale.”
Applies a configurable PQ signal framework to your BI feeds and outputs a ranked list of prospects every Monday, with trigger reasons attached. This is Prospy.
£10K-£16K build
“Reps duplicate qualification work because account intelligence lives in email threads and personal notes. A rep leaves and their knowledge leaves with them.”
A shared agent-assisted workspace where account qualification, stakeholder maps, and trigger events compound across the team. Owned by the business, not the rep.
£25K-£40K build
“We cap outbound at 200 sends a week because legal reviews every template by hand. Compliance is the bottleneck, not the team.”
Every outbound draft runs through an agent trained on ABPI, ASA, and MHRA rules. Non-compliant phrasing is flagged and rewritten before a human reviewer sees it.
£12K-£20K build
Track 2 · Marketing Leaders
Six tools a CMO or Head of Marketing would commission to retire subscription tools and redirect agency spend from production to strategy.
“We pay agencies £2K-£4K per brief for content we could generate ourselves if the system knew our voice and our rules.”
Trained on your content bible, product portfolio, and compliance rules. Produces on-brand LinkedIn posts, blog drafts, and product copy. Seventeen quality tests run automatically. This is Posty.
£15K-£28K build
“Compliance review is the bottleneck. Campaigns wait a week because the approval chain lives in email and spreadsheets.”
Routes every asset through claims, regulatory, and brand review with a full audit trail. Agent pre-flags risk so reviewers see only what needs human judgement.
£25K-£45K build
“Every congress needs exhibition graphics, KOL one-pagers, social cutdowns, and leave-behinds. Each format is a separate agency brief at £1K-£3K a time.”
Produces image, video, and infographic variants on brand, from a single prompt. Exports print-ready and channel-ready files. Human creative direction on top, agent handles the production.
£18K-£35K build
“Campaign sequences are bespoke every time. Each one takes a senior copywriter two days to draft and review.”
Generates segmented multi-channel sequences (email, LinkedIn, SMS) against a campaign brief, with merge fields and compliance pre-checks. Output ready for the CRM to schedule.
£12K-£20K build
“Our newsletter goes out when someone finds time, which means once a month if we're lucky. The list is warm. The output isn't consistent enough to use it.”
Drafts the weekly newsletter from your content calendar, calls, and industry feeds. Voice-matched, schema-complete, ready for one-pass human edit before send.
£8K-£15K build
“We have ad platforms, a CRM, Google Analytics, and a marketing-ops tool. Nothing talks. Reporting takes the analyst two days a month.”
Pulls ad platform, CRM, and web analytics into a single live dashboard with attribution rules you define. Replaces the Monday morning report-building ritual.
£20K-£40K build
Track 3 · Business Leadership
Four tools a CEO, MD, or Board would commission to replace ad-hoc research and quarterly consultant reports with a continuous intelligence layer.
“I hear about competitor launches from sales reps three weeks after they happened. I want to see movement as it happens.”
An agent watches industry RSS, press releases, clinical trial registries, and regulatory filings for signals tagged to your markets and competitors. Weekly digest, real-time alerts on flagged events.
£10K-£18K build
“We commission industry reports twice a year for £15K each. By the time we read them, half the data is stale.”
Generates a monthly or quarterly report on your chosen market (UK health tech, medical devices, a therapeutic area) with trends, deal activity, regulatory movement. Your voice, your framing.
£15K-£25K build
“We need customer research to make the next product decision. An agency quote is £25K and four months.”
Runs targeted desk research across forums, reviews, academic literature, and patient advocacy groups. Surfaces themes, quotes, and gaps. Faster and shallower than primary research. Good for framing the questions you then take to interviews.
£12K-£22K build
“Every board pack takes two weeks of finance, ops, sales, and marketing time. The data is dated by the time the board reads it.”
Pulls live data from finance, CRM, ad platforms, and ops dashboards into a board-ready pack. Written commentary generated against your framework. The team signs off, they don't build it.
£25K-£45K build
How we build
Most AI projects start with the tool and end at the demo. Ours starts with the workflow and ends when your team has run the platform for a week without us in the room.
01
Week 1
We map your commercial workflow end to end. Where time leaks. Where SaaS is overlapping. Where judgement is rare and rules-based work is common. No slides. One session and a walk-through of the real screens your team uses.
Deliverable · Workflow map + cost-to-replace matrix
02
Week 2
We write the functional specification for the agents or the platform. What each does, what it must not do, where the human-in-the-loop sits. The spec is the contract. No feature creep without a written change order.
Deliverable · Functional spec document (signed-off)
03
Week 3
We define who can see what, where data lives, how audit trails work, which models run under which accounts. Compliance review happens here, before a line of code. If your security team has questions, they're answered in this step.
Deliverable · Security + data model, model-provider matrix, access policy
04
Weeks 4-6
We build the smallest version of the platform that does the job. Real data, real workflow, ugly buttons. Your team uses it against real work inside a week. We learn what needs to change before we polish anything.
Deliverable · Working MVP your team operates against real work
05
Weeks 7-12
We tighten the agents, add the missing workflows, harden the platform. Visual polish comes last, not first. We leave when your team has run the platform for a full week without us in the room.
Deliverable · Production platform + documentation + handover week
The methodology transfers. Your team runs it after we leave.
Proof
Six internal agents run the commercial and content operations of AGENCY Bristol. The same architecture we deploy for clients. The productised version is Platty. The list below is what our own operating layer does every week.
Proposal and SOW drafting
Drafts commercial proposals in the AGENCY voice using the book frameworks. Replaced roughly 12 hours a week of senior commercial time. Output reviewed, never published raw.
Outreach and prospect qualification
Applies the PQ signal framework from the book to qualify prospects and draft first-contact outreach. Work a business development hire used to do. No BD hire needed.
Content production
Produces weekly LinkedIn content against the AGENCY content bible. Seventeen quality tests run automatically. A workflow that used to take 6 hours now takes 8 minutes with human review.
Operations and briefing
Runs the morning briefing, pulls the daily priorities, manages the project kanban state. Replaced a part-time operations role at AGENCY.
Campaign planning and scheduling
Applies the content bible to weekly scheduling across channels. Writes the Metricool CSVs. Work the marketing coordinator used to spend a day a week on.
Strategic positioning
Applies the MCT capstone pitch canvas to positioning work. Used inside pitch decks and brief creation. Operates as the strategic sounding board that used to require a paid advisor.
We would not sell you a platform we were not willing to run our own business on.
How we work
£5K · 2 weeks
Written assessment. We map your commercial workflows, cost out the saving (SaaS, hours, opportunity cost), and tell you what is worth building. Before you commit to anything else. Fee credited against the build if you commission one.
£15K-£35K · 4-6 weeks
One workflow automated. One agent, trained on your voice, your data, your compliance environment. Deployed, tested against real work, handed over to your team.
£25K-£75K · 8-12 weeks
The full build. Multiple agents, a platform for them to run in, workflow automation, handover, documentation. Your team producing real work on it for a week before we leave the room.
£1K/month
Optional. Model upgrades, new agent training, minor workflow additions, direct access to our team. Month to month. Cancel any time. Most teams run independently after six months.
Who owns the code?
You do. Outright. Built on open-source frameworks (Next.js, Supabase) with model APIs (Claude, GPT, Gemini) under your own accounts. No licence fees, no vendor lock-in, no per-seat pricing. If you ever want to hire a developer to extend it, they get the full codebase.
Who runs it after you leave?
Your staff. That is the whole point. The build programme ends with your team producing real work on the platform for a full week without us in the room. No theoretical handover. No PDF runbook as the only documentation. Your team operates it before we sign off.
How do you handle compliance?
Every agent is built inside your regulatory environment. ABPI, ASA, MHRA, GDPR. The human-in-the-loop is non-negotiable. Agents draft, people verify claims and evidence links, regulatory review stays where it is. We have shipped this inside regulated content operations and would not pretend the compliance layer is an afterthought.
What if a model gets deprecated?
The architecture is model-agnostic. Swap the API key, keep the workflow. Claude Sonnet today, something newer in six months. Most AI platforms lock you in. Yours will not.
How is this different from Copilot or HubSpot AI?
Those are features bolted onto software built for general marketing. Your platform is built around your commercial workflow, your product portfolio, your regulatory environment. Different starting point, different outcome. Bolted-on AI makes existing tools faster. A platform replaces the tools it needs to replace.
What happens to the SaaS we already pay for?
We keep what is load-bearing (usually your CRM and your scheduler) and retire what the agents replace. The diagnostic tells you which is which before you commit to a build. Typical outcome: two to four subscriptions retired, £20K-£60K a year of budget recovered, paid back inside the first year.
Not ready for a build?
Some teams want to try AI inside their existing workflow first. Our half-day workshop covers the compliant content workflow, the AI usage policy, and a working prompt library built around your products. £850. Credited against any build engagement booked within thirty days.
Most build commissions start here. Teams run the workshop, see what is possible inside their own workflow, and book the diagnostic within thirty days.
See the workshopThe productised version
Platty is the productised version of the operating layer we run AGENCY Bristol on. It's a faster way to see what an agent fleet looks like in practice. Some clients stay on Platty instead of commissioning a bespoke build. The bespoke work starts where Platty ends.
Ask about it on the triage call and we'll show you what it does before the conversation goes any further.
Ask about PlattyHealthcare innovations deserve recognition. We help them get it.
