AI-Native Founder

A small implementation cohort for founders of established companies

Build a company that gets smarter every week - alongside founders doing the same

Most companies are adding AI tools. Very few are building the shared context those tools need to become genuinely useful.

Your customers, decisions, processes and hard-won lessons are spread across software, meetings, messages and people’s heads. That means every new agent starts half-blind, and the company keeps relearning things it already knows.

Over eight Friday afternoons, you will begin turning that fragmented knowledge into a company brain - then use it to build dashboards, internal tools, agents and operating loops around the way your business actually works.

You will do it alongside a small group of founders who openly share what they are building, what worked and what they would do differently.

This is not an AI ideas course. It is a working founder community where every week combines learning, real examples, peer insight and practical implementation inside your own company.

  • Starts Friday 4 September 2026
  • Fridays, 1:00pm–4:00pm
  • Central London or live on Google Meet
  • Recordings for one year
  • £999 total
  • Maximum 20 approved companies

I am initially opening this to founders in my network. Applying is simply a way to check that the programme fits your company and that no direct competitors are placed together.

A note from Hugo

I built the programme I wish I had 18 months ago

When we started building more AI into ChargedUp, I assumed the big opportunity was replacing SaaS and automating tasks.

That helped, but it was not the real breakthrough.

The breakthrough was giving every new system access to the same company context.

Once our data, knowledge, goals, decisions and permissions began to connect, dashboards could explain what was happening, agents could act with useful context and each workflow could record what it learned for the next one.

I am running this cohort to help a small group of founders build that foundation without repeating all of our mistakes.

I will show the actual architecture, repositories, prompts, workflows and operating loops we use. The group will also share real implementations, lessons and best practices so we can learn faster from one another than any company could alone.

- Hugo Tilmouth, Founder and CEO of ChargedUp

The core idea

An agent is only as useful as the company it can understand

Generic AI is impressive, but it does not know why your best customers buy, what your team considers a good outcome, which exceptions matter or what decisions have already been made.

An AI-native company creates a central intelligence layer that connects those things - without replacing every existing system on day one.

System architecture

Inputs

  • CRM
  • Email
  • Slack
  • Finance
  • Support
  • Sales calls
  • Marketing
  • Documents
  • Operations

Central intelligence layer

Company Brain

  • Customers
  • Locations
  • Products
  • Goals
  • Decisions
  • Knowledge
  • Permissions
  • Memory

Powers

  • Founder agent
  • Management dashboard
  • Marketing loop
  • Support agent
  • Internal tools
  • Sales workflows
Every useful outcome and human correction improves what happens next.

Why this matters now

Your company knows more than its systems can use

Every week, your business produces valuable learning.

Customers explain why they buy. Sales calls reveal objections. Support exposes recurring friction. Managers intervene when a number moves in the wrong direction. Founders make decisions using years of accumulated judgment.

Most of that disappears into meetings, inboxes, call recordings and individual memory.

The result is not just wasted time. It is a company that cannot compound what it learns.

Where learning gets trapped

  • People’s heads
  • Slack threads
  • Email
  • Meeting notes
  • SaaS tools
  • Spreadsheets
  • Call recordings
  • Old decks

You already have the raw material. The next step is making it usable.

The mechanism

Move from automation to operating loops

An automation repeats an instruction.

An operating loop observes what happened, uses company context to decide what to do next, measures the outcome and records the learning.

  1. 01Connect context
  2. 02Interpret what is happening
  3. 03Take an approved action
  4. 04Measure the outcome
  5. 05Record the lesson
  6. 06Improve the system

Step six feeds back into step one

The ambition is not a magical self-running company. It is something more practical: a business where useful decisions and corrections stop evaporating and begin improving the next cycle.

Small improvements, captured and repeated, become a compounding advantage.

Example loops

Ad creative generation

Customer language + past results → new angles → creative generated → campaign live → results measured → next round starts sharper

Sales call feedback and training

Call recorded → objections and winning lines extracted → playbook updated → team trained → next calls measured against it

Support that fixes root causes

Ticket arrives → full customer context loaded → reply drafted or automated → resolution measured → recurring issues fixed at source

In the programme, we will show exactly how to build these - and many more. And if there is a loop your business needs that is not on the list, you can request it.

The 5-Layer Blueprint

Make everything agent-accessible: data, decisions, and the tools to act.

Traditional software gives humans interfaces to data. AI-native operations give agents access to everything they need to act autonomously. Not just read data, but understand context, make decisions, and execute. Zero friction between insight and action.

Layer 1

Data Accessibility

"Can agents see what they need to see?"

MCP servers expose every data source (Stripe, HubSpot, Posthog, your database) as tools agents can read AND write. Not dashboards. Actual ability to act.

Examples

  • Revenue metrics from Stripe
  • Customer health from CRM
  • Product usage from analytics
  • Support load from Intercom
Layer 2

Skills & Capabilities

"What can agents actually do?"

Codified expertise as SKILL.md files: triggers, procedures, tool dependencies, success metrics. Composable, so skills can invoke other skills.

Examples

  • Churn prevention playbook
  • Lead scoring and routing
  • Incident response
  • Board deck generation
Layer 3

Decision Framework

"How do agents decide what to do?"

Policies, thresholds, escalation rules. Confidence-based acting with clear boundaries: what's autonomous, what needs approval, what's blocked.

Examples

  • Discounts up to 20% → auto-approve
  • Refunds over £500 → human review
  • Public comms → always escalate
  • Support routing → confidence threshold
Layer 4

Governance & Traceability

"Can we trust what agents did?"

Every action generates an audit record: what was decided, why, what data was used, what policies allowed it, what happened as a result.

Examples

  • Full decision reasoning captured
  • Policy compliance logged
  • Outcome tracking
  • Explainable to auditors
Layer 5

Self-Improving Loops

"How do agents get better over time?"

Act → Measure outcome → Learn from results → Update skills → Repeat. Agents track false positives, analyse human overrides, and propose their own improvements.

Examples

  • Email timing optimisation
  • Threshold auto-tuning
  • Pattern extraction from corrections
  • Skill refinement proposals

When all 5 layers work together

Speed

Issues resolved in minutes, not days

Scale

1 human + agents = 10-person team

Quality

Consistent execution, continuous improvement

Trust

Full audit trail, clear accountability

Leverage

You focus on strategy, agents handle ops

Traditional company, Monday 9am

  1. CEO checks dashboards
  2. Notices churn spike
  3. Slacks CS team
  4. Meeting scheduled for Wednesday
  5. Analysis presented Friday
  6. Action plan following Monday
  7. Implementation over 2 weeks

Time to resolution: ~3 weeks

AI-native company, Monday 9am

  1. Agent detected churn spike on Saturday
  2. Root cause identified (billing issues from Friday deploy)
  3. Affected customers already contacted with apologies
  4. Engineering notified, fix deployed Sunday
  5. CEO gets summary: "Handled. 3 at risk, 2 retained, 1 escalated for your call."

Time to resolution: handled before you woke up

The outcome

A working foundation, built around your company

Each founder works on one real implementation throughout the programme. You will also set up a personal founder agent in the first week and improve it as your company context develops. By the end, you will have:

AI-Native Company Blueprint

A map of the data, knowledge, tools, agents and loops that should form your intelligence layer.

Central Brain Architecture

The entities, context sources, knowledge, permissions and memory your systems need to share.

A Deployed Company Application

A working application using Claude Code, GitHub, Vercel and Supabase.

A Management Intelligence Loop

Targets, live data, analysis, actions, owners, outcomes and recorded lessons.

An AI Marketing Loop

A system that turns customer language and campaign results into better future creative.

A Personal Founder Agent

Set up in week one, then progressively connected to your company context so it can support research, planning, meeting preparation, decisions and follow-through.

An Operational Agent

One narrow, useful agent connected to the right knowledge, tools and human approval points.

A 12-Month Roadmap

A sequenced plan for what to connect, build and automate next.

Fit

Designed for founders with a real company to redesign

The programme is best suited to founders of revenue-generating technology or technology-enabled businesses with customers, a team, established workflows and enough operational complexity for shared context to matter.

A good fit if

  • You can identify one or two costly workflows to improve
  • You are willing to openly share useful lessons and examples with a trusted peer group
  • You want company-owned capability, not another disconnected AI subscription
  • You can attend or watch each weekly session and implement between sessions

Not a fit if

  • You are looking only for AI inspiration or prompt tips
  • You do not yet have real company workflows or data
  • You want someone to build a complete bespoke platform for you
  • You cannot allocate time to implementation

You do not need to be a software engineer. You do need to be willing to build, test and make decisions.

The eight-week programme

Learn together. Build every week. Share what actually works.

Every Friday follows the same practical rhythm: a focused lesson, a show-and-tell of real systems, founders sharing what they built and learned, guided implementation, and a clear practical action to complete before the next session. The weeks build one connected company foundation rather than eight disconnected experiments.

Flexible by design

The weekly structure is the spine, not a fixed script. You can request specific systems for your business requirements - a loop, a dashboard, an agent, an integration - and we will cover exactly how to build them, adapting sessions to what the cohort most needs.

Built in a real operating company

The programme begins with the systems, not slides about them

The opening session starts with a detailed show-and-tell of the most powerful things we are building inside ChargedUp, followed by strong examples from across the industry. Throughout the programme, these real systems become working reference points rather than abstract case studies.

  • Daily performance numbers posted automatically to the team, with no manual reporting cycle.
  • Live management and investor reporting drawn from the same operational data.
  • Sales calls and meetings feeding searchable company memory.
  • Targets that trigger investigation, ownership, action and outcome measurement.
  • A founder agent for briefings, meeting preparation, decision tracking and follow-through.
  • Support workflows that understand the customer, location and history before acting.
  • Internal applications and presentations generated from current company context.

Sensitive customer and employee information is removed. The point is to understand the architecture and operating principles, not expose private data.

Your instructor

Lessons from building this inside ChargedUp

I am Hugo Tilmouth, Founder and CEO of ChargedUp. Over the last 18 months, I have been building a central operating system that connects more of our company’s data, knowledge, people and processes.

I am not teaching a theoretical “future of AI” framework. The programme is based on what worked, what broke and what became obvious only after we tried to deploy these systems in a real business.

My role is to give you the architecture, examples and practical support to build your first useful version - while creating the conditions for founders to teach one another through honest show-and-tell, shared failures and reusable best practices.

Hugo Tilmouth

Founder and CEO of ChargedUp · Forbes 30 Under 30 · Building an AI-native operating system inside a live company

Connect with Hugo on LinkedIn

Built with peers, not in isolation

The best practices will come from the room as well as the programme

Every company will approach the same AI-native principles from a different starting point. One founder may solve management reporting. Another may connect customer conversations. Another may discover a better way to evaluate an agent or structure permissions.

Those lessons become more valuable when they are shared. The cohort is designed around honest demonstrations of what people have built, what failed, what changed and what others can reuse.

The private founder community remains open after the eight-week programme - members keep sharing implementations, industry developments, tools and best practices as the field changes.

The aim is a trusted network of founders who are not merely consuming AI news, but building the operating practices that other companies will follow.

Community principles

  • Real examples over theory
  • Useful detail over polished success stories
  • Shared prompts, patterns and lessons
  • Confidentiality and competitor protection
  • Constructive peer review
  • Accountability to keep implementing

Hybrid by design

A live build room for founders

Each Friday session runs from 1:00pm–4:00pm UK time. You can attend in Central London or join through Google Meet, and you may change format from week to week subject to room capacity. Sessions are designed as working sessions rather than lectures.

In person

Build alongside a small group of founders, see what others are implementing, share your own progress and troubleshoot together.

Central London - exact venue shared with attendees

Online

Join the full working session live, participate in peer show-and-tell, ask questions and complete the same guided implementation.

Included in both

  • Focused teaching
  • ChargedUp and industry show-and-tell
  • Founder progress sharing
  • Best-practice exchange
  • Guided implementation
  • Q&A and technical troubleshooting
  • Weekly practical work
  • Templates and repositories
  • Recording access

The offer

The programme, materials and implementation foundation

The value is not in eight presentations. It is in leaving with a connected foundation, a trusted group of peers and reusable assets that help every member continue building after the cohort ends.

  • Eight live three-hour working sessions
  • Hybrid attendance
  • One year of recordings
  • Personal founder agent setup in Week 1
  • Weekly founder show-and-tell
  • Real ChargedUp system demonstrations
  • Current industry examples
  • Peer review and best-practice exchange
  • Weekly practical builds and implementation work
  • AI opportunity audit
  • Company systems and data audit
  • Phased migration plan
  • Company blueprint templates
  • Central brain architecture templates
  • Example database structures
  • Reusable code repositories
  • Dashboard and target templates
  • Cron and automation patterns
  • Agent specifications
  • Prompt libraries
  • Example skills and instructions
  • Marketing workflows
  • Technical setup guides
  • Private founder community
  • Between-session blocker support
  • Ongoing community access after the programme
  • Mutual confidentiality agreement
  • Commercial internal use of supplied code
  • 12-month implementation roadmap

Between-session support is for implementation questions and blockers. It does not include building a complete custom system on your behalf.

Build capability your company owns - with peers who keep helping one another improve it.

Founder pricing

A practical investment in company-owned capability

Founder Ticket

£999 total

No VAT will be added.

  • Eight live working sessions, in Central London or online
  • All templates, repositories and recordings
  • Personal founder agent set up in Week 1
  • Ongoing community access after the programme

The cohort is founders only.

It keeps the room honest - founders can speak openly about numbers, people and what is not working. You take the systems, templates and learnings back to your team, and you keep access to the founder community after the programme ends.

This first cohort is intentionally priced for founders in my network to join, implement and help shape the programme. Future cohorts may be priced differently.

The programme proceeds once 10 approved founders have paid. If that minimum is not reached, payments are refunded in full. Once the cohort is confirmed, places are non-refundable but may be transferred to another approved founder, subject to competitor restrictions.

A simple fit check

Request an invitation. Decide after you have the details.

The application is not intended to manufacture exclusivity. It helps create a trusted, useful peer group, avoid placing direct competitors together and make sure every participant has a real implementation and useful experience to contribute.

  1. Step 01

    Apply

    Tell me about your company and what you would like to build. It takes under two minutes.

  2. Step 02

    Hold your place

    Register on Luma to hold the place. A card is required, but nothing is charged while the application is being reviewed.

  3. Step 03

    Review

    I review company fit, implementation readiness and competitor conflicts.

  4. Step 04

    Approval

    If approved, your Luma registration is accepted and the card is charged.

  5. Step 05

    Cohort confirmation

    When 10 approved founders have joined, the cohort is confirmed and onboarding begins.

There is no sales call required. If I do not think the programme is right for your company, I will tell you.

FAQs

Frequently asked questions

September 2026 founder cohort

Build the foundation your next ten AI tools will need

Connect your company’s context, create useful operating loops and begin building software and agents around the way your business actually works.

Eight Friday afternoons. A maximum of 20 companies. One working foundation - plus a trusted peer community that continues after the programme ends.

Curious but unsure whether it fits? Submit the short application and describe what you are trying to build. I will give you an honest answer.

Starts Friday 4 September 2026. Applications are reviewed before places are offered.